which sentence best describes the difference between summarizing and paraphrasing

Paraphrasing vs. Summarizing (Differences, Examples, How To)

It can be confusing to know when to paraphrase and when to summarize. Many people use the terms interchangeably even though the two have different meanings and uses.

Today, let’s understand the basic differences between paraphrasing vs. summarizing and when to use which . We’ll also look at types and examples of paraphrasing and summarizing, as well as how to do both effectively.

Let’s look at paraphrasing first.

What is paraphrasing?

It refers to rewriting someone else’s ideas in your own words.

It’s important to rewrite the whole idea in your words rather than just replacing a few words with their synonyms. That way, you present an idea in a way that your audience will understand easily and also avoid plagiarism.

It’s also important to cite your sources when paraphrasing so that the original author of the work gets due credit.

When should you paraphrase?

The main purpose of paraphrasing is often to clarify an existing passage. You should use paraphrasing when you want to show that you understand the concept, like while writing an essay about a specific topic.

You may also use it when you’re quoting someone but can’t remember their exact words.

Finally, paraphrasing is a very effective way to rewrite outdated content in a way that’s relevant to your current audience.

How to paraphrase effectively

Follow these steps to paraphrase any piece of text effectively:

Read the full text and ensure that you understand it completely. It helps to look up words you don’t fully understand in an online or offline dictionary.
Once you understand the text, rewrite it in your own words. Remember to rewrite it instead of just substituting words with their synonyms.
Edit the text to ensure it’s easy to understand for your audience.
Mix in your own insights while rewriting the text to make it more relevant.
Run the text through a plagiarism checker to ensure that it does not have any of the original content.

Example of paraphrasing

Here’s an example of paraphrasing:

Original: The national park is full of trees, water bodies, and various species of flora and fauna.
Paraphrased: Many animal species thrive in the verdant national park that is served by lakes and rivers flowing through it.

What is summarizing?

Summarizing is also based on someone else’s text but rather than presenting their ideas in your words, you only sum up their main ideas in a smaller piece of text.

It’s important to not use their exact words or phrases when summarizing to avoid plagiarism. It’s best to make your own notes while reading through the text and writing a summary based on your notes.

You must only summarize the most important ideas from a piece of text as summaries are essentially very short compared to the original work. And just like paraphrasing, you should cite the original text as a reference.

When should you summarize?

The main purpose of summarizing is to reduce a passage or other text to fewer words while ensuring that everything important is covered.

Summaries are useful when you want to cut to the chase and lay down the most important points from a piece of text or convey the entire message in fewer words. You should summarize when you have to write a short essay about a larger piece of text, such as writing a book review.

You can also summarize when you want to provide background information about something without taking up too much space.

How to summarize effectively

Follow these steps to summarize any prose effectively:

Read the text to fully understand it. It helps to read it a few times instead of just going through it once.
Pay attention to the larger theme of the text rather than trying to rewrite it sentence for sentence.
Understand how all the main ideas are linked and piece them together to form an overview.
Remove all the information that’s not crucial to the main ideas or theme. Remember, summaries must only include the most essential points and information.
Edit your overview to ensure that the information is organized logically and follows the correct chronology where applicable.
Review and edit the summary again to make it clearer, ensure that it’s accurate, and make it even more concise where you can.
Ensure that you cite the original text.

Example of summarization

You can summarize any text into a shorter version. For example, this entire article can be summarized in just a few sentences as follows:

Summary: The article discusses paraphrasing vs. summarizing by explaining the two concepts. It specifies when you should use paraphrasing and when you should summarize a piece of text and describes the process of each. It ends with examples of both paraphrasing and summarizing to provide a better understanding to the reader.

Paraphrasing vs. summarizing has been a long-standing point of confusion for writers of all levels, whether you’re writing a college essay or reviewing a research paper or book. The above tips and examples can help you identify when to use paraphrasing or summarizing and how to go about them effectively.

Inside this article

Fact checked: Content is rigorously reviewed by a team of qualified and experienced fact checkers. Fact checkers review articles for factual accuracy, relevance, and timeliness. Learn more.

About the author

Dalia Y.: Dalia is an English Major and linguistics expert with an additional degree in Psychology. Dalia has featured articles on Forbes, Inc, Fast Company, Grammarly, and many more. She covers English, ESL, and all things grammar on GrammarBrain.

Core lessons

Abstract Noun
Accusative Case
Active Sentence
Alliteration
Adjective Clause
Adjective Phrase
Adverbial Clause
Appositive Phrase
Body Paragraph
Compound Adjective
Complex Sentence
Compound Words
Compound Predicate
Common Noun
Comparative Adjective
Comparative and Superlative
Compound Noun
Compound Subject
Compound Sentence
Copular Verb
Collective Noun
Colloquialism
Conciseness
Conditional
Concrete Noun
Conjunction
Conjugation
Conditional Sentence
Comma Splice
Correlative Conjunction
Coordinating Conjunction
Coordinate Adjective
Cumulative Adjective
Dative Case
Declarative Statement
Direct Object Pronoun
Direct Object
Dangling Modifier
Demonstrative Pronoun
Demonstrative Adjective
Direct Characterization
Definite Article
Doublespeak
Equivocation Fallacy
Future Perfect Progressive
Future Simple
Future Perfect Continuous
Future Perfect
First Conditional
Gerund Phrase
Genitive Case
Helping Verb
Irregular Adjective
Irregular Verb
Imperative Sentence
Indefinite Article
Intransitive Verb
Introductory Phrase
Indefinite Pronoun
Indirect Characterization
Interrogative Sentence
Intensive Pronoun
Inanimate Object
Indefinite Tense
Infinitive Phrase
Interjection
Intensifier
Indicative Mood
Juxtaposition
Linking Verb
Misplaced Modifier
Nominative Case
Noun Adjective
Object Pronoun
Object Complement
Order of Adjectives
Parallelism
Prepositional Phrase
Past Simple Tense
Past Continuous Tense
Past Perfect Tense
Past Progressive Tense
Present Simple Tense
Present Perfect Tense
Personal Pronoun
Personification
Persuasive Writing
Parallel Structure
Phrasal Verb
Predicate Adjective
Predicate Nominative
Phonetic Language
Plural Noun
Punctuation
Punctuation Marks
Preposition
Preposition of Place
Parts of Speech
Possessive Adjective
Possessive Determiner
Possessive Case
Possessive Noun
Proper Adjective
Proper Noun
Present Participle
Quotation Marks
Relative Pronoun
Reflexive Pronoun
Reciprocal Pronoun
Subordinating Conjunction
Simple Future Tense
Stative Verb
Subjunctive
Subject Complement
Subject of a Sentence
Sentence Variety
Second Conditional
Superlative Adjective
Slash Symbol
Topic Sentence
Types of Nouns
Types of Sentences
Uncountable Noun
Vowels and Consonants

Popular lessons

Stay awhile. Your weekly dose of grammar and English fun.

The world's best online resource for learning English. Understand words, phrases, slang terms, and all other variations of the English language.

Abbreviations
Editorial Policy

Key Differences

Know the Differences & Comparisons

Difference Between Summary and Paraphrase

On the other hand, paraphrase means the restatement of the passage, in explicit language, so as to clarify its hidden meaning, without condensing it. In paraphrasing, the written material, idea or statement of some other person is presented in your own words, which is easy to understand.

These two are used in an excerpt to include the ideas of other author’s but without the use of quotations. Let us talk about the difference between summary and paraphrase.

Content: Summary Vs Paraphrase

Comparison chart, definition of summary.

A summary is an abridged form of a passage, which incorporates all the main or say relevant points of the original text while keeping the meaning and essence intact. It is used to give an overview of the excerpt in brief, to the reader. In summary, the author’s ideas are presented in your own words and sentences, in a succinct manner.

A summary encapsulates the gist and the entire concept of the author’s material in a shorter fashion. It also indicates the source of the information, using citation. Basically the length of the summary depends on the material being condensed.

It encompasses the main idea of every paragraph and the facts supporting that idea. It does not end with a conclusion, however, if there is a message in the conclusion, it is included in the summary. It also uses the keywords from the original material, but it does not use the same phrases or sentences.

Summaries save a lot of time of the reader, as the reader need not go through the entire work to filter the most important information contained in it, rather the reader gets the most relevant information in hand.

Definition of Paraphrase

Paraphrasing is not a reproduction of a similar copy of another author’s work, rather it means to rewrite the excerpt in your own language, using comprehensible words and restructuring the sentences, but without changing the context. Hence, in paraphrasing, the original idea and meaning of the text are maintained, but the sentence structure and the words used to deliver the message would be different.

The paraphrased version of the text is simple and easily understandable. The length is almost similar to the original text, as it only translates the original text into simplest form. It is not about the conversion of the text in a detailed manner, rather it is presented in such a way that goes well with your expression.

In paraphrasing, someone else’s written material is restated or rephrased in your own language, containing the same degree of detail. It is the retelling of the concept, using a different tone to address a different audience.

Key Differences Between Summary and Paraphrase

The points discussed below, explains the difference between summary and paraphrase

To summarize means to put down the main ideas of the essential points of the excerpt, in your own words, while keeping its essence intact. On the contrary, to paraphrase means to decode the original text in your own words without distorting its meaning or essence.
A summary is all about emphasizing the central idea (essence) and the main points of the text. In contrast, paraphrasing is done to simplify and clarify the meaning of the given excerpt, so as to enhance its comprehension.
If we talk about the length of the summary in comparison to the original text, it is shorter, because summary tends to highlight the main points only and excludes the irrelevant material of the text. As against, in case of paraphrasing, the length is almost equal to the original text, because its aim is to decipher, i.e. to convert the complex text in a language which is easily understandable without excluding any material from the text.
The main objective of summarizing is to compile and present the gist of the author’s idea or concept in a few sentences or points. Conversely, the primary objective of paraphrasing is to clarify the meaning of author’s work in a clear and effective manner when the words used by him/her are not important or the words are too complex to understand.
A summary is used when you want to give a quick overview of the main ideas to the reader about the topic. On the contrary, Paraphrase is used when the idea or main point is more significant than the actual words used in the material and also when you want to use your own voice to explain the concept or idea.
A summary does not include lengthy explanations, examples and what the reader has understood. In contrast, a paraphrase does not include the exact same wordings or paragraphs used in the original source, so as to avoid plagiarism.

Steps for Summarizing

First of all, you need to read the entire passage twice or thrice to grasp the meaning and essence of the material.
Identify and underline all the important points, ideas and supporting facts which you have read.
Now, explain the material to yourself, for better understanding.
Rewrite in your own words, the salient points and central idea from the original text, in a few sentences.
Omit unnecessary detailing and examples.
Make a comparison of the original text and the summary which you’ve created.

Steps for Paraphrasing

Read the entire text carefully, twice or thrice, to absorb the meaning and essence.
Rewrite the author’s ideas in a unique language, i.e. in your own voice. Make sure that the sentences and words used are your own and it should not be a mere substitution or swapping of words and phrases.
Further, the sequence in which idea is presented, need not be different from the original source.
Compare the paraphrased version with the main text, and ensure that the essence clearly presented, as well as make sure that it is free from plagiarism.
Check that the words and phrases which are directly taken from the text are within quotation marks.
Provide references.

In a nutshell, a summary is nothing but a shorter version of an excerpt or passage. On the contrary, a paraphrase is the restatement of the original text or excerpt. One can use any of the two sources, as per the requirement, when the idea of any of the sources is relevant to your material, but the wording is not that important.

You Might Also Like:

Paraphrase and Summary

Paraphrase and summary are different writing strategies that ask you to put another author’s argument in your own words. This can help you better understand what the writer of the source is saying, so that you can communicate that message to your own reader without relying only on direct quotes. Paraphrases are used for short passages and specific claims in an argument, while summaries are used for entire pieces and focus on capturing the big picture of an argument. Both should be cited using the appropriate format (MLA, APA, etc.). See KU Writing Center guides on APA Formatting , Chicago Formatting , and MLA Formatting for more information.

When you paraphrase, you are using your own words to explain one of the claims of your source's argument, following its line of reasoning and its sequence of ideas. The purpose of a paraphrase is to convey the meaning of the original message and, in doing so, to prove that you understand the passage well enough to restate it. The paraphrase should give the reader an accurate understanding of the author's position on the topic. Your job is to uncover and explain all the facts and arguments involved in your subject. A paraphrase tends to be about the same length or a little shorter than the thing being paraphrased.

To paraphrase:

Alter the wording of the passage without changing its meaning. Key words, such as names and field terminology, may stay the same (i.e. you do not need to rename Milwaukee or osteoporosis), but all other words must be rephrased.
Retain the basic logic of the argument, sequence of ideas, and examples used in the passage.
Accurately convey the author's meaning and opinion.
Keep the length approximately the same as the original passage.
Do not forget to cite where the information came from. Even though it is in your own words, the idea belongs to someone else, and that source must be acknowledged.

A summary covers the main points of the writer’s argument in your own words. Summaries are generally much shorter than the original source, since they do not contain any specific examples or pieces of evidence. The goal of a summary is to give the reader a clear idea of what the source is arguing, without going into any specifics about what they are using to argue their point.

To summarize:

Identify what reading or speech is being summarized.
State the author’s thesis and main claims of their argument in your own words. Just like paraphrasing, make sure everything but key terms is reworded.
Avoid specific details or examples.
Avoid your personal opinions about the topic.
Include the conclusion of the original material.
Cite summarized information as well.

In both the paraphrase and summary, the author's meaning and opinion are retained. However, in the case of the summary, examples and illustrations are omitted. Summaries can be tremendously helpful because they can be used to encapsulate everything from a long narrative passage of an essay, to a chapter in a book, to an entire book.

When to Use Paraphrasing vs. Summarizing

Updated June 2022

Module 11: The Research Process—Using and Citing Sources

Paraphrasing, learning objective.

Describe when and how to paraphrase

There are two ways of integrating source material into your writing other than directly quoting from it: paraphrase and summary.

Paraphrasing and summarizing are similar. When we paraphrase, we process information or ideas from another person’s text and put it in our own words. The main difference between paraphrase and summary is scope : if summarizing means rewording and condensing, then paraphrasing means rewording without drastically altering length. However, paraphrasing is also generally more faithful to the spirit of the original; whereas a summary requires you to process and invites your own perspective, a paraphrase ought to mirror back the original idea using your own language.

What is Paraphrasing?

In a paraphrase, you use your own words to explain the specific points another writer has made. If the original text refers to an idea or term discussed earlier in the text, your paraphrase may also need to explain or define that idea. You may also need to interpret specific terms made by the writer in the original text.

Be careful not to add information or commentary that isn’t part of the original passage in the midst of your paraphrase. You don’t want to add to or take away from the meaning of the passage you are paraphrasing. Save your comments and analysis until after you have finished your paraphrase.

And be careful to remember that your paraphrase still requires a citation. Even when you use someone else’s ideas but put those ideas into your own words, you still need to acknowledge the source of those ideas!

What Does Good Paraphrasing Look Like?

In our first example, the writer is using MLA style to write a research essay for a literature class. Let’s compare two examples.

Example 1: While Gatsby is deeply in love with Daisy in The Great Gatsby , his love for her is indistinguishable from his love of his possessions (Callahan).
Example 2: John F. Callahan suggests in his article “F. Scott Fitzgerald’s Evolving American Dream” that while Gatsby is deeply in love with Daisy in The Great Gatsby , his love for her is indistinguishable from his love of his possessions (381).

In example 1, it’s hard to tell what exactly is being paraphrased. Is the entire source about the comparison between Gatsby’s love for Daisy and his love for his possessions? And, if this is the first or only reference to this particular piece of evidence in the research essay, the writer should include more information and signal phrasing about the source of this paraphrase in order to properly introduce it.

In example 2, the writer incorporates the names of the author and the title of the source material, so the credibility of the source material is now clear to the reader. Furthermore, because there is a page number at the end of this sentence, the reader understands that this passage is a paraphrase of a particular part of Callahan’s essay and not a summary of the entire essay.

Here’s another example, from an essay in APA style for a criminal justice class.

Example 1: Computer criminals have lots of ways to get away with credit card fraud (Cameron, 2002).
Example 2: Cameron (2002) points out that computer criminals intent on committing credit card fraud are able to take advantage of the fact that there aren’t enough officials working to enforce computer crimes. Criminals are also able to use technology to their advantage by communicating via email and chat rooms with other criminals.

In example 1, the main problem with this paraphrase is that it’s so vague and general that is doesn’t adequately explain to the reader what the point of the evidence really is or why that evidence matters. Remember: your readers have no way of automatically knowing why you as a research writer think that a particular piece of evidence is useful in supporting your point. This is why it is key that you introduce and explain your evidence.

The words "Think outside of the box" written in white chalk.

Figure 1 . Paraphrasing sources can sometimes be a tricky skill to develop, and often requires creative thinking. Use the practice examples below to fine-tune your paraphrasing skills.

In example 2, the writer includes additional information that introduces and explains the point of the evidence. In this particular example, the author’s name is also incorporated into the explanation of the evidence as well. In APA, it is preferable to weave in the author’s name into your essay, usually at the beginning of a sentence. However, it would also have been acceptable to end the paraphrase with just the author’s last name and the date of publication in parentheses.

What are the benefits of paraphrasing? It contextualizes the information (who said it, when, and where). It restates all the key points used by the source. Most importantly, it allows t he writer to maintain a strong voice while sharing important information from the source.

Paraphrasing is likely the most common way you will integrate your source information. Quoting should be minimal in most research papers.

Just remember, changing a few words here and there doesn’t count as a paraphrase. Students sometimes attempt to paraphrase by just changing a few words in the source material. The key is to understand the material put it into your own original words, but without changing the ideas. It’s tricky, but practice makes perfect!

Paraphrasing is a skill that takes time to develop. One way of becoming familiar with paraphrasing is by examining successful and unsuccessful attempts at paraphrasing. Read the quote below from page 179 of Howard Gardner’s book titled Multiple Intelligences and then examine the attempt at paraphrasing that follows.

“America today has veered too far in the direction of formal testing without adequate consideration of the costs and limitations of an exclusive emphasis on that approach.” [1]

A critical component of successful paraphrasing includes citing your original source. The citation may be made as an in-text citation, a footnote, or an endnote, but it must be included. You want to make clear and get credit for engaging with other thinkers in your work, and a correct citation foregrounds that strength. Failure to cite your sources is a violation of intellectual integrity (plagiarism), or taking someone else’s words or ideas and presenting them as your own. Sources should be always be cited in the appropriate way. Consider the following examples.

According to Levy (1997), the tutor-tool framework is useful.
According to Levy, the tutor-tool framework is useful.
Michael Levy, Computer-Assisted Language Learning: Context and Conceptualization (New York: Oxford), 178.

Take a look at the following paraphrased passages and determine if it is plagiarized or not plagiarized.

Gardner, Howard. Multiple Intelligences: New Horizons in Theory and Practice. BasicBooks, 2006 ↵
OWL at Excelsior College: Paraphrasing. Provided by : Excelsior College . Located at : https://owl.excelsior.edu/research/drafting-and-integrating/drafting-and-integrating-paraphrasing/ . Project : OWL at Excelsior College. License : CC BY: Attribution
Paraphrase Activity. Authored by : Excelsior OWL. Located at : https://owl.excelsior.edu/research/drafting-and-integrating/drafting-and-integrating-paraphrasing-activity/ . License : CC BY: Attribution
Avoiding Plagiarism Activity. Provided by : Excelsior OWL. Located at : https://owl.excelsior.edu/plagiarism/plagiarism-how-to-avoid-it/plagiarism-paraphrasing/plagiarism-paraphrasing-try-it-out/ . License : CC BY: Attribution
Image of writing in chalk. Authored by : Kaboompics. Provided by : Pexels. Located at : https://www.pexels.com/photo/think-outside-of-the-box-6375/ . License : Other . License Terms : https://www.pexels.com/photo-license/
Quoting examples from Paraphrasing in MLA Style. Authored by : Steven D. Krause. Located at : http://www.stevendkrause.com/tprw/chapter3.html . License : CC BY-NC-SA: Attribution-NonCommercial-ShareAlike
The Word on College Reading and Writing. Authored by : Carol Burnell, Jaime Wood, Monique Babin, Susan Pesznecker, and Nicole Rosevear. Located at : https://openoregon.pressbooks.pub/wrd/chapter/summarizing-a-text/ . License : CC BY-NC: Attribution-NonCommercial

Microsoft 365 Life Hacks > Writing > The Difference Between Summarizing & Paraphrasing

The Difference Between Summarizing & Paraphrasing

Summarizing and paraphrasing are helpful ways to include source material in your work without piling on direct quotes. Understand the differences between these approaches and when to use each.

Summarizing vs. Paraphrasing: The Biggest Differences

Though summarizing and paraphrasing are both tools for conveying information clearly and concisely, they help you achieve this in different ways. In general, the difference is rooted in the scale of the source material: To share an entire source at once, you summarize; to share a specific portion of a source (without quoting directly, of course), you paraphrase.

Write with Confidence using Editor

Elevate your writing with real-time, intelligent assistance

What is Summarizing?

Summarizing is simplifying the content of a source to its main points in your own words. You literally sum up something, distill it down to its most essential parts. Summaries cover whole sources rather than a piece or pieces of a source and don’t include direct quotes or extraneous detail.

How to Summarize

Understand the original thoroughly. You may start by scanning the original material, paying close attention to headers and any in-text summaries, but once you’re sure that this source is something you’re going to use in your research paper , review it more thoroughly to gain appropriate understanding and comprehension.
Take notes of the main points. A bulleted list is appropriate here-note the main idea of each portion of the source material. Take note of key words or phrases around which you can build your summary list and deepen your understanding.
Build your summary. Don’t just use the list you’ve already created—this was a first draft . Craft complete sentences and logical progression from item to item. Double check the source material to ensure you’ve not left out any relevant points and trim anything extraneous. You can use a bulleted or numbered list here or write your summary as a paragraph if that’s more appropriate for your use. Make sure to follow the rules of parallelism if you choose to stay in list form.

What is Paraphrasing?

Paraphrasing is rephrasing something in your own words; the word comes from the Greek para -, meaning “beside” or “closely resembling”, 1 combined with “phrase,” which we know can mean a string of words or sentences. 2 Paraphrasing isn’t practical for entire sources—just for when you want to highlight a portion of a source.

How to Paraphrase

Read actively . Take notes, highlight or underline passages, or both if you please-whatever makes it easiest for you to organize the sections of the source you want to include in your work.
Rewrite and revise. For each area you’d like to paraphrase, take the time to rewrite it in your own words. Retain the meaning of the original text, but don’t copy it too closely; take advantage of a thesaurus to ensure you’re not relying too heavily on the source material.
Check your work and revise again as needed . Did you retain the meaning of the source material? Did you simplify the language of the source material? Did you differentiate your version enough? If not, try again.

Summarizing and paraphrasing are often used in tandem; you’ll likely find it appropriate to summarize an entire source and then paraphrase specific portions to support your summary. Using either approach for including sources requires appropriate citing, though, so ensure that you follow the correct style guide for your project and cite correctly.

Get started with Microsoft 365

It’s the Office you know, plus the tools to help you work better together, so you can get more done—anytime, anywhere.

Topics in this article

What is independent publishing?

Avoid the hassle of shopping your book around to publishing houses. Publish your book independently and understand the benefits it provides for your as an author.

What are literary tropes?

Engage your audience with literary tropes. Learn about different types of literary tropes, like metaphors and oxymorons, to elevate your writing.

What are genre tropes?

Your favorite genres are filled with unifying tropes that can define them or are meant to be subverted.

What is literary fiction?

Define literary fiction and learn what sets it apart from genre fiction.

Everything you need to achieve more in less time

Get powerful productivity and security apps with Microsoft 365

Explore Other Categories

What is the difference between paraphrasing & summarizing

Differences between paraphrasing & summarizing, definition and purpose.

Paraphrasing involves rewriting someone else's ideas or a specific text in your own words while maintaining the original meaning and often keeping a similar length to the source material. The primary purpose of paraphrasing is to use another person's ideas in your work without resorting to a direct quotation, thereby showing your understanding of the source while integrating it smoothly into your own narrative.

Summarizing , on the other hand, is the process of distilling a longer piece of text down to its essential points, significantly reducing its length. The goal of summarizing is to provide a broad overview of the source material, capturing only the main ideas in a concise manner, which helps in clarifying the overall theme or argument of the text for the reader.

Detail and Length

The level of detail and the length of the text are key differences between paraphrasing and summarizing. Paraphrasing retains a level of detail similar to the original text, and the paraphrased passage is typically about the same length or slightly shorter than the source. This approach is suitable when specific details or points from the source are necessary to understand the reader in the context of your work.

In contrast, summarizing significantly reduces the length of the original text, focusing only on the central themes or main ideas. This makes summaries particularly useful for giving an overview of long texts, such as books, articles, or comprehensive reports, where only the core message is needed.

Usage in Academic Writing

Both paraphrasing and summarizing are crucial skills in academic writing. They help one effectively incorporate the ideas of others into one's work. Paraphrasing is often preferred when specific evidence or a detailed understanding of the source is required to support one's argument without overusing direct quotes, which can clutter the text and disrupt the flow of the narrative.

Summarizing is a time-saving and efficient strategy when referring to broader concepts or discussing a source's general scope. It's particularly useful during literature reviews or when providing background information where detailed support is unnecessary. Allowing you to succinctly conveys the essence of a sourcerelieves you from overwhelming your readers with unnecessary details while still ensuring they grasp the relevance of the referenced works.

In summary, choosing between paraphrasing and summarizing depends mainly on the writer's intent, the importance of the details in the source material, and how they wish to integrate this information into their own writing.

How to paraphrase in a few steps

Paraphrasing refers to rewriting content in our own words while keeping the original meaning. Here are some steps and tips for effective paraphrasing:

Reading & Understanding the content

Ensure you fully understand the meaning of the text while identifying the essential concepts.

Taking Notes

Write the main ideas without looking at the original text.

Using Synonyms

This step involves replacing words with relevant synonyms; keeping some technical words that can't be replaced is essential.

Changing Sentence Structure

Alter the sentence structure, such as changing active to passive voice or shortening long sentences.

Rephrasing Concepts

Explain the concepts differently, using your own words and style.

Comparing with the original content

Ensure your paraphrased version accurately reflects the original meaning and is not too similar.

Cite the Source

It's essential to credit the source even when paraphrased.

The original text:

Students frequently overuse direct quotations in taking notes, and as a result, they overuse quotations in the final [research] paper. Probably only about 10% of your final manuscript should appear as a directly quoted matter. Therefore, you should strive to limit the exact transcribing of source materials while taking notes. Lester, James D. Writing Research Papers. 2nd ed., 1976, pp. 46-47.

The paraphrased text:

In research papers, students often quote excessively, failing to keep quoted material down to a desirable level. Since the problem usually originates during note-taking, it is essential to minimize the material recorded verbatim (Lester 46-47).

Here are some tips to paraphrase:

Avoid Plagiarism: Don’t copy the text verbatim without quotation marks and proper attribution.
Maintain Original Meaning: Ensure the paraphrased text conveys the same message as the original.
Use a Different Structure: Change the information order and rephrase sentences.
Simplify the Language: Use more straightforward language to make the text more understandable, if appropriate.

Common paraphrasing mistakes

Not changing enough to avoid plagiarism

One of the most complex parts of paraphrasing a sentence is changing enough to avoid copying and not lose the original meaning. This can be a tricky balancing act, especially if you have to keep some of the wording.

Distorting the meaning

Likewise, changing the words and sentence structure can accidentally change the meaning. That’s fine if you want to write an original sentence, but if you’re trying to convey someone else’s idea, it's best to rewrite and adequately describe it.

Review your paraphrase to confirm that all the words are correctly used and placed in the correct order for your intended meaning. If unsure, you can ask someone to read the passage to see how they interpret it.

Forgetting the citation

Some people think that if you put an idea into your own words, you don’t need to cite where it came from—but that’s not true. Even if the wording is your own, the ideas are not. That means you need a citation.

If you have many paraphrased sentences from the exact location in a source, you need only one citation at the end of the passage. Otherwise, you need a citation for each paraphrased sentence from another source in your writing, without exception.

How to summarize in a few steps

Focusing on the main ideas.

Read through the entire piece you want to summarize and identify the most important concepts and themes. Ignore minor details and examples. Focus on capturing the essence of the critical ideas.

If it's an article or book, read introductions, headings, and conclusions to understand the central themes. As you read, ask yourself, "What is the author trying to convey here?" to determine what's most significant.

Keeping it short and straightforward

A summary should be considerably shorter than the original work. Aim for about 1/3 of the length or less. Be concise by eliminating unnecessary words and rephrasing ideas efficiently. Use sentence fragments and bulleted lists when possible.

Maintaining objectivity

Summarize the work factually without putting your own personal spin or opinions on the information. Report the key ideas in an impartial, balanced manner. Do not make judgments about the quality or accuracy of the content.

Using a summarizing tool

As AI continues gaining popularity, leveraging dedicated to benefit from their advantage is essential.

Among these tools lies Wiseone , an innovative AI tool that transforms how we read and search for information online.

Wiseone's "Summarize" feature allows you to understand the main points of an article or a PDF document efficiently without the need to read the entire piece by generating thorough summaries with key takeaways.

Don’ts of paragraph summarization

Similarly, keep these in mind as things to avoid:

Plagiarizing the original paragraph. It’s perfectly fine to include direct quotes, but if you do, cite them properly. However, most of the summary should be in your own words.
Paraphrasing rather than summarizing. Here’s a way to think of the difference: a summary is a “highlight reel,” and paraphrasing condenses the entire paragraph.
Omitting key information. When summarizing a paragraph, you might need to mention information from its preceding or following paragraphs, or even other sections from the original work, to give the reader appropriate context for the other information included in the summary.

What is paraphrasing?

What is summarizing.

Summarizing is the process of distilling a longer text down to its essential points, significantly reducing its length. The goal of summarizing is to provide a broad overview of the source material, capturing only the main ideas in a concise manner, which helps in clarifying the overall theme or argument of the text for the reader.

What are the steps to paraphrase?

Reading & Understanding the content: Ensure you fully understand the meaning of the text while identifying the essential concepts.
Taking Notes: Write the main ideas without looking at the original text.
Using Synonyms: This step involves replacing words with relevant synonyms; keeping some technical words that can't be replaced is essential.
Changing Sentence Structure: Alter the sentence structure, such as changing active to passive voice or shortening long sentences.
Rephrasing Concepts: Explain the concepts differently, using your own words and style.
Comparing with the original content: Ensure your paraphrased version accurately reflects the original meaning and is not too similar.
Cite the Source: It's essential to credit the source even when paraphrased.

What are some tips to paraphrase?

What are the common paraphrasing mistakes.

Not changing enough to avoid plagiarism: One of the most complex parts of paraphrasing a sentence is changing enough to avoid copying and not lose the original meaning. This can be a tricky balancing act, especially if you have to keep some of the wording.
Distorting the meaning: Changing the words and sentence structure can accidentally change the meaning. That’s fine if you want to write an original sentence, but if you’re trying to convey someone else’s idea, it's best to rewrite and adequately describe it. Review your paraphrase to confirm that all the words are correctly used and placed in the correct order for your intended meaning. If unsure, you can ask someone to read the passage to see how they interpret it.
Forgetting the citation: Some people think that if you put an idea into your own words, you don’t need to cite where it came from—but that’s not true. Even if the wording is your own, the ideas are not. That means you need a citation.

What are the steps to summarize content?

Focusing on the main ideas : Read through the entire piece you want to summarize and identify the most important concepts and themes. Ignore minor details and examples. Focus on capturing the essence of the critical ideas. If it's an article or book, read introductions, headings, and conclusions to understand the central themes. As you read, ask yourself, "What is the author trying to convey here?" to determine what's most significant.
Keeping it short and straightforward: A summary should be shorter than the original work. Aim for about 1/3 of the length or less. Be concise by eliminating unnecessary words and rephrasing ideas efficiently. Use sentence fragments and bulleted lists when possible.
Maintaining objectivity: Summarize the work factually without putting your own personal spin or opinions on the information. Report the key ideas in an impartial, balanced manner. Do not make judgments about the quality or accuracy of the content.

Enhance your web search, Boost your reading productivity with Wiseone

The Sheridan Libraries

Avoiding Plagiarism
Sheridan Libraries

Paraphrasing & Summarizing

What is Plagiarism?
School Plagiarism Policies
Common Knowledge
Minimizing Your Plagiarism Risk
Student Help
Helping Prevent Plagiarism in Your Classroom
Avoiding Plagiarism Course
Course FAQs

To help the flow of your writing, it is beneficial to not always quote but instead put the information in your own words. You can paraphrase or summarize the author’s words to better match your tone and desired length. Even if you write the ideas in your own words, it is important to cite them with in-text citations or footnotes (depending on your discipline’s citation style ).

Definitions

Paraphrasing allows you to use your own words to restate an author's ideas.
Summarizing allows you to create a succinct, concise statement of an author’s main points without copying and pasting a lot of text from the original source.

What’s the difference: Paraphrasing v. Summarizing

Explore the rest of the page to see how the same material could be quoted, paraphrased, or summarized. Depending on the length, tone, and argument of your work, you might choose one over the other.

Bad Paraphrase
Good Paraphrase
Reread: Reread the original passage until you understand its full meaning.
Write on your own: Set the original aside, and write your paraphrase on a note card.
Connect: Jot down a few words below your paraphrase to remind you later how you envision using this material.
Check: Check your rendition with the original to make sure that your version accurately expresses all the essential information in a new form.
Quote: Use quotation marks to identify any unique term or phraseology you have borrowed exactly from the source.
Cite: Record the source (including the page) on your note card or notes document so that you can credit it easily if you decide to incorporate the material into your paper.

Explore the tabs to see the difference between an acceptable and unacceptable paraphrase based on the original text in each example.

Original Text

“Business communication is increasingly taking place internationally – in all countries, among all peoples, and across all cultures. An awareness of other cultures – of their languages, customs, experiences and perceptions – as well as an awareness of the way in which other people conduct their business, are now essential ingredients of business communication” (Chase, O’Rourke & Wallace, 2003, p.59).

More and more business communication is taking place internationally—across all countries, peoples, and cultures. Awareness of other cultures and the way in which people do business are essential parts of business communication (Chase, O’Rourke & Wallace, 2003, p.59)

Compare the Original and Paraphrase

Too much of the original is quoted directly, with only a few words changed or omitted. The highlighted words are too similar to the original quote:

More and more business communication is taking place internationally —across all countries, peoples, and cultures . Awareness of other cultures and the way in which people do business are essential parts of business communication (Chase, O’Rourke & Wallace, 2003, p.59)

Original Text

The importance of understanding the traditions, language, perceptions, and the manner in which people of other cultures conduct their business should not be underestimated, and it is a crucial component of business communication (Chase, O’Rourke & Wallace, 2003, p. 59).

The original’s ideas are summarized and expressed in the writer’s own words with minimal overlap with the original text's language:

Bad Summary
Good Summary
Find the main idea: Ask yourself, “What is the main idea that the author is communicating?”
Avoid copying: Set the original aside, and write one or two sentences with the main point of the original on a note card or in a notes document.
Connect: Jot down a few words below your summary to remind you later how you envision using this material.

Business communication is worldwide, and it is essential to build awareness of other cultures and the way in which other people conduct their business. (Chase, O’Rourke & Wallace, 2003, p.59).

Compare the Original and Summary

Too much of the original is quoted directly, with only a few words changed or omitted. The highlighted words are too similar to the original text:

Business communication is worldwide, and it is essential to build awareness of other cultures and the way in which other people conduct their business . (Chase, O’Rourke & Wallace, 2003, p.59).

In a world that is increasingly connected, effective business communication requires us to learn about other cultures, languages, and business norms (Chase, O’Rourke & Wallace, 2003, p.59).

The original’s ideas are summarized and expressed in the writer’s own words with minimal overlap:

In a world that is increasingly connected, effective business communication requires us to learn about other cultures , languages , and business norms (Chase, O’Rourke & Wallace, 2003, p.59).

No matter what the source or style, you need to cite it both in-text and at the end of the paper with a full citation! Write down or record all the needed pieces of information when researching to ensure you avoid plagiarism.

Cheat Sheet

Paraphrasing and Summarizing Download this helpful cheat sheet covering "Paraphrasing and Summarizing."
<< Previous: Quoting
Next: Minimizing Your Plagiarism Risk >>
Last Updated: Aug 7, 2023 2:00 PM
URL: https://guides.library.jhu.edu/avoidingplagiarism

Writing Center FAQs

Capella FAQs Home
Capella FAQs
Writing Center

Q: What is the difference between summarizing and paraphrasing?

Career Center
Disability Support
Doctoral Support
Learner Records
Military Support
Office of Research & Scholarship
Quantitative Skills Center
Scholarships & Grants
Technical Support
13 APA Style
16 Composition
8 References
22 Resources
24 Writing Center
6 Writing Studio

You Search Writing Center FAQs

Search All FAQs

Answer Last Updated: May 22, 2017 Views: 411

Summarizing and paraphrasing are very similar. Both involve taking ideas, words, or phrases from a source and crafting them into new sentences within your writing.

Summarizing is when you briefly state the main concept/s of a topic in your own words.

Paraphrasing is when you briefly recount a specific passage from a piece in your own words.

Whether summarizing or paraphrasing, credit is always given to the author.

Share on Facebook

Was this helpful? Yes 0 No 0

Still have questions? Email us. [email protected]

Summarizing vs. Paraphrasing: What’s the Real Difference?

By Joshua Turner

September 6, 2023

Summarizing and paraphrasing are two essential skills in writing. They are often used interchangeably, but they have distinct differences. Summarizing is the process of condensing a text into a shorter version, highlighting the main points, and leaving out the details.

On the other hand, paraphrasing is rewording a text in your own words, retaining the original meaning and message.

Understanding summarizing involves identifying the key ideas and concepts in a text and presenting them in a concise and clear manner. It requires a good understanding of the text and the ability to distinguish between essential and non-essential information.

Summarizing is useful when you want to provide a brief overview of a longer text or when you want to highlight the main ideas.

Understanding paraphrasing involves rewording a text in a way that retains the original meaning but uses different words and sentence structures.

It requires a good understanding of the text and the ability to express the ideas in your own words. Paraphrasing is useful when you want to avoid plagiarism or when you want to clarify the meaning of a text.

Key Takeaways

Summarizing involves condensing a text into a shorter version, highlighting the main points and leaving out the details.
Paraphrasing involves rewording a text in a way that retains the original meaning but uses different words and sentence structures.
Summarizing is useful when you want to provide a brief overview of a longer text, while paraphrasing is useful when you want to avoid plagiarism or clarify the meaning of a text.

Definition of Summarizing

Summarizing is the process of condensing a longer piece of text into a shorter, more concise version while retaining the main points and key concepts. It involves creating an overview of the text that captures the gist of the original content.

Purpose of Summaries

The purpose of summaries is to provide readers with a condensed version of a longer text that highlights the main points and key concepts. Summaries are useful for quickly understanding the content of a longer piece of writing, such as an article or book, without having to read the entire text.

Main Points in Summarizing

The main points in summarizing include identifying the key concepts and ideas in the original text, condensing the information into a shorter version, and ensuring that the summary accurately represents the main points of the original text.

Steps in Summarizing

The steps in summarizing include reading the original text carefully, identifying the main points and key concepts, condensing the information into a shorter version, and reviewing the summary to ensure that it accurately represents the main points of the original text. It is important to use your own words when creating a summary and to avoid copying phrases or sentences directly from the original text.

In summary, summarizing is the process of condensing a longer piece of text into a shorter, more concise version while retaining the main points and key concepts. It involves creating an overview of the text that captures the gist of the original content. The purpose of summaries is to provide readers with a condensed version of a longer text that highlights the main points and key concepts.

Understanding Paraphrasing

Paraphrasing is the act of rephrasing a text in your own words while maintaining the original meaning. It is an essential skill in academic writing , as it allows you to incorporate information from other sources while avoiding plagiarism. Paraphrasing involves interpreting the main ideas in the original text and presenting them in your own voice.

Purpose of Paraphrases

The purpose of paraphrasing is to present information from other sources in a way that is more accessible or relevant to your intended audience. It also allows you to integrate information from multiple sources into a cohesive argument. Paraphrasing can also help you to clarify complex ideas and concepts.

Main Ideas in Paraphrasing

The main ideas in paraphrasing are to understand the original text, interpret the main ideas, and rephrase them in your own words. It is important to maintain the original meaning and avoid changing the author’s intended message. Paraphrasing should also be done in your own voice to avoid plagiarism.

Steps in Paraphrasing

The steps in paraphrasing include reading and understanding the original text, identifying the main ideas, and rephrasing them in your own words. You should also check your paraphrase against the original text to ensure that you have maintained the original meaning. It is also important to cite the original source to avoid plagiarism.

Comparison of Summarizing and Paraphrasing

Summarizing and paraphrasing are two different techniques used to convey information from one source to another.

Length and Detail

Summarizing involves condensing a large amount of information into a concise version while maintaining the main points. On the other hand, paraphrasing involves rephrasing the text in your own words while retaining the original meaning. Summaries are shorter than the original text and omit details, while paraphrases are usually the same length as the original text and include more details.

Quoting and Citation

When summarizing, you don’t need to use direct quotes or citations because you are putting the information into your own words. However, when paraphrasing, you still need to give credit to the original source by using citations and quotation marks when necessary.

Structure and Concepts

Summarizing involves restructuring the original text to make it more concise, while paraphrasing involves rewording the original text. Summarizing focuses on the main points while paraphrasing focuses on the details.

When summarizing, you may need to rearrange the concepts to make them more understandable, while paraphrasing may require you to explain the concepts more clearly.

The audience and purpose of the text can influence whether summarizing or paraphrasing is appropriate. Summarizing is useful when the audience needs a quick overview of the main points, while paraphrasing is useful when the audience needs a more detailed understanding of the text. The purpose of the text can also determine whether summarizing or paraphrasing is appropriate. Summarizing is useful when the purpose is to provide a brief overview, while paraphrasing is useful when the purpose is to explain the details.

Avoiding Plagiarism

Using someone else’s work without proper credit is not only unethical, but it can also have serious consequences. By understanding plagiarism, citing your source material, and using a plagiarism checker, you can ensure that your work is original and free of plagiarism.

Understanding Plagiarism

Plagiarism is the act of using someone else’s work without giving them proper credit. It can be intentional or unintentional, and it can have serious consequences. To avoid plagiarism, understand what it is and how to avoid it.

Citing Source Material

Citing your source material is an essential part of avoiding plagiarism. When you use someone else’s work, you must give them credit by citing the original source. There are different citation styles, such as APA, MLA, and Chicago, so make sure to use the appropriate one for your work.

Using a Plagiarism Checker

Using a plagiarism checker is a great way to ensure that your work is original and free of plagiarism. There are many free and paid tools available online that can help you check your work for plagiarism. These tools compare your work to other sources on the internet and highlight any similarities.

In summary, while summarizing and paraphrasing are similar in that they both involve condensing or rewording information, there are some key differences between them. Summarizing involves reducing a text to its essential points, while paraphrasing involves restating the central idea in your own words.

Accuracy is crucial in both cases, but it is especially important when paraphrasing since it involves conveying information in a new way. Paraphrasing is useful when you want to highlight specific insights or takeaways from a text while summarizing is better suited for providing an overview of the essential information.

When deciding whether to summarize or paraphrase, it’s important to consider the function of the text and the audience you are writing for. Summarizing is useful when you want to provide a quick overview of a text’s most relevant information, while paraphrasing is better suited for conveying the central idea in a new way.

Overall, whether you choose to summarize or paraphrase, the goal is to convey relevant information in a clear and concise manner that helps the reader gain insights and takeaways from the text.

Frequently Asked Questions

Here are some common questions about this topic.

What are some examples of paraphrasing and summarizing, and how do they differ?

Paraphrasing involves restating a passage in your own words while summarizing involves condensing a larger text into a shorter version. For example, paraphrasing a quote in an essay would involve rephrasing it in a way that still conveys the original meaning, while summarizing a news article would involve highlighting the main points in a few sentences.

What are the similarities and differences between summarizing and paraphrasing?

Both summarizing and paraphrasing involve rephrasing information in your own words. However, summarizing involves condensing a larger text into a shorter version, while paraphrasing involves restating a passage in your own words. Both techniques are useful for avoiding plagiarism and presenting information in a clear and concise way.

How do you paraphrase a quote in an essay?

To paraphrase a quote in an essay, you should rephrase the quote in your own words while still maintaining its original meaning. This involves understanding the main idea of the quote and expressing it in a way that fits with the rest of your essay. It is important to properly cite the original source of the quote to avoid plagiarism.

When using a source, should you quote, paraphrase, or summarize it?

The choice between quoting, paraphrasing, or summarizing a source depends on the purpose of your writing. If you want to include a specific passage word-for-word, you should quote it. If you want to restate a passage in your own words, you should paraphrase it. If you want to condense a larger text into a shorter version, you should summarize it.

What is the definition of summarizing?

Summarizing is the act of condensing a larger text into a shorter version that highlights the main points of the original. This technique is useful for presenting information in a clear and concise way and can be applied to a variety of texts, such as news articles, research papers, and books.

The Ultimate Guide: What Is the Best Definition of Sobriety?

Psychoanalysis vs. behavior therapy: what’s the difference, unlocking the power of music: how it shapes your behavior, stop guessing here’s which phase of strategic conflict management you can ignore.

Purdue Online Writing Lab Purdue OWL® College of Liberal Arts

Quoting, Paraphrasing, and Summarizing

Welcome to the Purdue OWL

This page is brought to you by the OWL at Purdue University. When printing this page, you must include the entire legal notice.

Copyright ©1995-2018 by The Writing Lab & The OWL at Purdue and Purdue University. All rights reserved. This material may not be published, reproduced, broadcast, rewritten, or redistributed without permission. Use of this site constitutes acceptance of our terms and conditions of fair use.

This handout is intended to help you become more comfortable with the uses of and distinctions among quotations, paraphrases, and summaries. This handout compares and contrasts the three terms, gives some pointers, and includes a short excerpt that you can use to practice these skills.

What are the differences among quoting, paraphrasing, and summarizing?

These three ways of incorporating other writers' work into your own writing differ according to the closeness of your writing to the source writing.

Quotations must be identical to the original, using a narrow segment of the source. They must match the source document word for word and must be attributed to the original author.

Paraphrasing involves putting a passage from source material into your own words. A paraphrase must also be attributed to the original source. Paraphrased material is usually shorter than the original passage, taking a somewhat broader segment of the source and condensing it slightly.

Summarizing involves putting the main idea(s) into your own words, including only the main point(s). Once again, it is necessary to attribute summarized ideas to the original source. Summaries are significantly shorter than the original and take a broad overview of the source material.

Why use quotations, paraphrases, and summaries?

Quotations, paraphrases, and summaries serve many purposes. You might use them to:

Provide support for claims or add credibility to your writing
Refer to work that leads up to the work you are now doing
Give examples of several points of view on a subject
Call attention to a position that you wish to agree or disagree with
Highlight a particularly striking phrase, sentence, or passage by quoting the original
Distance yourself from the original by quoting it in order to cue readers that the words are not your own
Expand the breadth or depth of your writing

Writers frequently intertwine summaries, paraphrases, and quotations. As part of a summary of an article, a chapter, or a book, a writer might include paraphrases of various key points blended with quotations of striking or suggestive phrases as in the following example:

In his famous and influential work The Interpretation of Dreams , Sigmund Freud argues that dreams are the "royal road to the unconscious" (page #), expressing in coded imagery the dreamer's unfulfilled wishes through a process known as the "dream-work" (page #). According to Freud, actual but unacceptable desires are censored internally and subjected to coding through layers of condensation and displacement before emerging in a kind of rebus puzzle in the dream itself (page #).

How to use quotations, paraphrases, and summaries

Practice summarizing the essay found here , using paraphrases and quotations as you go. It might be helpful to follow these steps:

Read the entire text, noting the key points and main ideas.
Summarize in your own words what the single main idea of the essay is.
Paraphrase important supporting points that come up in the essay.
Consider any words, phrases, or brief passages that you believe should be quoted directly.

There are several ways to integrate quotations into your text. Often, a short quotation works well when integrated into a sentence. Longer quotations can stand alone. Remember that quoting should be done only sparingly; be sure that you have a good reason to include a direct quotation when you decide to do so. You'll find guidelines for citing sources and punctuating citations at our documentation guide pages.

What’s the Difference? Summarizing, Paraphrasing, & Quoting

Posted on November 29, 2023 November 29, 2023

What’s the Difference? Summarizing , Paraphrasing , & Quoting

Quoting, paraphrasing , and summarizing are three methods for including the ideas or research of other writers in your own work. In academic writing , such as essay writing or research papers , it is often necessary to utilize other people’s writing.

Outside sources are helpful in providing evidence or support written claims when arguing a point or persuading an audience. Being able to link the content of a piece to similar points made by other authors illustrates that one’s writing is not based entirely off personal thoughts or opinions and has support found from other credible individuals. In scientific work such as reports or experiment related writing, being able to point to another published or peer-reviewed writer can strengthen your personal research and even aid in explaining surprising or unusual findings. In all situations, referencing outside sources also elevates the integrity and quality of your work.

When pulling information from an outside source it is critical to properly use quotations, paraphrasing , or summarizing to avoid plagiarizing from the original passage . Plagiarism is portraying another’s work, ideas, and research as one’s own, and is an extremely serious disciplinary offense. Without using proper quotations, paraphrasing and summarizing , it can be easy to unintentionally plagiarize from the original source . Including citations that reference the author also helps ensure proper credit is given, and no accidental plagiarism occurs. Regardless of if APA , MLA or Chicago style are used, a citation must accompany the work of another author.

This article will compare these three concepts, to help users become more comfortable with each of them and the differing scenarios to utilize each. The article will also provide examples and give pointers to further increase familiarity with these essential techniques and prevent the happening of plagiarism .

What is Quoting?

Quoting is the restatement of a phrase, sentence, thought, or fact that was previously written by another author. A proper direct quotation includes the identical text without any words or punctuation adjusted.

One might use a quotation when they want to use the exact words from the original author , or when the author has introduced a new concept or idea that was of their conception. Oftentimes, the author already used concise, well-thought-out wording for an idea and it may be difficult to restate without using a direct quote .

However when repeating content from someone else’s work, one must use quotation marks with a corresponding citation or it will be considered plagiarism . The proper citation may also vary based on the citation style being used.

Examples of Quoting

In order to further the understanding of how to utilize quotes, some examples of incorrect and correct quotation are provided below.

Original Text: As natural selection acts solely by accumulating slight, successive, favorable variations, it can produce no great or sudden modification; it can act only by very short and slow steps

Incorrect Quotation Example: “Because natural selection acts only by accumulating slight, successive favorable variations. It can produce no greater or sudden modification and can only act by very short and slow steps

Correct Quotation Example: “As natural selection acts solely by accumulating slight, successive, favorable variations, it can produce no great or sudden modification; it can act only by very short and slow steps,” (Darwin 510).

The bad example provided does not include the identical text or identical grammar and punctuation to that of the original source . The quote is also lacking one quotation mark and a citation to attribute the initial author. Meanwhile, the good example i s completely identical to the original text and features a correct citation, making it a great example of a quote in use.

What is Paraphrasing ?

Paraphrasing is taking the written work, thoughts, or research of another author and putting it in one’s own words . Correct paraphrasing is done through the restatement of key ideas from another person’s work, but utilizing different words to avoid copying them. Oftentimes, finding synonyms to the words used by the original author helps to paraphrase .

One would use paraphrasing when they hope to capture the key points of a written work in their own writing . Paraphrasing should also be employed when the content of the original source is more important than the wording used. This writing technique is a good strategy to maintain one’s personal writing style throughout a written work.

Similar to quoting, even paraphrased material should be accompanied by the proper citation to avoid plagiarizing the initial author.

Examples of Paraphrasing

Original Content: The Statue of Liberty, one of the most recognizable symbols of freedom and democracy across the world, was a gift of friendship to America from France. Inaugurated in 1886, the statue is 305 feet tall and represents Libertas, the Roman liberty goddess, bearing a torch in her right hand and a tablet in her left hand with the date of the US Declaration of Independence. Broken shackles lay underneath the statue’s drapery, to symbolize the end of all types of servitude and oppression.

Incorrect Paraphrasing Example: The Statue of Liberty is an evident display of freedom and democracy for the whole world, and was created by France for America to represent their friendship. The 305 foot statue of the Roman liberty goddess Libertas was installed in 1886. The Statue of Liberty has a tablet with the US Declaration of Independence date in one hand and a torch in her other. She also has broken shackles on the ground to represent an end to enslavement and oppression.

Correct Paraphrasing Example: France presented the United States with the Statue of Liberty in 1886 to commemorate the two countries friendship. The Roman goddess of liberty, Libertas, stands 305 feet tall as a well-known tribute to freedom and democracy. The statue commemorates the US Declaration of Independence though the tablet in her left hand that accompanies a torch in her right. The Statue of Liberty also celebrates an end to oppression and servitude, indicated by broken chains by her feet ( Diaz, 2019 ).

The incorrect example provided featured a sentence structure that followed too closely to that of the original text. Additionally, the writer only swapped out a few words for very common synonyms so the paraphrased content is ultimately too similar to the original text. An academic work that used this paraphrase would be cited for plagiarism .

On the other hand, the correct example featured paraphrased content that is properly cited, with variety to the sentence structure and text that includes words beyond just synonyms to words in the original content. This example also contains the main ideas, but is ultimately slightly condensed from the original text.

What Is Summarizing ?

Summarizing is providing a brief description of the key ideas from a written work. This description should be in one’s own writing , and is typically significantly shorter than the source material because it only touches on the main points .

Summaries are commonly used when a writer hopes to capture the central idea of a work, without relying on the specific wording that the original author used to explain the idea. They also can provide a background or overview of content needed to understand a topic being discussed. This strategy still captures the meaning of the original text without straying from one’s personal tone and writing style.

Unlike paraphrasing and quoting, a summary does not require an in- text citation and only occasionally needs accreditation to the original writer’s work .

Examples of Summarizing

In order to further the understanding of how to summarize content in your writing, some examples of incorrect and correct summaries for the short children’s story Goldilocks and The Three Bears are provided below.

Incorrect Summary Example: Once upon a time, Goldilocks went for a walk on the beach when she saw a house and went in it. In the house she found three bowls of soup and decided to try them all, but one was too hot, one was too cold and one was just right. Next, Goldilocks tried to sit in three different chairs but only found one that fit her perfectly. Lastly, she went to the back of the house and found three beds. Just like the soup and chairs she tested all of them before picking one that she liked the best and taking a nice long nap. The End.

Correct Summary Example: In Goldilocks and the Three Bears by Robert Southy, a young girl wanders into the house of three bears where she tastes three different porridges; sits in three different chairs; and naps in three different beds before finding one of each that fits her. Goldilocks is eventually found by the bears who are upset about her intrusion and usage of their personal belongings.

The incorrect example provided would not be considered a good summary for a few reasons. Primarily, this summary does not summarize well, as provides too much unnecessary detail and an individual would still be able to comprehend the main point of the story without it. The summary also ends without touching on the most important point , which is the lesson of the story. This summary also provides inaccurate information, and lacks a citation.

Meanwhile, the correct example is a good summary because it does not spend too much time on any certain aspect of the story. The reader is still able to understand exactly what happens to Goldilocks without consuming any non-essential details. This summary also provides completely accurate information and touches on the main point or lesson from the story.

Differences and Similarities

There are a few major differences and similarities between the three writing techniques discussed.

Quoting, paraphrasing , and summarizing are similar in that they are all writing techniques that can be used to include the work of other authors in one’s own writing . It is common for writers to use these strategies collectively in one piece to provide variety in their references and across their work. These three strategies also share the similarity of helping to prevent plagiarizing the content from the original source . All three of these methods require some form of citation and attribution to the original author to completely avoid plagiarizing.

Oppositely, the main difference between quoting, paraphrasing , and summarizing is that quoting is done word for word from the original work . Both paraphrasing and summarizing only touch on the key points and are written with some variation from the initial author’s work , usually in the style and tone of the new author. When comparing just the latter two, paraphrased material tends to be closer in length to the actual material, because it only slightly condenses the original passage . On the other hand, a summary is most likely significantly shorter than the original author’s work since this method only pulls from the most important points .

Final Thoughts

It is extremely common to utilize the previous writing of others, especially in academic writing . These original works enhance the quality and honesty of one’s work while also providing backing and emphasis to the points made.

Quoting, paraphrasing , and summarizing are all strategies for incorporating the thoughts, ideas, research, and writing from another author in one’s own work. The three methods explained are also safe strategies to employ to avoid accidental plagiarism of the original passage .

Another strategy to ensure one’s writing is properly quoted, paraphrased, and summarized is by using a plagiarism checker. Quetext provides an easy-to-use plagiarism checker that verifies the originality of work and can create citations for any sources cited throughout the paper.

Sign Up for Quetext Today!

Click below to find a pricing plan that fits your needs.

A Safer Learning Environment: The Impact of AI Detection on School Security

Posted on May 17, 2024

Rethinking Academic Integrity Policies in the AI Era

Posted on May 10, 2024 May 10, 2024

Jargon Phrases to Avoid in Business Writing

Posted on May 3, 2024 May 3, 2024

Comparing Two Documents for Plagiarism: Everything You Need to Know

Posted on April 26, 2024 April 26, 2024

Mastering Tone in Email Communication: A Guide to Professional Correspondence

Posted on April 17, 2024

Mastering End-of-Sentence Punctuation: Periods, Question Marks, Exclamation Points, and More

Posted on April 12, 2024

Mastering the Basics: Understanding the 4 Types of Sentences

Posted on April 5, 2024

Can You Go to Jail for Plagiarism?

Posted on March 22, 2024 March 22, 2024

Input your search keywords and press Enter.

Home / Guides / Citation Guides / Citation Basics / Quoting vs. Paraphrasing vs. Summarizing

Quoting vs. Paraphrasing vs. Summarizing

If you’ve ever written a research essay, you know the struggle is real. Should you use a direct quote? Should you put it in your own words? And how is summarizing different from paraphrasing—aren’t they kind of the same thing?

Knowing how you should include your source takes some finesse, and knowing when to quote directly, paraphrase, or summarize can make or break your argument. Let’s take a look at the nuances among these three ways of using an outside source in an essay.

What is quoting?

The concept of quoting is pretty straightforward. If you use quotation marks, you must use precisely the same words as the original , even if the language is vulgar or the grammar is incorrect. In fact, when scholars quote writers with bad grammar, they may correct it by using typographical notes [like this] to show readers they have made a change.

“I never like[d] peas as a child.”

Conversely, if a passage with odd or incorrect language is quoted as is, the note [sic] may be used to show that no changes were made to the original language despite any errors.

“I never like [sic] peas as a child.”

The professional world looks very seriously on quotations. You cannot change a single comma or letter without documentation when you quote a source. Not only that, but the quote must be accompanied by an attribution, commonly called a citation. A misquote or failure to cite can be considered plagiarism.

When writing an academic paper, scholars must use in-text citations in parentheses followed by a complete entry on a references page. When you quote someone using MLA format , for example, it might look like this:

“The orphan is above all a character out of place, forced to make his or her own home in the world. The novel itself grew up as a genre representing the efforts of an ordinary individual to navigate his or her way through the trials of life. The orphan is therefore an essentially novelistic character, set loose from established conventions to face a world of endless possibilities (and dangers)” (Mullan).

This quote is from www.bl.uk/romantics-and-victorians/articles/orphans-in-fiction , which discusses the portrayal of orphans in Victorian English literature. The citation as it would look on the references page (called Works Cited in MLA) is available at the end of this guide.

What is paraphrasing?

Paraphrasing means taking a quote and putting it in your own words.

You translate what another writer has said into terms both you and your reader can more easily understand. Unlike summarizing, which focuses on the big picture, paraphrasing is involved with single lines or passages. Paraphrasing means you should focus only on segments of a text.

Paraphrasing is a way for you to start processing the information from your source . When you take a quote and put it into your own words, you are already working to better understand, and better explain, the information.

The more you can change the quote without changing the original meaning , the better. How can you make significant changes to a text without changing the meaning?

Here are a few paraphrasing techniques:

Use synonyms of words
Change the order of words
Change the order of clauses in the sentences
Move sentences around in a section
Active – passive
Positive – negative
Statement-question

Let’s look at an example. Here is a direct quote from the article on orphans in Victorian literature:

“It is no accident that the most famous character in recent fiction – Harry Potter – is an orphan. The child wizard’s adventures are premised on the death of his parents and the responsibilities that he must therefore assume. If we look to classic children’s fiction we find a host of orphans” (Mullan).

Here is a possible paraphrase:

It’s not a mistake that a well-known protagonist in current fiction is an orphan: Harry Potter. His quests are due to his parents dying and tasks that he is now obligated to complete. You will see that orphans are common protagonists if you look at other classic fiction (Mullan).

What differences do you spot? There are synonyms. A few words were moved around. A few clauses were moved around. But do you see that the basic structure is very similar?

This kind of paraphrase might be flagged by a plagiarism checker. Don’t paraphrase like that.

Here is a better example:

What is the most well-known fact about beloved character, Harry Potter? That he’s an orphan – “the boy who lived”. In fact, it is only because his parents died that he was thrust into his hero’s journey. Throughout classic children’s literature, you’ll find many orphans as protagonists (Mullan).

Do you see that this paraphrase has more differences? The basic information is there, but the structure is quite different.

When you paraphrase, you are making choices: of how to restructure information, of how to organize and prioritize it. These choices reflect your voice in a way a direct quote cannot, since a direct quote is, by definition, someone else’s voice.

Which is better: Quoting or paraphrasing?

Although the purpose of both quoting and paraphrasing is to introduce the ideas of an external source, they are used for different reasons. It’s not that one is better than the other, but rather that quoting suits some purposes better, while paraphrasing is more suitable for others.

A direct quote is better when you feel the writer made the point perfectly and there is no reason to change a thing. If the writer has a strong voice and you want to preserve that, use a direct quote.

For example, no one should ever try to paraphrase John. F. Kenney’s famous line: “Ask not what your country can do for you; ask what you can do for your country.”

However, think of direct quotes like a hot pepper: go ahead and sprinkle them around to add some spice to your paper, but… you might not want to overdo it.

Conversely, paraphrasing is useful when you want to bring in a longer section of a source into your piece, but you don’t have room for the full passage . A paraphrase doesn’t simplify the passage to an extreme level, like a summary would. Rather, it condenses the section of text into something more useful for your essay. It’s also appropriate to paraphrase when there are sentences within a passage that you want to leave out.

If you were to paraphrase the section of the article about Victorian orphans mentioned earlier, you might write something like this:

Considering the development of the novel, which portrayed everyday people making their way through life, using an orphan as a protagonist was effective. Orphans are characters that, by definition, need to find their way alone. The author can let the protagonist venture out into the world where the anything, good or bad, might happen (Mullan).

You’ll notice a couple of things here. One, there are no quotation marks, but there is still an in-text citation (the name in parentheses). A paraphrase lacks quotation marks because you aren’t directly quoting, but it still needs a citation because you are using a specific segment of the text. It is still someone else’s original idea and must be cited.

Secondly, if you look at the original quote, you’ll see that five lines of text are condensed into four and a half lines. Everything the author used has been changed.

A single paragraph of text has been explained in different words—which is the heart of paraphrasing.

What is summarizing?

Next, we come to summarizing. Summarizing is on a much larger scale than quoting or paraphrasing. While similar to paraphrasing in that you use your own words, a summary’s primary focus is on translating the main idea of an entire document or long section.

Summaries are useful because they allow you to mention entire chapters or articles—or longer works—in only a few sentences. However, summaries can be longer and more in-depth. They can actually include quotes and paraphrases. Keep in mind, though, that since a summary condenses information, look for the main points. Don’t include a lot of details in a summary.

In literary analysis essays, it is useful to include one body paragraph that summarizes the work you’re writing about. It might be helpful to quote or paraphrase specific lines that contribute to the main themes of such a work. Here is an example summarizing the article on orphans in Victorian literature:

In John Mullan’s article “Orphans in Fiction” on bl.uk.com, he reviews the use of orphans as protagonists in 19 th century Victorian literature. Mullan argues that orphans, without family attachments, are effective characters that can be “unleashed to discover the world.” This discovery process often leads orphans to expose dangerous aspects of society, while maintaining their innocence. As an example, Mullan examines how many female orphans wind up as governesses, demonstrating the usefulness of a main character that is obligated to find their own way.

This summary includes the main ideas of the article, one paraphrase, and one direct quote. A ten-paragraph article is summarized into one single paragraph.

As for giving source credit, since the author’s name and title of the source are stated at the beginning of the summary paragraph, you don’t need an in-text citation.

How do I know which one to use?

The fact is that writers use these three reference types (quoting, paraphrasing, summarizing) interchangeably. The key is to pay attention to your argument development. At some points, you will want concrete, firm evidence. Quotes are perfect for this.

At other times, you will want general support for an argument, but the text that includes such support is long-winded. A paraphrase is appropriate in this case.

Finally, sometimes you may need to mention an entire book or article because it is so full of evidence to support your points. In these cases, it is wise to take a few sentences or even a full paragraph to summarize the source.

No matter which type you use, you always need to cite your source on a References or Works Cited page at the end of the document. The MLA works cited entry for the text we’ve been using today looks like this:

Mullan, John. Orphans in Fiction” www.bl.uk/romantics-and-victorians/articles/orphans-in-fiction. Accessed 20. Oct. 2020

————–

See our related lesson with video: How to Quote and Paraphrase Evidence

Citation Guides

Annotated Bibliography
Block Quotes
Citation Examples
et al Usage
In-text Citations
Page Numbers
Reference Page
Sample Paper
APA 7 Updates
View APA Guide
Bibliography
Works Cited
MLA 8 Updates
View MLA Guide

How useful was this post?

Click on a star to rate it!

We are sorry that this post was not useful for you!

Let us improve this post!

Tell us how we can improve this post?

Citation Basics

Harvard Referencing

Plagiarism Basics

Plagiarism Checker

Upload a paper to check for plagiarism against billions of sources and get advanced writing suggestions for clarity and style.

Get Started

Doug Lemov's field notes

Teaching & Schools
Coaching & Practice

03.09.16 On the Difference Between Paraphrase & Summary

Paraphrase, Colleen, Erica and I point out in Reading Reconsidered , is not the same as summary. A paraphrase is a restatement of the sentence in simpler, clarified terms that still capture all of the explicit meaning and as much of the connotation as possible.

A paraphrase recreates a facsimile of the passage from the narrator’s point of view. A summary describes the passage from the reader’s point of view.

In the book, we use the example of this short passage from To Kill a Mockingbird in which Scout is talking about her Aunt Alexandra who speaks a coded language about the importance of heredity:

I have never understood her preoccupation with heredity. Somewhere, I had received the impression that Fine Folks were people who did the best with the sense they had, but Aunt Alexandra was of the opinion, obliquely expressed, that the longer a family had been squatting on one patch of land the finer it was.

We observed that writing something like, “Scout is reflecting on her interpretation of how class was determined, as contrasted to that of her Aunt Alexandra” is not a paraphrase but rather a summary. A paraphrase would not describe the passage but replicate it- it would be written in the first person and take on Scout’s point of view. It might start: “Somehow I had come to believe that respectable people were . . .”

In addition to that a good paraphrase would:

Restate all parts of the excerpt in new words—the hard parts as well as the easy parts; the key ideas and the sub-textual ones
Mimic the excerpt’s Point of View, perspective and tense (First person versus third person, say; past versus present)
Match the excerpt’s tone the degree possible

So it might sound something like this:

“I had come to believe, though I’m not sure when or how, that people of stature were people who lived as wisely and well as they could given their circumstances but Aunt Alexandra believed, though she wouldn’t come out and say it directly, that status was based on how long you had been living on your land.”

Paraphrasing might seem to some like the most straightforward and mundane of activities—banal, even, but if you try to do all of the things we describe to the passage from To Kill a Mockingbird you would likely find that paraphrasing a worthy segment of complex text is a rigorous task. What does it mean that Aunt Alexandra’s opinions about race and class were “obliquely expressed”? What does “Fine Folks” mean and why is it capitalized?

Paraphrasing is a worthy and challenging passage is a worth task in teaching students to deal with challenging text. In light of that our colleague, Maggie Johnson, put together this training activity.

Close Reading_Paraphrase Activity_2.9

It’s designed to allow teachers to reflect on the difference between paraphrase and summary and to prepare to use them (differently) in their classrooms. I thought I’d share it here in case you found it useful.

Name (required)

Mail (will not be published) (required)

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

In order to pass the CAPTCHA please enable JavaScript

SEARCH OUR SITE

louisville.edu
PeopleSoft HR
PeopleSoft Campus Solutions
PeopleSoft Financials
Business Ops
Cardinal Careers

Undergraduate
International
Online Learning

University of Louisville Writing Center

University Writing Center FAQs
Virtual Writing Center FAQs
HSC Writing Center FAQs
Writing FAQs
Handouts and Videos
Graduate Student Writing
Spring Dissertation Writing Retreat
Graduate Student Writing Workshops
Faculty and Graduate Student Writing Group
Creative Writing Group
Accessibility and Accommodations
LGBTQ+ Writing Group
The University Writing Center and Your Students
Request a Presentation about the University Writing Center
Resources for Teaching Writing
The Writing Center and Your Writing
University Writing Center Mission Statement
Meet Our Staff
Statement on Diversity, Inclusion, and Equity
Research at the University Writing Center
How I Write Blog Posts
Our Community Writing Values and Approaches
Community Writing Internships and Volunteering
Family Scholar House
Western Branch Library
How can I make myself a stronger writer?
What makes college writing different than the writing I’ve done up to this point?
How are the papers I'm asked to write in my major different from those in English 101, 102, and 105 courses?
What can I do if I don’t completely understand the writing assignment?
I want to get started writing early, but how do I begin?
How do I get started writing a personal statement?
I have a lot to say, but how can I organize my thoughts?
How can I learn how to write in a new genre (for example, personal statement, resume, or literature review)?
How do I expand a rough draft to make it meet the assignment’s length requirement?
How can I find good sources for my research paper?
What are some strategies for working sources into my research paper?
What is the difference between quotation, paraphrase, and summary?
How can I revise my draft if it doesn’t seem to “flow”?
What does my teacher mean by “substantial revision?”
How do I write an essay that makes an “argument”?
How can I avoid plagiarizing?
What are some strategies for improving my grammar and punctuation?
How can I format my document properly in Word, PowerPoint or Excel?
How should I approach writing a literature review at the graduate level?
/ Resources for Students
/ Writing FAQs
/ What is the difference between quotation, paraphrase, and summary?

Writing in college often means using ideas from other sources. There are times when it may be best to quote the sources directly, while other times may be better served by paraphrasing or summary. In order to decide which technique to use, it is helpful to think about how you are using the information in your paper.

Definitions

Quotation reproduces a statement word-for-word as it appears in its original source
Paraphrase explains a statement by using your own words and sentence structure
Summary explains a statement using your words, but typically condenses a larger statement into a shorter explanation

How to decide which approach to use

Direct quotations can be useful when the exact wording of a statement is important. The exact wording of a quotation may be significant to your claim. In example 1 below, the contrast between adjectives are important to the claim. Also, direct quotation may be important when you want to make sure you are being precise in representing the author’s position. Finally, you might choose to use a direct quotation when the original statement is particularly well written or structurally persuasive. If a statement uses elements such as parallelism or alliteration, you might not be able to recreate that same effect. An important element of the quotation in example 1 is the parallel structure between "lowest and vilest alleys" and "smiling and beautiful countryside."

When Sherlock tells Watson "the lowest and vilest alleys in London do not present a more dreadful record of sin than does the smiling and beautiful countryside,” he intensifies suspense by equating innocence with evil ("The Adventure of the Copper Beeches" 502).

Paraphrasing is usually expected in research and argumentative essays. These type of papers benefit from paraphrasing because it shows that you understand the source and are therefore a reliable voice on that source. Paraphrasing can make the evidence more straightforward. Another reason to paraphrase is to adjust your tone for your audience. If the assignment asks you to write a presentation for your classmates, you do not want to quote scientific jargon. Your source is only persuasive and supportive if your readers understand it. The paraphrase of the quotation below is shorter, and more direct.

Original quotation: “In the case of Facebook, it has changed its format multiple times, and merged other literacy practices – email, instant messaging, games – into its structure in an attempt to keep users on the site” (Keller 2014, 74).

Paraphrase: Facebook has tried to hold on to its users by incorporating new functions like games and email (Keller 2014).

Summaries can also be used in reviews, research papers, and argumentative essays. They have a similar purpose as paraphrasing, but they condense a large work (i.e. an entire chapter, article, or book) into a shorter text such as a paragraph or a short essay. Summaries allow you to focus your description on the parts that are relevant to your discussion. Example 3 briefly summarizes Anne of Green Gables, focusing on Anne as a strong female character and could lead into a discussion of how the series teaches girls self-respect while also cherishing romance.

Anne of Green Gables is a book series that follows the life of an unruly red-headed orphan as she grows from an romantic adolescent into an independent young woman.

What can the Writing Center do to help?

Writing Center consultants can help you if you aren't sure what style of source integration works best for an assignment. Some essays require a mix of methods. Consultants can help you determine if your writing needs a better balance of integration methods. If you are less familiar with one of the three uses of sources, the Writing Center can give you additional pointers.

See our section on how to incorporate sources for more on punctuating and introducing quotations. Also see our section on avoiding plagiarism to learn how to paraphrase and summarize.

Conceptualizing Trauma-Informed Consulting in the University Writing Center Apr 17, 2024

Creating Art: A Painter’s Journey Into the World of Writing Mar 25, 2024

Getting Comfortable with Directive Practices in the Writing Center Mar 08, 2024

International Mother Language Day 2024 Mar 04, 2024

University and High School Writing Centers Feb 26, 2024

How Time Affects a Writing Center Session Feb 16, 2024

UofL Writing Center Blog - More…

University Writing Center

Ekstrom Library 132

Kornhauser Library 221

University of Louisville

Louisville, Kentucky 40292

Summer 2024

Ekstrom Library

M -F 9 am - 4 pm

Closed on student breaks and holidays

(502) 852-2173

[email protected]

Social Media

Direct Quotes vs. Paraphrasing vs. Summarizing: Know the Difference

Written By Lorraine Roberte
Updated: February 22, 2024

Knowing the differences between direct quotes, paraphrasing, and summarizing is crucial no matter your occupation, from business owner to content marketer.

Why? Because it can prevent you from accidentally plagiarizing in the work you do for yourself and from breeching best practices.

Incorporating a mixture of these elements in your content can also help you tell a better story, so your audience keeps reading.

Direct quotes vs. paraphrasing vs. summarizing — understanding the difference

We’re breaking down the differences between direct quotes vs. paraphrasing and summarizing and how you can use them in your writing.

From press releases for your business to engaging blog posts for your target audience, you can make your writing more interesting by including trustworthy sources.

Direct quotes

Direct quotes include the exact words that someone said, with quotation marks and name attribution. They’re especially common when writing about people .

Example: “Elon Musk said in a tweet that Starlink’s satellite broadband service coverage will be available on ‘most of Earth by end of year,’ although he noted that ‘cellular will always have the advantage in dense urban areas.'”

When to use direct quotes

According to the APA style guide , you’ll need to use direct quotes when:

Copying an exact definition
The author’s words are memorable and succinct
Responding or reacting to someone’s exact words

How to use direct quotes

In general, direct quotes are written verbatim. But you can make these small changes without alerting your readers:

Changing the first letter of the quote to an upper or lowercase so that the quotation matches the context sentence’s syntax. Can also modify the punctuation at the end of the quote.
Swapping single quotation marks to double quotation marks and vice versa
Omitting footnote or endnote number references

Paraphrasing

Paraphrasing is when you restate someone else’s words, but not word for word.

Example (original quote): “It’s risky trusting employees as much as we do. Giving them as much freedom as we do. But it’s essential in creative companies where you have much greater risk from lack of innovation.” — Reed Hastings, Netflix CEO and co-founder .

Example (paraphrase): “Netflix’s CEO and co-founder, Reed Hastings, feels that micromanaging workers can stifle innovation in creative businesses.”

When to paraphrase

It can be helpful to paraphrase if you want to keep your writing more conversational. It’s also useful when breaking up direct quotes or explaining the original source in simpler terms. That way, the information better fits the tone and style of your writing.

How to paraphrase

Paraphrasing involves putting a section of the source information entirely into your own words while staying true to its original meaning. You can link to the source in the place that makes the most sense, such as “report” for an industry report.

You can keep from plagiarizing when paraphrasing by using synonyms for words mentioned in the source. It’s important to restate phrases differently (even if they’re just a few words) to avoid the same sentence structure. If you don’t, you could still be plagiarizing, despite crediting the source.

If you use exact words from the original material while paraphrasing, you must put the word or words in quotes. The exception is generic terms that are difficult to find synonyms for.

Summarizing

When you summarize, you use your own words to describe the critical points of what someone else said or that you heard or read in a source.

Example (original quote): “In a diverse population of older patients who were hospitalized for acute decompensated heart failure, an early, transitional, tailored, progressive rehabilitation intervention that included multiple physical-function domains resulted in greater improvement in physical function than usual care.” — Study in the New England Journal of Medicine

Example (summary): “A recent study shows physical rehabilitation programs to be helpful for older populations with hospitalizations from heart failure.”

When to summarize

Summaries are excellent at giving readers the key insights they need from a longer text when proving your point. They also add context while keeping at a manageable length whatever type of article you’re writing.

How to summarize

You don’t need to include any quotes or attribution when summarizing, just a brief overview that often links back to the original material for more details. It may also introduce essential points from the original text, allowing readers to understand the source without clicking through it.

Now that you know the difference between direct quotes, paraphrasing, and summarizing, you can confidently write content for your business.

Need help creating engaging blog posts for your business? Talk to a content specialist at ClearVoice today about your needs.

Stay in the know.

We will keep you up-to-date with all the content marketing news and resources. You will be a content expert in no time. Sign up for our free newsletter.

Elevate Your Content Game

Transform your marketing with a consistent stream of high-quality content for your brand.

Why ClearVoice: Your Flexible Content Partner

ClearVoice Content Roadmap: The Pain Points

ClearVoice Content Roadmap: The Product

Content Production
Build Your SEO
Amplify Your Content
For Agencies

Why ClearVoice

Talent Network
How It Works
Freelance For Us
Statement on AI
Talk to a Specialist

Get Insights In Your Inbox

Privacy Policy
Terms of Service
Intellectual Property Claims
Data Collection Preferences

school Campus Bookshelves
menu_book Bookshelves
perm_media Learning Objects
login Login
how_to_reg Request Instructor Account
hub Instructor Commons

Margin Size

Download Page (PDF)
Download Full Book (PDF)
Periodic Table
Physics Constants
Scientific Calculator
Reference & Cite
Tools expand_more
Readability

selected template will load here

This action is not available.

7.1: Using and Organizing Information Details

Last updated
Save as PDF
Page ID 100032

$ \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } $

$ \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} $

$ \newcommand{\id}{\mathrm{id}}$ $ \newcommand{\Span}{\mathrm{span}}$

( \newcommand{\kernel}{\mathrm{null}\,}\) $ \newcommand{\range}{\mathrm{range}\,}$

$ \newcommand{\RealPart}{\mathrm{Re}}$ $ \newcommand{\ImaginaryPart}{\mathrm{Im}}$

$ \newcommand{\Argument}{\mathrm{Arg}}$ $ \newcommand{\norm}[1]{\| #1 \|}$

$ \newcommand{\inner}[2]{\langle #1, #2 \rangle}$

$ \newcommand{\Span}{\mathrm{span}}$

$ \newcommand{\id}{\mathrm{id}}$

$ \newcommand{\kernel}{\mathrm{null}\,}$

$ \newcommand{\range}{\mathrm{range}\,}$

$ \newcommand{\RealPart}{\mathrm{Re}}$

$ \newcommand{\ImaginaryPart}{\mathrm{Im}}$

$ \newcommand{\Argument}{\mathrm{Arg}}$

$ \newcommand{\norm}[1]{\| #1 \|}$

$ \newcommand{\Span}{\mathrm{span}}$ $ \newcommand{\AA}{\unicode[.8,0]{x212B}}$

$ \newcommand{\vectorA}[1]{\vec{#1}} % arrow$

$ \newcommand{\vectorAt}[1]{\vec{\text{#1}}} % arrow$

$ \newcommand{\vectorB}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } $

$ \newcommand{\vectorC}[1]{\textbf{#1}} $

$ \newcommand{\vectorD}[1]{\overrightarrow{#1}} $

$ \newcommand{\vectorDt}[1]{\overrightarrow{\text{#1}}} $

$ \newcommand{\vectE}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash{\mathbf {#1}}}} $

The Scholarly Conversation

As a student you’re participating in a scholarly conversation anytime you write a paper, complete a research project, or make a speech describing, arguing, or synthesizing information on a topic you’ve studied. In this conversation, you are listening to lots of different voices such as your professors, the authors of your textbooks, journalists reporting on an issue, and researchers who have performed studies or analyzed that issue.

Through your academic assignments, you’re adding your voice to the discussion. Depending on the purpose of the assignment you may be:

synthesizing and drawing connections between what others have said on an issue
explaining your personal experiences related to the topic, and/or
expressing your own opinions or conclusions about what the research shows.

For this scholarly conversation to be successful, you have to approach your academic assignments with honesty,use information sources effectively , and give proper credit to the other voices in the conversation through attribution and citation.

These terms - attribution and citation - are going to come up a lot over the next two weeks. Before moving on, here are some quick definitions of the terms:

Attribution (noun): Crediting ideas, words, or works to a particular author, artist, or person.
Cite (verb): To give information from a source as support or evidence in a research paper or project.
Citation (noun): A note describing the original source of ideas, words, or works included in a research paper or project.

Manage Pillar

Being a part of the scholarly conversations means you know how to effectively use and organize information. This is where the Manage pillar comes in. It focuses on the need to use and organize information professionally and ethically. This means:

Individuals understand:

Their responsibility to be honest in all aspects of information handling and dissemination (e.g. copyright, plagiarism, and intellectual property issues).
The need to adopt appropriate data-handling methods.
The role they play in helping others in information seeking and management.
The need to keep systematic records.
The importance of storing and sharing information and data ethically.
The role of professionals, such as data managers and librarians, who can advise, assist, and support with all aspects of information management.

They are able to

Use reference management software, if appropriate, to manage information.
Cite printed and electronic sources using suitable referencing styles.
Create appropriately formatted bibliographies.
Demonstrate awareness of issues relating to the rights of others including ethics, data protection, copyright, plagiarism, and any other intellectual property issues.
Meet standards of conduct for academic integrity.
Use appropriate data management software and techniques to manage data.

In order to manage information and take part in a scholarly conversation, we often rely on the works of others. We are using information to help build our knowledge, form educated opinions, and create our own contributions.

So how should we handle this product of creativity (a.k.a information)? Let’s think about a simple example: apple picking in the fall. It is a popular thing to do, especially in our Sierra foothills. Have you ever been to Apple Hill? People come to orchards, get bags or baskets, gather apples, and then line up to weigh them and pay. The farmers’ hard work is being rewarded.

Now imagine a different situation. You worked hard to research and write a very good article that is being published in the college newsletter. You've discovered that your roommate has copied a couple of paragraphs and inserted them into one of her assignments because the topics were related. Was this fair? How were you rewarded for your hard work? Bottom line - she should not have used your intellectual capital without attribution to you. What she did was an act of plagiarism.

Once your ideas become tangible - meaning they're written down on paper, on a computer, on your phone, they become your intellectual personal property and you hold the copyright. This means that no one has the right to reproduce all or any part of it (i.e. copy it) without your permission. If your roommate decides to use some information from your article in her own paper, she should provide a citation. If she is using direct quotes from your article, again, she would need to put double quotes around your words and provide information about the author (you, in this instance) to avoid plagiarism.

There is a lot to learn about using information legally and ethically, copyright and plagiarism are just two aspects of intellectual property. This knowledge will empower you in your academic work and help you be successful.

Plagiarism Defined

Plagiarism is using someone else’s words or ideas in your academic work without giving credit to the original author. In college, the way we most often give credit is in the form of a citation. Plagiarism can include:

Summarizing the ideas, statements of fact, or conclusions made by another without citing the source.
Word-for-word copying (CTRL + V) of work written by someone else without quotation marks and/or a citation.
Close or extended paraphrasing of another person’s work without acknowledging the source. In other words, you’ve taken someone else’s ideas and changed a few words here and there, but haven’t acknowledged where you got the idea from.
Using an image, video, or music without permission and/or proper attribution.

In the classroom (and in the world of publishing), documenting your information sources is the only way others can tell how thorough and careful you’ve been in researching your topic. If you don’t tell readers where your information came from, they may think (and many do) that you either made up the information or “stole” it . Failing to cite your sources is plagiarism.

Unintentional Plagiarism

Plagiarism can be intentional, meaning you knowingly copied the words of another person in your paper. Or it can be unintentional, meaning you weren’t trying to pass off someone else’s ideas as your own, but you didn’t properly cite your sources. Unintentional plagiarism often happens when students don’t know citation rules, fail to keep track of their research, or are rushing and/or disorganized when completing an assignment.

Jackie was working on her 10-page research paper at the last minute. It was 3:30 am and her paper was due in class at 9:00 am. She finished the last sentence at 5:15 am, did a spellcheck and voila! Done! Groggy yet awake she went to class, turned in the paper and waited for her grade. She received an email from her professor that read, “There are some major issues with your research paper that I need to discuss with you. Please see me.” Uh oh.

When she nervously went to see him, the professor told Jackie that she hadn’t cited any of her sources, and because she included a lot of direct quotes in her paper, she was guilty of plagiarism. She received an F on her paper and may be referred to the school administration for academic dishonesty.

Academic Honesty

In your college courses, it’s expected that you (and only you) will do the assigned work. By completing the assignments, papers, quizzes, and exams, you are showing that you have mastered the content of the course - meaning you have learned the information and gained the skills covered in that class. (Your grade shows how well you’ve done that.)

When you are dishonest in your academic work, either by cheating or plagiarizing another person’s work, the learning process breaks down.

Plagiarism Detection

You should be aware that our Learning Management System (Canvas) includes a plagiarism detection tool called Vericite. If your instructor is using this tool, your papers and assignments could be compared to a large database of material to search for signs of matching text. If you have not quoted, paraphrased, or cited sources properly, you could be accused of academic dishonesty.

Consequences of Academic Dishonesty

There are big consequences for academic dishonesty. If you plagiarize or cheat at a Los Rios college, you could face one or more of these consequences:

Receiving a failing grade on the assignment
Receiving a lower overall grade in the class
Failing the class altogether
Being referred to the College’s Disciplinary Officer and/or
Being placed on disciplinary probation, suspension or even expulsion from the College.

Even if you are not caught cheating or plagiarizing, you are still damaging your opportunity to learn from the coursework.

Real World Scenario

Students often feel that they are being singled out in regard to plagiarism and academic dishonesty. But that is far from the case. There are numerous examples of scholars and other professionals who have been caught plagiarizing.

Take a look at these headlines...

Another example, with a dramatic outcome, is that of Eugene Tobin. He was the president of Hamilton College in New York State, when it was discovered that he had included plagiarized material in speeches he had given over the course of almost a decade. He resigned from his position as the head of this prestigious institution, admitting his guilt (Isserman, 2003). Other college presidents and administrators have also been caught violating academic trust. If you try a search using the terms plagiarism and college president , you may be dismayed at the number of results.

If you plagiarize in your career, you could face job penalties like firing or suspension, your reputation could be damaged, and you could even face lawsuits.

Clearly, plagiarism and cheating are things you want to avoid.

Plagiarism can be avoided when you learn to give proper credit through attribution and citation and when you learn to use information sources effectively by staying organized throughout the research and writing process. More on that is coming up next.

Why We Cite

One of the best ways to avoid plagiarism and to follow the rules of academic writing is to cite your sources. Students often think about plagiarism and wanting to avoid it when they talk about the reasons to cite a source. It is more than that though. This very short video from North Caroline State University Library explains exactly what a citation is and why we use them.

Citation a very brief introduction tile that links to video.

In your research projects, it’s important to seek out a variety of sources like books, articles, websites, films, etc. These sources will help you define and describe your topic, identify new developments, current events, or emerging research, and see what conclusions or opinions other people have come up with related to your topic.

When you write your paper and present the information you found in your research, you need to say where this information came from. This is called attribution or acknowledging your sources, and it’s usually done through a citation. Again, a citation is a brief note that describes the original source of the ideas.

Citing your sources serves a few purposes:

It shows your preparation for your research assignment and illustrates the conversation surrounding your topic.
When you use credible sources , it helps strengthen your argument.
It allows your reader to learn more by referring them back to the original source.
It helps you avoid plagiarism by giving credit to the thinkers and authors whose works you are drawing from.

A citation is required anytime you:

Summarize someone’s ideas on a topic.
Use a direct quote or copy and paste from a source.
Paraphrase information you read in a source - meaning you’ve taken someone else’s idea and put it into your own words.
State a fact that is not considered common knowledge.

Summarizing

Summarizing is when you describe or explain the central ideas or most important information found in a source. You might read a whole 5-page article about an issue, but in your research paper you just describe the main points of that article in one sentence. Summarizing is taking a lot of information and explaining it in as few words as possible. But because you are explaining what you learned from a source , you need to include a citation at the end of the summarizing sentence.

How to Summarize: An Overview

A summary is a brief explanation of a longer text. Some summaries, such as the ones that accompany annotated bibliographies, are very short, just a sentence or two. Others are much longer, though summaries are always much shorter than the text being summarized in the first place.

Summaries of different lengths are useful in research writing because you often need to provide your readers with an explanation of the text you are discussing. This is especially true when you are going to quote or paraphrase from a source.

Of course, the first step in writing a good summary is to do a thorough reading of the text you are going to summarize in the first place. Beyond that important start, there are a few basic guidelines you should follow when you write summary material:

Stay “neutral” in your summarizing. Summaries provide “just the facts” and are not the place where you offer your opinions about the text you are summarizing. Save your opinions and evaluation of the evidence you are summarizing for other parts of your writing.
Don’t put quotes inside your summary. Summaries will be more useful to you and your colleagues if you write them in your own words.
Don’t “cut and paste” from database abstracts. Many of the periodical indexes that are available as part of the library’s database system include abstracts of articles. Do no “cut” this abstract material and then “paste” it into your own annotated bibliography. For one thing, this is plagiarism. Second, “cutting and pasting” from the abstract defeats one of the purposes of writing summaries and creating an annotated bibliography in the first place, which is to help you understand and explain your research.

Machine readable text description of preceding graphic

Summarizing information graphic description.

The information graphic shows a definition of summarizing:

Summarizing is taking the main themes or ideas from your source and explaining or describing them in just a few sentences. An entire chapter of a book or a 5-page article might be summarized in just one or two sentences.

The book, Geek Girl Rising , by Heather Cabot and Samantha Walravens is the featured example. In a paper, a summary in APA style would be written as follows:

In their book, Geek Girl Rising , Heather Cabot and Samantha Walravens, highlight success stories of women working in technology and provide inspiration for girls wanting to break into this male-dominated industry (2017).

Writer's quote (and paraphrase) from research in order to support their points and to persuade their readers. A quote from a piece of evidence in support of a point answers the reader's question, says who?"

This is especially true in academic writing since scholarly readers are most persuaded by effective research and evidence. For example, readers of an article about a new cancer medication published in a medical journal will be most interested in the scholar’s research and statistics that demonstrate the effectiveness of the treatment. Conversely, they will not be as persuaded by emotional stories from individual patients about how a new cancer medication improved the quality of their lives. While this appeal to emotion can be effective and is common in popular sources, these individual anecdotes do not carry the same sort of “scholarly” or scientific value as well-reasoned research and evidence.

Of course, your instructor is not expecting you to be an expert on the topic of your research paper. While you might conduct some primary research, it’s a good bet that you’ll be relying on secondary sources such as books, articles, and Web sites to inform and persuade your readers. You’ll present this research to your readers in the form of quotes and paraphrases.

A direct quote includes the exact word-for-word sentences or phrases that you found in a source. When you copy and paste text into your paper, you are directly quoting that source. A direct quote must have quotation marks around it and it must include a citation to show the reader where those words came from.

While quotes and paraphrases are different and should be used in different ways in your research writing (as the examples in this section suggest), they do have a number of things in common. Both quotes and paraphrases should:

be “introduced” to the reader, particularly the first time you mention a source;
include an explanation of the evidence which explains to the reader why you think the evidence is important, especially if it is not apparent from the context of the quote or paraphrase; and
include a proper citation of the source.

When to Quote

the question of when to quote and when to paraphrase depends a great deal on the specific context of the writing and the effect you are trying to achieve. Learning the best times to quote and paraphrase takes practice and experience.

In general, it is best to use a quote when:

The exact words of your source are important for the point you are trying to make. This is especially true if you are quoting technical language, terms, or very specific word choices.
You want to highlight your agreement with the author’s words. If you agree with the point the author of the evidence makes and you like their exact words, use them as a quote.
You want to highlight your disagreement with the author’s words. In other words, you may sometimes want to use a direct quote to indicate exactly what it is you disagree about. This might be particularly true when you are considering the antithetical positions in your research writing projects.

Tips for Quoting

Introduce your quotes to your reader, especially on first reference.
Explain the significance of the quote to your reader.
Cite your quote properly according to the rules of style you are following in your essay.
Quote when the exact words are important, when you want to highlight your agreement or your disagreement.

Quoting in APA Style - An Example

Consider this BAD example in APA style, of what NOT to do when quoting evidence:

“If the U.S. scallop fishery were a business, its management would surely be fired, because its revenues could readily be increased by at least 50 percent while its costs were being reduced by an equal percentage.” (Repetto, 2001, p. 84).

Again, this is a potentially valuable piece of evidence, but it simply isn’t clear what point the research writer is trying to make with it. Further, it doesn’t follow the preferred method of citation with APA style.

Here is a revision that is a GOOD or at least BETTER example:

Repetto (2001) concludes that in the case of the scallop industry, those running the industry should be held responsible for not considering methods that would curtail the problems of over-fishing. “If the U.S. scallop fishery were a business, its management would surely be fired, because its revenues could readily be increased by at least 50 percent while its costs were being reduced by an equal percentage” (p. 84).

Paraphrasing

Paraphrasing is when you use your own words to explain something that you learned from a source. Paraphrasing can be a useful way to clearly explain the meaning of information you uncovered through your research. It’s a good way to describe what you’ve learned and also help the reader understand the significance of the information. Paraphrasing is a common writing technique, but it’s also where many students unintentionally plagiarize.

Students unintentionally plagiarize because they:

Take a sentence from a source and rearrange the words.
Use the thesaurus tool to change a few words in a sentence.

To do a good job of paraphrasing, you have to make sure you are using only your words, not the author’s. First, read the original source and think about it. Make a few notes of what you think it means. After that, try explaining the author’s ideas in your own words. It’s helpful if you wait a little while between reading the source and trying to paraphrase it so that the author’s words aren’t quite so fresh in your mind. If there’s a word or phrase that you keep repeating when you try to paraphrase, then you probably should just quote it (with quotation marks and a citation).

When to Paraphrase

In general, it is best to paraphrase when:

There is no good reason to use a quote to refer to your evidence. If the author’s exact words are not especially important to the point you are trying to make, you are usually better off paraphrasing the evidence.
You are trying to explain a particular piece of evidence in order to explain or interpret it in more detail. This might be particularly true in writing projects like critiques.
You need to balance a direct quote in your writing. You need to be careful about directly quoting your research too much because it can sometimes make for awkward and difficult to read prose. So, one of the reasons to use a paraphrase instead of a quote is to create balance within your writing.

Tips for Paraphrasing

Introduce paraphrases to your reader, especially on first reference.
Explain the significance of the paraphrased material to your reader.
Cite your paraphrased ideas properly according to the rules of style you are following in your essay.
Paraphrase when the exact words aren’t important, when you want to explain the point of your evidence, or when you need to balance the direct quotes in your writing.

Paraphrasing in APA Style - An Example

Paraphrasing in APA style is slightly different from MLA style as well. Consider first this BAD example of what NOT to do in paraphrasing from a source in APA style:

Computer criminals have lots of ways to get away with credit card fraud (Cameron, 2002).

The main problem with this paraphrase is there isn’t enough here to adequately explain to the reader what the point of the evidence really is. Remember: your readers have no way of automatically knowing why you as a research writer think that a particular piece of evidence is useful in supporting your point. This is why it is key that you introduce and explain your evidence.

Here is a revision that is GOOD or at least BETTER:

Cameron (2002) points out that computer criminals intent on committing credit card fraud are able to take advantage of the fact that there aren’t enough officials working to enforce computer crimes. Criminals are also able to use the technology to their advantage by communicating via email and chat rooms with other criminals.

Again, this revision is better because the additional information introduces and explains the point of the evidence. In this particular example, the author’s name is also incorporated into the explanation of the evidence as well. In APA, it is preferable to weave in the author’s name into your essay, usually at the beginning of a sentence. However, it would also have been acceptable to end an improved paraphrase with just the author’s last name and the date of publication in parentheses.

Common Knowledge

Common Knowledge is a term for facts that are generally well-known, not controversial, and easy to look up. When you state something that is common knowledge in your paper, you don’t have to include a citation because you are assuming that the reader already knows this information. Common knowledge can vary somewhat depending on who the audience for your research project is. If you aren’t sure whether something you are stating in your paper is common knowledge, always play it safe and include a citation.

Machine readable text description of preceding graphic.

How to Cite

Now that we know when we have to cite our sources, let’s talk about how we cite our sources.

Citation styles are sets of rules to follow for citing sources and formatting research papers.

Professionals in different disciplines or subject areas have put together these styles to help writers in their field write well, convey important information, and organize their research.

You don’t need to memorize all the rules of each citation style, you really just need to know where to go to get guidance and help. We'll be focusing more on this next week.

Regardless of which style you are using for your research assignment, there are generally two parts to a citation in academic writing. The first part tells the reader when you have included information from a source in your paper. This could be done through an in-text/parenthetical citation (MLA, APA) or through a footnote/endnote (Chicago, etc.).

In this first part of the citation, you are signalling to the reader that you’ve just provided information or ideas that you gained from someone else. You show this by including information about who’s responsible for the information. This could include things like the author’s last name, title, date of publication and/or page number.

The second part of a citation is generally a full description of the source organized in a list at the end of your paper. This list may be called a Works Cited List (MLA style), a References list (APA style) or a Bibliography (Chicago style). Include a citation for each source that you’ve referenced in your paper. This full citation will allow the reader to find the source if they want to read it.

We'll be using APA Style in this class. It is used by many disciplines, including Early Childhood Education, Psychology, Sociology, Communication Studies, and Library Science.

Ethical Issues and Intellectual Property

Knowing how to properly cite sources shows that you understand the practice of professional and ethical use of information. Ethical treatment of information assumes that you are treating an author’s rights appropriately and avoiding an act of academic dishonesty such as plagiarism.

As a creator of information yourself, you should understand the importance of respecting other authors’ rights and following the general rules of academic integrity, including being cognizant of copyright, and other issues associated with intellectual property.

Anatomy of An Academic Journal Article

In order to use and cite information effectively, you must understand what it is you're reading. Before we get into the assignments this week, I want to take a minute to explore more deeply the content of academic journal articles.

We've talked about these sources in earlier weeks of the course. These are the sources your instructors will often call "scholarly articles" or "scholarly sources." College instructors love it when you use academic journal articles because many of them see this as the most "valuable" type of academic information. We'll be exploring the value of information in coming weeks, but what I mean by that gets to the heart of most instructors' passions.

They are academics and scholars. They have chosen a field that allows them to teach other people about their discipline expertise every day. Because of this, they think very highly of academic writing and research and it will benefit you greatly as a student if you understand how to read and use these types of sources. I am going to break down one of the most common types of academic journal article, which is to present empirical research - or research that the authors have designed and collected and then presented in the form of an academic paper.

Common Sections of an Empirical Research Academic Journal Article

All journal articles are going to have a title. Sometimes titles are quite descriptive and you can immediately understand what the article is about. Other times, you might need to read the abstract to get a true picture of what's included in the article.

An abstract is the very first thing the reader will see under the article's title. It is a one-paragraph summary and can be very useful for researchers. You should know immediately after reading the abstract whether the article is relevant to your research or if it's something you should pass up. In the databases, you can click on the little magnifying glass to the right of an article's title to read the abstract without having to open up the entire article. As you're searching for sources, don't spend a lot of time reading beyond the abstract - you'll do that later as you begin organizing your thoughts and how you will use each source in your research assignment.

Introduction

This part of the paper will introduce the reader to the author(s)' research question. Just like in science class, researchers will pose a question, decide how they're going to go about answering that question (usually a survey or treatment of subjects), administer their study, and then analyze the results. The introduction will lay all of this out for you so you know exactly what information the researchers were trying to find.

Literature Review

The literature review may be its own section or might be part of the introduction. This is the section in which the authors take time to summarize other research that has been done that relates to their research question. This section can be very powerful to you as a researcher because you will often find other sources that are mentioned that could also be powerful evidence for your topic. Sometimes, researchers will write an entire paper that is a review of the literature. The embedded research guide below includes an example of a single paper that is a literature review.

This is the section in which the author(s) describe their research subjects and how they went about researching and coming to conclusions. They will sometimes describe specific research methods and ways of analysis. At this level in your college career, those sections aren't necessarily the most useful. It might be useful to describe the types and number of research subjects, and perhaps even include specific questions that were asked, but getting into the minute details of the research method probably isn't necessary for the level of academic writing you will be faced with in community college. If you continue onto graduate and/or post graduate school, these sections will become more useful.

This is the section in which authors describe what they found. This can be one of the most useful sections for you as a novice researcher. This is where you might grab quotes or evidence that could be used to make points in your own writing.

The discussion section allows the authors to lay out how their findings impact the body of research in that area. They might show how their research is in line with other similar studies or they might show how their findings dispute other published research. Authors will often point out limitations to their research and make suggestions of other studies that could be conducted to help flesh out questions that remain.

This is the section in which the authors summarize their entire paper. They will often restate their research question, give a summary of their findings, and again point out where further research can be done. This section can be very useful to novice researchers. You might even want to start here.

On that note, I think the most useful sections to novice researchers are the Introduction, Results/Discussion, and Conclusion. You'll be able to practice this in this week's assignments and later in the course.

The embedded information below is from the Distance Education Research Guide has more information on types of academic journal articles, differences between scholarly and popular sources, and defines primary, secondary, and tertiary sources. You might want to bookmark this page as a reference for other research assignments you might encounter.

Original sources used to create content (also licensed under CC BY-NC 4.0 unless otherwise noted):

Krause, S.D. (2016). Quoting, paraphrasing, and avoiding plagiarism. J. Kepka (Ed.) In Oregon writes open writing text . https://openoregon.pressbooks.pub/oregonwrites/chapter/quoting-paraphrasing-and-avoiding-plagiarism

Los Rios Libraries. (2020). Avoiding plagiarism and citing sources. Los Rios libraries information literacy tutorials. https://lor.instructure.com/resources/44fe428e10b347bea9892a63482f55fd?shared

Manage: Organizing information effectively and ethically. (2016). In G. Bobish & T. Jacobson (Eds.), The information literacy user's guide. Milne Publishing. https://milnepublishing.geneseo.edu/the-information-literacy-users-guide-an-open-online-textbook/chapter/manage-organizing-information-effectively-and-ethically/

North Carolina State University Libraries. (2014). Citation: A (very) brief introduction. https://youtu.be/IMhMuVvXCVw . NOTE: This source is licensed under Creative Commons 3.0 BY-NC-SA US

Isserman, Maurice. (2003). Plagiarism: A lie of the mind. Chronicle of Higher Education 49 (34), p. B12.

What are Tone Words? List of 300+ Useful Words

Table of contents

1 What is a Tone?
2.1 Positive Tone Words
2.2 Negative Tone Words
2.3 Neutral Tone Words
3 How to Find the Right Tone for Your Work
4 Tone Words Examples in Various Texts

In writing, tone plays a significant role in conveying emotions, setting the atmosphere, and establishing a connection with readers. The tone of every piece of writing – whether an argumentative essay, a page-turning novel, or a moving poem – can be set by the author’s choice of words.

It would be best to balance the subtleties of language and freedom of thought to get the desired effect. Today, we share a definitive collection of tone words to help you and every writer achieve their goals in this area.

What is a Tone?

Tone, in writing, is the author’s emotional response with deep respect to either the subject matter or the readers. It helps to create a particular atmosphere and direct the reader’s feelings by revealing the author’s point of view.

The tone is the author’s voice (in this case, their attitude, not their personality), and how the words on the page are emotion-tinted to make the reader feel a specific way.

Word choice, tone list sentence structure, imagery, and figurative language are all examples of literary methods that can be used to communicate tone. It comprises more than just the words themselves.

Also, it includes the author’s general tone and approach. To provoke various feelings and responses from the reader, a talented writer might switch tones within the same piece of writing.

Recognizing and employing tone is fundamental to effective communication because it influences the writer’s attitude and how the recipient interprets the message. It can potentially affect a message’s readability, retention, and recall. When writers use the right tone, they may impress readers, hold their attention, and get their point over.

As they delve further into the nuances of tone and tone words in literature, authors gain access to many expressive vocabularies.

The Ultimate List of Tone Words

Now for our ultimate list of tone words in literature. This comprehensive compilation encompasses a range of words that can help authors like you effectively communicate their intended emotions and engage readers on a profound level. And we are talking about a deep level indeed!

Positive Tone Words

Positive tone words inspire happy thoughts and feelings in the reader. They spark positive emotions and are incredibly powerful. Powerful? How so? Positive tone words can make you think well of someone and have a favorable impression of them even without meeting. Here are some other good tone words, word examples, and their meanings:

Joyful: Filled with happiness, delight, and great pleasure.
Optimistic: Expecting positive outcomes, hopeful, and confident about the future.
Enthusiastic: Displaying intense excitement, passion, and eagerness.
Serene: Calm, peaceful, and tranquil.
Endearing: Inspiring warmth or affection
Grateful: Feeling or expressing appreciation and thankfulness.
Empowered: Feeling confident, capable, and in control of one’s life or circumstances.
Radiant: Emitting brightness, happiness, and positivity.
Inspirational: Providing motivation, encouragement, and a sense of upliftment.
Energetic: Full of vigor, liveliness, and vitality.
Confident: Having self-assurance, belief in one’s abilities, and a positive outlook.
Empathetic: showing empathy; feeling the emotions of others.

Negative Tone Words

The use of negative tone words can considerably influence the perceptions produced by readers, regardless of whether they describe a location, a literary work, or a collective of individuals. Simply put, just as positive tone words spark positive emotions, negative ones spark negative ones.

Hence, when applying negative tone phrases, you must exercise caution and ensure your criticisms remain fair, constructive, and objective, and that is very important.

By attentively picking your words, you can provide a well-rounded perspective while keeping your message’s moral sense and integrity. Striking the correct balance between negative and positive judgments provides for a full understanding and encourages effective interactions.

Miserable: Extremely unhappy, sorrowful, or distressed.
Disgusting: Causing intense revulsion, repugnance, or strong aversion.
Aggressive: Hostile, confrontational, or prone to initiating conflict.
Chaotic: Marked by disorder, confusion, and lack of organization or control.
Repulsive: Evoking strong feelings of dislike, aversion, or disgust.
Arrogant: Displaying an exaggerated sense of superiority, self-importance, or entitlement.
Hateful: Full of intense dislike, animosity, or prejudice towards someone or something.
Depressing: Causing feelings of sadness, dejection, or hopelessness.
Inferior: Of lower quality, value, or importance compared to others.

Neutral Tone Words

Neutral tone words help maintain a fair and unbiased approach, allowing readers to form their opinions without undue influence. Here are examples of neutral tone words along with their meanings:

Ordinary: Usual, commonplace, or unremarkable.
Standard: Conforming to established norms, expectations, or criteria.
Common: Frequently occurring, widely experienced, or generally known.
Typical: Representing a characteristic example or pattern.
Balanced: Exhibiting equality, equilibrium, or impartiality.
Objective: Unbiased, based on facts, and devoid of personal opinions or feelings.
Rationale: Logical, reasoned, and based on sound judgment or reasoning.
Neutral: Impartial, unbiased, or not taking sides.
Moderate: Neither excessive nor extreme, characterized by a reasonable and measured approach.
Unbiased: Fair, impartial, and free from prejudice or favoritism.
Indifferent: Having no particular preference or bias, lacking interest or concern.
Matter-of-fact: Devoid of emotional embellishment, straightforward, and focused on facts.

How to Find the Right Tone for Your Work

So far, we have established that finding the right, powerful tone words is the key to successfully delivering your message through writing. To figure out which examples of tone words used work best for your goal, ask these reflective questions:

Why am I writing this?

It helps to know why you’re writing in the first place to set the right words of tone in your writing. What effect do you hope to have on the reader? Each goal requires specific words and unique words for the tone of the literature to be adequate.

Who exactly am I writing for?

When deciding on a tone for your writing, it is essential to keep your readers in mind. Is it an official report, a casual blog post, or a rigorous research paper? The interest and understanding of your readers will increase if you modify your author’s tone to match their expectations and preferences.

What do I hope the reader will take away from this experience?

The tone of your communication can be improved by first determining the key idea or message you wish to express. Whatever you’re trying to do through your writing – persuade, inform, or evoke an emotion – matching your tone to your intended result is crucial!

The tone of any formal writing must be clear, concise, confident, and courteous. Attempt refinement without straying into pretentiousness; balance formality and readability.

However, creative writing allows for greater latitude in tone. Still, focus on effectively using tone words, regardless of the genre. The tone you go for will be influenced by the genre you’re writing in, but ultimately, you want to get your point across, make the reader feel something, and pull them into the story.

Suppose writers take the time to evaluate these factors carefully. In that case, they can master using an appropriate tone to describe words that will connect with their target audience, improve readability, and have the desired effect. We hope you become one such writer.

Tone Words Examples in Various Texts

Again, tone words play a vital role in shaping the atmosphere and evoking emotions in different types of writing. The choice of tone words varies based on the formality of the writing and is further influenced by the genre, whether creative or formal.

Let’s explore some examples of tone words in various types of writing:

Melancholic: Conveys a tone of sadness, longing, or introspection, often found in poems exploring themes of loss or nostalgia.
Whimsical: Creates a playful and lighthearted tone, commonly used in poems that embrace imagination and fantasy.
Serene: Establishes a calm and peaceful tone, often seen in verses that celebrate nature or meditate on inner peace.
Euphoric: Creates a tone of extreme joy, vitality, or ecstasy, frequently found in poems expressing moments of joy or bliss.

Romantic Novel

Passionate: Sets an intense and vibrant tone, typically associated with love, desire, and emotional depth.
Sentimental: Evokes tender and nostalgic emotions, often found in romantic novels emphasizing deep emotional connections.
Yearning: Conveys a tone of longing or desire, frequently employed in books that explore unrequited love or longing for a lost connection.
Tender: Establishes a gentle and affectionate tone, often used to depict tender moments of intimacy or vulnerability.

Horror Fiction

Eerie: Creates a tone of unease, suspense, and impending dread, frequently found in horror fiction to heighten tension.
Sinister: Evokes a dark, malicious, or hateful tone, often depicting evil or foreboding circumstances.
Terrifying: Establishes a tone of extreme fear, horror, or terror intended to invoke a visceral response from the reader.
Macabre: Conveys a tone of fascination with the gruesome death or the supernatural, often seen in horror fiction that explores the darker aspects of human existence.

Informative News Article

Objective: Establish a neutral and unbiased tone, presenting facts and information without personal opinions or emotions.
Authoritative: Conveys a tone of expertise, credibility, and confidence, frequently employed in news articles to establish trustworthiness.
Informative: Sets a tone of clarity, providing straightforward and concise information to educate the readers.
Balanced: Establishes a fair and even-handed tone, presenting multiple perspectives and avoiding bias or favoritism.

What Is the Difference Between Tone and Voice?

Tone and voice are sometimes misunderstood or used interchangeably by writers. However, they couldn’t be more dissimilar. We’ve established that tone conveys the author’s or a character’s feelings about the subject.

Tone words in nonfiction writing reveal the author’s point of view. Tone words are helpful in fiction because they indicate a scene’s or conversation’s emotional state, whether the reader might expect tension, happiness, sadness, etc.

While the term “voice” describes the overall character of a piece of writing. One author’s scathing tone may be another’s instructive or friendliness.

What Is the Difference Between Tone and Mood?

The word for tone in literature is the mood or atmosphere that the author intends for the reader to experience while reading the text, while mood is the experience that the reader has while reading the story.

How the author feels about the setting or the character and how he wants the reader to feel determines the tone. On the contrary, the reader’s emotional response establishes the mood. Setting the mood at the opening of a story prepares the audience for what is to come.

How do you identify tone words?

Identifying tone words in a piece of writing is essential for understanding the intended emotions and attitudes conveyed by the author. Here are some key approaches to identifying tone words:

Contextual Clues: Pay attention to the overall context and the writer’s purpose. Examine the subject matter, the writer’s attitude, and the intended audience. These factors can provide valuable hints about the tone.
Word Choice: Look for words that evoke emotions or convey a particular attitude. Tone words often carry emotional weight or reveal the author’s perspective. Words with strong connotations, such as “brutal,” “uplifting,” or “serene,” can indicate the tone.
Go with your Gut: Sometimes, the best way to explain a tone is to say that you just “get it.” The text makes you feel a specific way, whether it’s urgent or melancholy. After reading it, you feel angry and get the impression that the author is also angry. Or sometimes, there is nothing specifically humorous about the writing, yet you find yourself laughing anyhow. As a result, go with your instinct when deciding how to interpret the author’s tone in passages like these.

Readers also enjoyed

WHY WAIT? PLACE AN ORDER RIGHT NOW!

Just fill out the form, press the button, and have no worries!

We use cookies to give you the best experience possible. By continuing we’ll assume you board with our cookie policy.

Beyond Trend and Periodicity: Guiding Time Series Forecasting with Textual Cues

This work introduces a novel Text-Guided Time Series Forecasting (TGTSF) task. By integrating textual cues, such as channel descriptions and dynamic news, TGTSF addresses the critical limitations of traditional methods that rely purely on historical data. To support this task, we propose TGForecaster , a robust baseline model that fuses textual cues and time series data using cross-attention mechanisms. We then present four meticulously curated benchmark datasets to validate the proposed task, ranging from simple periodic data to complex, event-driven fluctuations. Our comprehensive evaluations demonstrate that TGForecaster consistently achieves state-of-the-art performance, highlighting the transformative potential of incorporating textual information into time series forecasting. This work not only pioneers a novel forecasting task but also establishes a new benchmark for future research, driving advancements in multimodal data integration for time series models.

1 Introduction

Time series forecasting (TSF) is crucial in many fields, thereby attracting significant attention from both academia and industry. Despite extensive research dedicated to this task, recent studies have shown that simple linear models (Zeng et al., 2023 ; Xu et al., 2023 ; Toner & Darlow, 2024 ) that barely extract trend and periodicity information from time series data frequently achieve performance close to that of state-of-the-art complex models (Nie et al., 2023 ; Liu et al., 2023 ; Jin et al., 2023 ) , indicating that current approaches have reached a saturation point, possibly overfitting historical data.

To achieve higher forecasting accuracy, it is essential to provide models with additional, critical information—such as holiday information, consumer sentiment, and climate changes (depending on the specific forecasting task)—that are not present in the time series data itself. Without this additional information, models cannot extract the necessary causal relationships, leading to oversimplification or overfitting. Moreover, the lack of insight about channels and underlying systems prevents the modeling of inter-channel relationships, often resulting in unreliable forecasts.

As such valuable information is usually hard to quantify as auxiliary time series data, we consider them as textual cues and introduce a novel task for TSF: Text-Guided Time Series Forecasting (TGTSF) . TGTSF leverages two main components: channel descriptions and dynamic news messages. Channel descriptions provide static knowledge about the underlying systems, enabling the model to differentiate between channels and better understand inter-channel correlations. News messages offer dynamic and external insights, helping the model adjust to shifts in data distribution caused by external events. By modeling the joint distribution of future values conditioned on these textual cues, TGTSF has the potential to enhance forecasting accuracy and reliability. Furthermore, TGTSF enables scenario-based forecasting, which is particularly valuable for business decision-making. For instance, a company can forecast sales based on the assumption of a successful marketing campaign, incorporating relevant news and channel-specific descriptions to refine predictions.

We then introduce TGForecaster , a simple Transformer-based baseline multimodal model designed to realize the potential of the proposed TGTSF framework. Utilizing a cross-attention mechanism, TGForecaster fuses information from textual cues and time series data to guide the forecasting process. To rigorously evaluate and demonstrate the efficacy of our model, we carefully craft four benchmark datasets categorized into synthetic , captioned existing , and real-world datasets. Each dataset is uniquely designed to validate the task and test the model’s capability from different perspectives. Specifically, the synthetic dataset features simple yet variable time series patterns, such as sinusoidal waves with random frequency changes, to test the model’s ability to utilize textual cues in forecasting and mitigating distribution shifts. Two existing datasets are captioned with different levels of textual information, used to challenge the model in complex, non-periodic environments by integrating detailed external data. Lastly, our real-world dataset, collected from actual sales scenarios, showcases the practical application value of TGTSF.

Experimental results show that TGForecaster consistently demonstrates state-of-the-art performance across our datasets, affirming the soundness of the TGTSF task definition. Ablation studies reveal that without textual assistance, TGForecaster’s performance reverts to that of PatchTST (Nie et al., 2023 ) , its time series encoder backbone, underscoring that the performance enhancement is driven by the additional information provided by textual data, not merely by a more sophisticated architecture.

Our contributions are summarized as follows:

We identify the roadblock of TSF as information insufficiency and propose TGTSF as a new forecasting approach that integrates textual data to enrich the models with external causal information and system knowledge.

We establish the first TGTSF benchmark containing four uniquely designed datasets.

We design a simple baseline model for TGTSF, TGForecaster. TGForecaster achieves state-of-the-art performance by utilizing textual information and effectively validates our proposed TGTSF task.

2 Insights and Motivation

2.1 existing tsf solutions suffer from information insufficiency.

Time series forecasting typically involves predicting future segments based on historical data, specifically forecasting subsequent time series segments from prior ones. However, the inherent sparsity of information within time series data makes achieving accurate forecasts challenging, or even impossible at times. Standard decomposition of time series data identifies three main components: trend, periodicity, and noise (Box et al., 2015 ) . The noise component is intrinsically unpredictable. Trends, which are slowly changing patterns, are often impacted by external events that influence the underlying system, making it less predictable without external information. As a result, the periodicity component is usually the main source of reliable prediction.

Recent advancements in linear models have validated these challenges. For instance, the DLinear (Zeng et al., 2023 ) model utilizes a simple linear layer to effectively capture basic periodic patterns, outperforming most complex Transformer-based models. Similarly, FITS (Xu et al., 2023 ) , designed to simply extract periodicity, achieves comparable or superior results to state-of-the-art (SOTA) methods using fewer than 10 ⁢ k 10 𝑘 10k 10 italic_k parameters. Further research demonstrates that employing a closed-form forecasting matrix (Toner & Darlow, 2024 ) , calculated from training data, can produce SOTA outcomes. These findings indicate that the minimal information in time series data allows models to recognize patterns with few or no learnable parameters, suggesting that training larger models with insufficient information can lead to severe overfitting.

The issue of information insufficiency extends beyond the data itself to the absence of external information. External factors like holidays, consumer sentiment, or climate changes can significantly alter patterns in time series data, affecting variables such as electricity usage, sales trends, or regional weather conditions. However, these factors are not contained within the time series data, hindering models from incorporating causal relationships. This deficiency can compromise model training, causing models to either learn overly simplified average shortcut or overfit on the training data, as shown in Fig. 1 . For example, if a dataset frequently presents days without rain, the model may consistently predict no rain for future days. However, rainfall predictions are largely influenced by external factors like climate change, not included in the time series, leading models to opt for shortcuts that minimize overall loss by predicting the most common outcome.

Moreover, the lack of knowledge about the underlying systems exacerbates the problem of information insufficiency, especially in datasets with multiple time series channels. Without a deeper understanding of these systems or the characteristics of each channel, it is challenging to leverage such information for improved modeling performance. The indistinctiveness among channels complicates modeling efforts, making it difficult to accurately capture and model correlations.

These challenges underscore the need to integrate external information to address the fundamental limitations of time series forecasting methods.

2.2 RIN & Weight Sharing are Compromises under Information Insufficiency

Researchers have developed various methods to tackle information insufficiency in time series forecasting. One such method is Reversible Instance Normalization (RIN) (Kim et al., 2021 ) , which normalizes each instance to align mean and variance across the dataset, as shown in Fig. 1 . RIN effectively reduces distribution shifts caused by trends shifting or external events, forcing the model to learn a unified distribution. While it standardizes data inputs, RIN also strips away essential information such as trend intensity and amplitude variations, which diminishes the model’s capacity to detect relative biases or amplitude shifts beyond the training scope. However, RIN merely hides the issue of information insufficiency by neutralizing the effects of external events.

Another strategy is weight sharing (Nie et al., 2023 ; Zeng et al., 2023 ; Xu et al., 2023 ) . This technique treats all input channels as equivalent, leveraging common patterns across all channels to alleviate the problem of limited training samples. In other words, this approach enhances performance by learning from the shared periodicities across all the channels, typical within a single dataset. However, since the model is trained to minimize loss across all channel distributions with a single set of weights, it prioritizes common patterns, failing to capture unique distributions for each channel, as illustrated in Fig. 1 , which may lead to averaged results. These average results further reveal that the model lacks sufficient information to distinguish between different channels effectively.

Despite these advancements, both methods provide only temporary solutions to the pervasive issue of information insufficiency. They do not resolve the core problem and still suffer from the limitations imposed by insufficient information.

2.3 Motivation for The Proposed TGTSF Framework

External factors significantly impact time series behavior such as trending and periodicity. A line of work apply and investigate how to embed the external information as a new channel, such as Autoformer (Wu et al., 2021 ) , FedFormer (Zhou et al., 2022a ) add auxilary channels to indicate the day-of-week or even the exact date. Temporal Fusion Transformer (TFT) (Lim et al., 2021 ) and TimeXer (Wang et al., 2024 ) , however, focus on how to select the given variables and how to embed and fuse them with other channels. However, these methods only bring limited performance boost. Basically, they are still trying to squeeze out the information in the original time series dataset which contains limited information itself.

However, other forms of information such as news articles, social media updates, global climate change report are crucial for accurately predicting time series pattern in domains like sales and local weather report. These information are often in forms of text and hard to be quantified as numerical data alone. Moreover, the real-world application of forecasting frequently requires conditional predictions based on specific scenarios. For example, businesses may need to assess the impact of a marketing campaign on future sales, necessitating a deep understanding of external textual factors to forecast outcomes effectively.

To address these challenges, it is essential to include text as an additional input modality. This approach not only allows models to react to historical data but also to proactively adapt to potential future events and trends. It enhances forecasting by providing insights into the underlying dynamics and causal relationships often missing from time series data alone. Guiding time series forecasting with textual data also helps clarify inter-channel relationships, reflecting systemic correlations or domain-specific knowledge that are crucial for accurate predictions. This textual context enhances models’ ability to interpret complex data interactions, vital in environments where data channels are interconnected.

In summary, guiding time series forecasting with textual data addresses the critical issue of information insufficiency. This approach transcends the limitations of traditional TSF methodologies, enhancing model accuracy and effectiveness. By incorporating textual insights, forecasting becomes more robust and practically valuable for real-world applications.

3 TGTSF Task Formulation

We introduce a novel task within the domain of time series analysis, termed Text-Guided Time Series Forecasting (TGTSF) , designed to address the prevalent issue of information insufficiency in time series forecasting.

As illustrated in Fig. 1 , text guidance in TGTSF operates on two aspects. Firstly, channel descriptions serve as identifiers for each channel, aiding the model in distinguishing between them while learning shared features. These descriptions can also incorporate static knowledge about the underlying system, enhancing the model’s ability to recognize inter-channel correlations. Secondly, news messages provide dynamic and external insights into known (in training) or hypothesized (in inference) future events, which assist the model in adapting to event-driven distribution shifts. 1 1 1 In creating a TGTSF dataset, it is advisable to avoid directly generating the description out of the forecasting horizon time series pattern as news messages, as this could lead to information leakage. News messages should instead contain relevant, known information from other sources.

The TGTSF task is defined by three principal inputs: the time series data, news messages , and channel descriptions . The challenge lies in the fact that news items are not directly linked to specific time series channels. Consequently, the model must infer the impact of each news item on individual channels based on their descriptions. This setup not only capitalizes on dynamic events but also integrates domain knowledge via channel descriptions. Such integration is crucial for the model to comprehend and learn the spatial correlations among channels, thereby enhancing the forecasting accuracy with nuanced contextual understanding.

Formally, the TGTSF task seeks to model the potentially complex joint distribution of future values in a multi-channel sequence. This modeling involves the integration of historical time series data, textual information from news messages, and channel descriptions over a specified look-back window:

where L 𝐿 L italic_L denotes the look-back window length, h ℎ h italic_h stands for the prediction length, the j 𝑗 j italic_j means the channel number, c 𝑐 c italic_c is the total number of channels.

4 TGForecaster: A Baseline Model for The TGTSF Task

To validate the TGTSF task, we developed TGForecaster, a streamlined transformer model designed for multimodal fusion with cross-attention. Illustrated in Fig. 2 , this model harnesses textual information to enhance the accuracy and relevance of time series forecasting. It employs a reimplemented PatchTST encoder for processing time series data and leverages off-the-shelf, pretrained text models for pre-embedding textual inputs into vector sequences across the time dimension, allowing for effective modality fusion in the embedding space.

A key innovation in TGForecaster is the integration of a cross-attention layer to achieve the ’text guided channel independent’. This layer is pivotal in fusing news content with channel descriptions. Taking cues from the attention mechanism’s original use in recommendation systems, we treat the news as both the key and value, with the channel descriptions serving as queries. Through this mechanism, we calculate the relevance of each news item to every channel. The resulting scores are then used to aggregate the news, culminating in a composite embedding for each channel. Another crucial aspect of our design is the modality fusion, achieved through a separate layer of cross-attention where text embeddings function as query, while temporal embeddings act as the key and value. This ensures that the output dimensions align naturally with the query which has the length of forecasting horizon. The attention maps for these two cross-attention layers are provided in Appendix E . These maps compellingly demonstrate that the layers perform in accordance with our design specifications.

As decoder, we use the token-wise decoding to avoid the overfitting on the last layer as observed in previous work (Lee et al., 2023 ) . Each token is project back to its original patch length and weighted sumed to get the final forecasting result. Notably, TGForecaster does not incorporate the RIN. Our experiments, detailed in Appendix D , demonstrate that the inclusion of RIN degrades performance when sufficient information is already available.

This design strategy not only enables the integration of textual data into the time series forecasting model but also facilitates a direct comparison of the impact of text on forecasting accuracy.

5 TGTSF Benchmark Datasets

We have designed four TGTSF benchmark datasets across three categories: synthetic, captioned existing, and real-world datasets. We plan to release these TGTSF datasets along with captions, captioning scripts, and pre-embeddings, adhering to all applicable intellectual property constraints.

5.1 Synthetic Toy Dataset

The Synthetic Toy Dataset evaluates models’ ability to utilize textual information for time series forecasting. It features diverse patterns, including segments of sinusoidal waves with variations in frequency, amplitude, and trend. Textual descriptions resembling news reports precede each change point, outlining upcoming alterations to prompt the TGTSF model to adjust its forecasts accordingly. This setup tests the model’s effectiveness in mitigating distribution shifts and harnessing textual cues for improved forecasting accuracy.

5.2 Electricity-Captioned Dataset

The Electricity-Captioned Dataset builds on a commonly-used electricity utility dataset (Zhou et al., 2021 ) that tracks appliance usage in an office building, with daily patterns largely influenced by whether it is a workday. This dataset has been enhanced with straightforward, publically-known textual information such as the type of day (e.g., day of the week, public holiday, workday), using the channel name as the descriptor. This enhancement serves a dual purpose: to illustrate whether minimal textual data can improve prediction accuracy and to facilitate a direct performance comparison with traditional time series models.

5.3 Weather-Captioned Dataset

The Weather-Captioned Dataset is designed to overcome the limitations of commonly used datasets in time series forecasting (TSF), especially the lack of predictability and periodicity in variables like rainfall, wind speed, and direction. By incorporating external climate information, this dataset demonstrates how integrating textual data can address the challenges of information insufficiency in traditional TSF, enabling enhanced control over forecasting outcomes.

Originally limited to a single year (Zhou et al., 2021 ) , we have expanded the weather dataset to encompass a decade of detailed weather data from 2014 to 2023, sourced consistently from the same weather station. The Weather-Captioned dataset now includes weather forecasting reports from a publicly accessible service, detailing the climate conditions in Jena, Germany, where the weather station is approximately located. These reports, updated every six hours and daily, cover a range of parameters including weather conditions, temperature, humidity, wind speed, and direction, providing a comprehensive overview for our analysis. We directly use the channel name as channel description.

To transform this quantitative data into actionable insights for TSF, we employed GPT-4 to summarize each forecasting report into a set of seven thematic sentences, as demonstrated in Tab. 1 . GPT-4 is prompted to avoid specific numerical details from the original reports to prevent information leakage and to simplify the model’s learning process. To further enrich the dataset, GPT-4 generated three distinct summary versions for each report, resulting in 3 7 = 2187 superscript 3 7 2187 3^{7}=2187 3 start_POSTSUPERSCRIPT 7 end_POSTSUPERSCRIPT = 2187 possible unique textual captions for each time step. The full details of the prompt used for generating these summaries are provided in the appendix.

We divided this extensive dataset into two subsets for detailed analysis: Weather-Captioned-Medium, covering data from 2014 to 2018, and Weather-Captioned-Large, which includes the entire dataset spanning ten years. This dataset encompasses over 525,600 weather data records across 21 different channels, with more than 85,000 unique sentences. This vast corpus provides millions of potential training samples for captioned time series forecasting. Further information is detailed in the appendix.

5.4 Steam Game Dataset

A key application of TGTSF is in sales forecasting and decision-making. To illustrate this, we compiled data on online gamer counts from several of the most popular games on Steam, the leading online video game distribution platform. This dataset also includes records of all game updates and events. The time series exhibits a weekly periodicity, with gamer activity typically peaking on weekends. Announcements from game developers frequently trigger spikes in activity, reflecting player enthusiasm for new content. However, the varied management strategies of game developers and the diverse reactions of gamers to updates introduce significant distribution shifts, even when textual information is considered. This variability makes the Steam Dataset the most challenging and intricate dataset within the TGTSF framework. 2 2 2 Due to intellectual property constraints, we may not be able to directly release this dataset. Detailed instructions are available in the appendix for researchers interested in replicating or extending our data.

6 Experiments

We evaluate the TGForecaster model across four datasets to demonstrate the feasibility of the TGTSF task on the proposed datasets. As discussed in Section 5 , each dataset is uniquely designed to test the task from different perspectives.

Baseline Models

TGForecaster is benchmarked against state-of-the-art (SOTA) methods including DLinear, FITS, PatchTST, iTransformer, and Time-LLM (Zeng et al., 2023 ; Xu et al., 2023 ; Nie et al., 2023 ; Liu et al., 2023 ; Jin et al., 2023 ) . We specifically contrast it with the linear-based models DLinear and FITS to illustrate how TGForecaster surpasses periodicity-focused models. Comparisons with PatchTST highlight how textual information can enhance forecasting performance, even when using the same time series encoder. Time-LLM "reprograms" the LLaMa2 (Touvron et al., 2023 ) , a large language model, which gives it capability of understanding text information. We use it as a baseline of multi-modal model. Specific settings differences, if any, are noted accordingly.

6.1 Evaluation on Toy Dataset

Experiment settings.

All of the models are following the same experimental setup with prediction length H ∈ { 14 , 28 , 60 , 120 } 𝐻 14 28 60 120 H\in\{14,28,60,120\} italic_H ∈ { 14 , 28 , 60 , 120 } and LBW length T = 60 𝑇 60 T=60 italic_T = 60 .

Statistical Results

Table 2 presents the performance comparison of various models on the toy dataset, with TGForecaster significantly outperforming all baseline models. Specifically, TGForecaster achieved an 80% improvement in Mean Squared Error (MSE) over the best performing transformer-based model, and a 96% increase over DLinear. These results underscore the considerable benefits of integrating textual guidance in TSF. Notably, a version of TGForecaster without news data, which lacks the ability to utilize auxiliary textual information, demonstrated substantially lower performance. This underscores the essential role of text in enhancing forecast accuracy.

Figure 3 illustrates a segment of the toy dataset where the frequency changes within the forecasting horizon. In this visualization, the PatchTST model maintains the frequency observed in the look-back window, indicating an inability to adapt to new frequencies. Similarly, DLinear displays a collapsed pattern. Notably, without news input, TGForecaster shows the same behavior of PatchTST, suggesting that it relies solely on its time series encoder in the absence of textual cues. The models without external data also highlights the decreasing amplitude of predictions, a tendency to revert to safer, average predictions when faced with uncertainty in the far future—an issue known as the ’average shortcut’ phenomenon discussed in Section 2.1 . Conversely, with the integration of external text information eliminating uncertainty, TGForecaster adeptly adapts to new frequencies at the appropriate moments, demonstrating the substantial benefits of incorporating textual data into the forecasting process.

6.2 Evaluation on Electricity-Captioned Dataset

We follow the experiment settings in previous works as follows: forecasting horizon H ∈ { 96 , 192 , 336 , 720 } 𝐻 96 192 336 720 H\in\{96,192,336,720\} italic_H ∈ { 96 , 192 , 336 , 720 } , look back window length of 288. For fair comparison, we directly compare with the results report in the baseline original paper. And the TGForecaster is trained with the captioned version. On this dataset, we also test the impact of a shorter look-back window on the TGForecaster. We will report accordingly in Tab. 2 .

As depicted in Tab. 2 , TGForecaster demonstrates SOTA performance on the Electricity dataset, particularly effective at shorter forecasting horizons using minimal textual cues. However, TimeLLM slightly outperforms TGForecaster over longer forecasting periods. Given TimeLLM’s backbone is a large language model that also incorporates textual information, its edge may stem from its use of basic date and statistical data within its text inputs.

6.3 Evaluation on Weather-Captioned Dataset

We follow the experiment settings in previous works as follows: forecasting horizon H ∈ { 96 , 192 , 336 , 720 } 𝐻 96 192 336 720 H\in\{96,192,336,720\} italic_H ∈ { 96 , 192 , 336 , 720 } , look back window length of 360. We trained all other models on the Weather-Medium and -large dataset with their setting on the original weather dataset accordingly. And the TGForecaster is trained with the captioned version. We will report accordingly in Tab. 3 .

As demonstrated in Tab. 3 , TGForecaster significantly outperforms other models across both the Weather-Medium and Weather-Large datasets, substantiating the efficacy of incorporating external text information in addressing the information insufficiency inherent in TSF models. Notably, despite its simplicity, the FITS model performs exceptionally well, second only to TGForecaster. This suggests that simply adding more training time series to complex models does not necessarily lead to improved performance, as more time series does not necessarily contain more information. The results highlight that strategic integration of textual data can provide a more substantial performance boost than merely increasing the quantity of time series data. Further, we also report the channel-wise performance in Appendix B . We notice a groundbreaking performance boost of over 60% on certain channels which were not predictable with historical time series data alone.

Ablation Study

We conducted ablation studies to evaluate the impact of different embedding models and the integration of textual inputs on the performance of TGForecaster. We tested three embedding models: OpenAI Embedding ( ope, ) , paraphrase-MiniLM-L6 (Wang et al., 2020 ) , and all-mpnet-base (Song et al., 2020 ) 3 3 3 Sentence Transformer: https://huggingface.co/sentence-transformers . The results, presented in Tab. 4 , indicate minimal performance differences between the embedding models. Notably, the removal of channel descriptions led to a significant performance decrease, underscoring the model’s reliance on this feature for distinguishing between channels. Similarly, omitting news text resulted in performance dropping to levels comparable to the baseline PatchTST model, confirming that the improvements in forecasting accuracy are primarily driven by the inclusion of external textual information. This observation validates our hypothesis that textual data plays a crucial role in compensating for information deficiencies in traditional time series forecasting.

Visualization and Controllability Test

Fig. 4 visualizes three channels from the Weather-Caption-Medium dataset, full result see Appendix C . The first channel, atmospheric pressure (p), which is influenced by regional climate conditions, typically exhibits slowly changing trends that are challenging for TSF models to predict due to their subtle fluctuations without periodicity. However, with the integration of external information, TGForecaster accurately predicts these trends, whereas PatchTST tends to predict a constant average value, failing to capture the gradual changes.

The second channel, time of raining, measured in seconds per 10 minutes, presents unique challenges: it lacks periodicity, displays no causality, and often exists in a binary state. Traditional models like PatchTST often fail, typically predicting no rainfall - the averaged shortcut. TGForecaster adjusts its forecasts based on text prompts about upcoming rain. However, it sometimes misses predictions when the text weather forecasting report does not perfectly align with actual weather events, as noted with the missed second rainfall period.

The third channel, SWDR (solar radiation), directly reflects the amount of solar power reaching the ground. TGForecaster accurately forecasts SWDR, even though these are not directly mentioned in the news texts. It effectively infers inter-channel dependencies, enabling it to predict shifts in SWDR associated with weather changes, in contrast to PatchTST’s basic waveform predictions.

Additionally, a controllability test involving the swapping of news inputs for the second and fourth days illustrates TGForecaster’s responsive accuracy: it predicts rain on the fourth day and clear conditions on the second, aligning forecasts with the altered news data. This test underscores the model’s adaptability and the effectiveness of text-guided forecasting.

6.4 Evaluation on Steam-100 Dataset

We evaluate the performance of the TGForecaster on our Steam-100 dataset, utilizing an input window of 60 days and an output horizon of 14 days. The findings indicate that the TGForecaster outperforms baseline models such as PatchTST, achieving a performance enhancement of over 12.6 % percent 12.6 12.6\% 12.6 % . This superior performance is consistently observed across over 59.6 % percent 59.6 59.6\% 59.6 % of all games, ranking as the best among all evaluated methods. Given the significant stylistic variations among different game developers and the potential risk of temporal distribution shifts, traditional time series forecasting methods often struggle to capture the commonalities in temporal features. In contrast, the TGForecaster leverages textual information to significantly augment its predictive capabilities. We report the full result in the Appendix G .

7 Conclusions and Discussion

This paper addresses a critical roadblock in time series forecasting: information insufficiency. We introduced Text-Guided Time Series Forecasting (TGTSF), a new approach that integrates textual cues to enrich the models with external information and system knowledge. We developed and released four TGTSF datasets, each crafted to validate different aspects of the task and model. Our straightforward yet effective TGForecaster model demonstrates that textual guidance can significantly enhance time series modeling by mitigating the average predictions typically resulting from information scarcity.

While the TGForecaster effectively validates the TGTSF task, it does not fully comprehend the semantics of the text, such as extracting correlations among channels automatically. Future work will focus on advancing the model’s semantic understanding and its ability to autonomously discern intricate relationships within the data.

(1) Openai embedding api. https://platform.openai.com/docs/guides/embeddings. URL https://platform.openai.com/docs/guides/embeddings .
Box et al. (2015) George EP Box, Gwilym M Jenkins, Gregory C Reinsel, and Greta M Ljung. Time series analysis: forecasting and control . John Wiley & Sons, 2015.
Jin et al. (2023) Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, et al. Time-llm: Time series forecasting by reprogramming large language models. arXiv preprint arXiv:2310.01728 , 2023.
Kim et al. (2021) Taesung Kim, Jinhee Kim, Yunwon Tae, Cheonbok Park, Jang-Ho Choi, and Jaegul Choo. Reversible instance normalization for accurate time-series forecasting against distribution shift. In International Conference on Learning Representations , 2021. URL https://openreview.net/forum?id=cGDAkQo1C0p .
Lee et al. (2023) Seunghan Lee, Taeyoung Park, and Kibok Lee. Learning to embed time series patches independently. arXiv preprint arXiv:2312.16427 , 2023.
Lim et al. (2021) Bryan Lim, Sercan Ö Arık, Nicolas Loeff, and Tomas Pfister. Temporal fusion transformers for interpretable multi-horizon time series forecasting. International Journal of Forecasting , 37(4):1748–1764, 2021.
Liu et al. (2023) Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. itransformer: Inverted transformers are effective for time series forecasting. arXiv preprint arXiv:2310.06625 , 2023.
Nie et al. (2023) Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. A time series is worth 64 words: Long-term forecasting with transformers. In International Conference on Learning Representations , 2023.
Song et al. (2020) Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, and Tie-Yan Liu. Mpnet: Masked and permuted pre-training for language understanding. Advances in neural information processing systems , 33:16857–16867, 2020.
Toner & Darlow (2024) William Toner and Luke Darlow. An analysis of linear time series forecasting models. arXiv preprint arXiv:2403.14587 , 2024.
Touvron et al. (2023) Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 , 2023.
Wang et al. (2020) Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, and Ming Zhou. Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. Advances in Neural Information Processing Systems , 33:5776–5788, 2020.
Wang et al. (2024) Yuxuan Wang, Haixu Wu, Jiaxiang Dong, Yong Liu, Yunzhong Qiu, Haoran Zhang, Jianmin Wang, and Mingsheng Long. Timexer: Empowering transformers for time series forecasting with exogenous variables. arXiv preprint arXiv:2402.19072 , 2024.
Wu et al. (2021) Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting. Advances in Neural Information Processing Systems , 34:22419–22430, 2021.
Wu et al. (2023) Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. Timesnet: Temporal 2d-variation modeling for general time series analysis. In International Conference on Learning Representations , 2023.
Xu et al. (2023) Zhijian Xu, Ailing Zeng, and Qiang Xu. Fits: Modeling time series with 10 ⁢ k 10 𝑘 10k 10 italic_k parameters. arXiv preprint arXiv:2307.03756 , 2023.
Zeng et al. (2023) Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. Are transformers effective for time series forecasting? 2023.
Zhou et al. (2021) Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. Informer: Beyond efficient transformer for long sequence time-series forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 35, pp. 11106–11115, 2021.
Zhou et al. (2022a) Tian Zhou, Ziqing Ma, Qingsong Wen, Xue Wang, Liang Sun, and Rong Jin. Fedformer: Frequency enhanced decomposed transformer for long-term series forecasting. In International Conference on Machine Learning , 2022a.
Zhou et al. (2022b) Tian Zhou, Ziqing Ma, xue wang, Qingsong Wen, Liang Sun, Tao Yao, Wotao Yin, and Rong Jin. FiLM: Frequency improved legendre memory model for long-term time series forecasting. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho (eds.), Advances in Neural Information Processing Systems , 2022b. URL https://openreview.net/forum?id=zTQdHSQUQWc .

Appendix A Data and Code Availability

Given the extensive size of the TGTSF dataset, especially the ones with pre-embeddings, an immediate anonymous release may not be feasible. Therefore, we plan to make the full datasets publicly available ASAP.

Additionally, the script used to create the toy dataset will be released alongside our code base. You can access the code repository at: https://github.com/VEWOXIC/TGTSF .

We will update the github repo here later and will update the dataset in it later.

Appendix B Channel-wise performance on Weather-Captioned Dataset

The difficulty in predicting each channel varies, therefore, we present channel-wise performance in Table 5 . The results demonstrate that TGForecaster, with the aid of external textual climate reports, significantly enhances forecasting accuracy across all channels. Notably, the model achieves over a 60% performance improvement in channels such as atmospheric pressure (p (mbar)), relative humidity (rh(%)), and vapor pressure deficit (VPdef (mbar)), which typically cannot be predicted reliably using historical time series data alone. The integration of external text cues has led to groundbreaking improvements in forecasting these parameters.

However, the wind velocity channel shows minimal variance in performance across all models, each achieving similar results with slight losses. This phenomenon is attributed to the presence of extreme values in this channel, which, after normalization, diminish the impact of more typical values on the overall gradient. Consequently, all models struggle to learn detailed patterns in this channel due to the reduced contribution to the global gradient.

Another noteworthy observation is that while TGForecaster is capable of predicting rainfall—unlike models that default to predicting near-zero averages—the performance improvement in these channels is modest. This is because rainfall is relatively scarce in this dataset, leading to large losses when rain is inaccurately predicted at the wrong times. Conversely, predicting the average value results in a smaller overall loss. This tendency explains why other models often opt for the average, avoiding the complex task of learning rainfall patterns. Nevertheless, accurate rainfall forecasting remains crucial in meteorological applications, underscoring our commitment to enhancing predictive accuracy in this area. The same principle also applies to other channels.

Appendix C Visualization Weather-Captioned Dataset

We provide the full visualization as Fig. 5 . The TGForecaster shows great performance across all the channels. Even very hard ones such as Wind dir. It can also model the time series that totally independent with the weather such as the CO2 channel.

Appendix D Weather Results w. w/o. RIN

We compared the performance of models with and without Reversible Instance Normalization (RIN) on the weather-captioned-medium dataset, focusing on a 720-hour forecasting horizon. The model with RIN enabled achieved an MSE of 0.3428, whereas the model without RIN achieved a lower MSE of 0.2814. Results visualized in Fig. 6 show that the RIN-enabled model exhibits significant biases in many channels, particularly those with gradual trend shifts. This occurs because RIN removes the bias term from all instances, leaving the model unable to recognize relative bias and trend values. For instance, with RIN, temperature patterns in winter and summer are treated similarly, ignoring the typically higher and more variable temperatures in summer. Additionally, we noted pronounced shifting behavior coinciding with changes in captions, suggesting that the absence of bias information leads the model to over-rely on textual prompts, compensating for the missing data.

Appendix E Attention Map Visualization on Weather-Captioned

We further visualize two cross-attention blocks to further investigate the TGForecaster. You are strongly advised to check the Tab. 1 , Appendix K.3 and Fig. 5 while reading this part.

Figure 7 illustrates the attention map of the "text-guided channel independent" cross-attention block in the text encoder across three layers. In the first layer, attention is predominantly focused on the first sentence, which specifies the month and time. This sentence is crucial as it provides temporal context that significantly impacts the prediction of both daily and annual periodicity. While other sentences receive moderate attention, the sixth sentence, which describes atmospheric pressure as detailed in Table 1 , consistently receives no attention across all channels.

In the second layer, however, there is a notable shift in attention dynamics. All channels, particularly channel 0, show intense focus on the sixth sentence. According to the channel definitions in Appendix K.3 , channel 0 directly corresponds to atmospheric pressure. Channels 10 and 20, which are related to air density and CO2 concentration respectively—factors closely associated with pressure—also display relatively high attention scores. This suggests that the TGForecaster is capable of discerning the underlying relationships among the channels.

The separation of attention focus between the first and second layers suggests that the influence of atmospheric pressure on the model’s predictions is independent of time. In the third layer, a diversity of attention patterns emerges; channel 0 focuses exclusively on the sixth sentence, while other channels predominantly attend to the first sentence.

Since we take the output of previous layer as query and input news embeddings as key and value, the information lies in the news are progressively added to the channel embeddings. Thus, the model can focus on different perspective in separate cross attention layers.

Figure 8 presents the attention map of the modality mixer layer cross attention block in the weather-captioned dataset. The map, averaged across three cross attention layers, illustrates distinct patterns of attention for each channel. This diversity underscores the TGForecaster’s ability to adaptively extract time series embeddings tailored to the unique distribution characteristics of each channel, facilitated by textual inputs.

Notably, the channels for SWDR, PAR, and max.PAR display clear periodic patterns in their attention maps, aligning with observations from waveform visualizations. These patterns suggest that the TGForecaster effectively captures and utilizes periodic information from these environmental variables.

Furthermore, the channels labeled rain and raining show a particularly interesting behavior; they assign significantly higher attention scores to the exact time periods of rainfall within the look-back window. This behavior indicates that the TGForecaster is adept at identifying and prioritizing crucial temporal events specific to each channel, further enhancing its forecasting accuracy by focusing on relevant patterns where needed. This level of detail in attention allocation demonstrates the model’s capability to integrate contextual cues from textual data and further guide the time series forecasting.

Appendix F Comparison with More Baselines on Electricity-Captioned

We further compare with more baselines on Electricity-Captioned, including Autoformer, Fedformer, Informer, FiLM and TimesNet (Wu et al., 2021 ; Zhou et al., 2022a , 2021 , b ; Wu et al., 2023 ) . TGForecaster shows dominant superior performance across these baselines, as shown in Tab. 6 .

Appendix G Full Result on Steam-100 Dataset

We show the comparison on Steam-100 Dataset in Tab. 7 , and Tab, 8 with PatchTST. We use de-normed MAE as metric since the base volume of players of each game varies drastically, using normed metrics can lead to unfair comparison. The pretrain indicate that the model is jointly trained on all the games and each game is labeled by the channel discription. The Gnorm indicate that we apply the global normalization to preserve the player variation mentioned before. But it seems bring limited boost.

Appendix H Error Bar & Critical Difference Diagram

We run the experiments on Toy and Electricity for five times with different randomly chosen random seeds. And Weather-Medium for three times because of the large amount of data can result in very long training time on our devices. We report the mean and standard deviation as follows with comparison with FITS, the most stable model.

As Tab. 9 indicate, TGForecaster shows stable performance across the benchmark. Even with extreme condition, it still maintains superior performance. It worth note that, we thought the relative large variance on weather dataset is caused by the different combination of the text description. But the FITS also shows large variance on this dataset which indicate it is hard to converge on this dataset.

We generate the critical difference plot on our result of four datasets (toy, Electricity, Weather-Medium, Weather-Large) with the default alpha as 0.05 as shown in Fig. 9 . TGForecaster’s placement at the top of the critical difference plot, without intersecting with other lines, demonstrates its consistent and superior performance in terms of MSE compared to the other models. It indicates that with the help of external textual information, TGForecaster can handle complicated datasets.

Appendix I Toy Dataset Details

I.1 data source.

We directly generate this dataset with sinusoidal wave that randomly changes frequency. Before each changing point, we add 10 captions as ’Channel 1 will change to frequency x in y timesteps.’ After each changing point, we add 5 captions as ’Channel 1 will keep steady with frequency of x.’ In other timesteps, we caption it as ’The waveform will go steady.’

We will publish this dataset with CC BY-NC-SA 4.0 licence.

I.2 Dataset Structure

Appendix j electricity-caption details, j.1 data source.

We caption the day of week with the given time stamp. But we somehow find the original time stamp is incorrect. Instead of the year of 2016, it should be collected in year 2012. Without knowing the exact location of this building, we cannot identify the specific public holiday. We then uses channel 319, which shows obvious patterns of workday and holiday as indicator, when the average value lower than a specific value, we caption it with public holiday.

J.2 Dataset Structure

Appendix k weather-caption details, k.1 data source.

We get the weather time series data from: https://www.bgc-jena.mpg.de/wetter/ and weather report from https://www.timeanddate.com/weather/germany/jena/historic . We will publish this dataset with CC BY-NC-SA 4.0 licence since the data source forbids commercial use.

We generate the summary of climate report using following prompt on GPT4.

K.3 Channel Details

The meaning of each channel are as follows. We ask GPT4 to generate this document. There might be incorrectness. These sentences are then embedded as channel description embeddings used in the dataset.

p (mbar): Atmospheric pressure measured in millibars. It indicates the weight of the air above the point of measurement.

T (degC): Temperature at the point of observation, measured in degrees Celsius.

Tpot (K): Potential temperature, given in Kelvin. This is the temperature that a parcel of air would have if it were brought adiabatically to a standard reference pressure, often used to compare temperatures at different pressures in a thermodynamically consistent way.

Tdew (degC): Dew point temperature in degrees Celsius. It’s the temperature to which air must be cooled, at constant pressure and water vapor content, for saturation to occur. A lower dew point means dryer air.

rh (%): Relative humidity, expressed as a percentage. It measures the amount of moisture in the air relative to the maximum amount of moisture the air can hold at that temperature.

VPmax (mbar): Maximum vapor pressure, in millibars. It represents the maximum amount of moisture that the air can hold at a given temperature.

VPact (mbar): Actual vapor pressure, in millibars. It’s the current amount of water vapor present in the air.

VPdef (mbar): Vapor pressure deficit, in millibars. The difference between the maximum vapor pressure and the actual vapor pressure; it indicates how much more moisture the air can hold before saturation.

sh (g/kg): Specific humidity, the mass of water vapor in a given mass of air, including the water vapor. It’s measured in grams of water vapor per kilogram of air.

H2OC (mmol/mol): Water vapor concentration, expressed in millimoles of water per mole of air. It’s another way to quantify the amount of moisture in the air.

rho (g/m³): Air density, measured in grams per cubic meter. It indicates the mass of air in a given volume and varies with temperature, pressure, and moisture content.

wv (m/s): Wind velocity, the speed of the wind measured in meters per second.

max. wv (m/s): Maximum wind velocity observed in the given time period, measured in meters per second.

wd (deg): Wind direction, in degrees from true north. This indicates the direction from which the wind is coming.

rain (mm): Rainfall amount, measured in millimeters. It indicates how much rain has fallen during the observation period.

raining (s): Duration of rainfall, measured in seconds. It specifies how long it has rained during the observation period.

SWDR (W/m²): Shortwave Downward Radiation, the amount of solar radiation reaching the ground, measured in watts per square meter.

PAR (umol/m2̂/s): Photosynthetically Active Radiation, the amount of light available for photosynthesis, measured in micromoles of photons per square meter per second.

max. PAR (umol/m2̂/s): Maximum Photosynthetically Active Radiation observed in the given time period, indicating the peak light availability for photosynthesis.

Tlog (degC): Likely a logged temperature measurement in degrees Celsius. It could be a specific type of temperature measurement or recording method used in the dataset.

CO2 (ppm): Carbon dioxide concentration in the air, measured in parts per million. It’s a key greenhouse gas and indicator of air quality.

K.4 Dataset Structure

Appendix l steam details, l.1 data source.

We are not directly publishing this dataset because of the intellectual property restrictions.

For those who are interested in reproducing the dataset: We directly crawl all the event data on Steam News of each Game. We may release the script for this later.

As for the online players, we use a randomly picked publicly available database that provide the downloading of csv file containing online gamer number. We will not release any data or tools to get these data.

Appendix M Implementation Details and Hyper-Parameters

We train our model on single NVIDIA A800 GPU.

For electricity dataset, we directly report the result from the original paper. For weather dataset, we uses the exact set of hyper-parameter for the original weather datasets provided by each baseline model.

In most of the experiments, we simply use a patch length of 6 and stride of 3. For Toy dataset, we use patch length of 16 and stride of 8.

We follow the previous works, split all the dataset by 7:1:2 for training, validation and testing.

Except the performance on the Weather-Captioned Dataset, all other experiments are ran on the MiniLM Embedding.

Further detailed hyperparameter settings are provided in the training scripts in our codebase. We did not perform comprehensive hyper-parameter tuning because of the constraint of compute power. Thus, we may report a sub-optimal result of TGForecaster.

Search Menu
Sign in through your institution
Advance articles
Editor's Choice
Focus Issue Archive
Open Access Articles
JAMIA Journal Club
Author Guidelines
Submission Site
Open Access
Call for Papers
About Journal of the American Medical Informatics Association
About the American Medical Informatics Association
Journals Career Network
Editorial Board
Advertising and Corporate Services
Self-Archiving Policy
Dispatch Dates
For Reviewers
Journals on Oxford Academic
Books on Oxford Academic

Article Contents

Background and significance, acknowledgments, author contributions, supplementary material, conflicts of interest, data availability, evaluation of gpt-4 ability to identify and generate patient instructions for actionable incidental radiology findings.

Article contents
Figures & tables
Supplementary Data

Kar-mun C Woo, Gregory W Simon, Olumide Akindutire, Yindalon Aphinyanaphongs, Jonathan S Austrian, Jung G Kim, Nicholas Genes, Jacob A Goldenring, Vincent J Major, Chloé S Pariente, Edwin G Pineda, Stella K Kang, Evaluation of GPT-4 ability to identify and generate patient instructions for actionable incidental radiology findings, Journal of the American Medical Informatics Association , 2024;, ocae117, https://doi.org/10.1093/jamia/ocae117

Permissions Icon Permissions

To evaluate the proficiency of a HIPAA-compliant version of GPT-4 in identifying actionable, incidental findings from unstructured radiology reports of Emergency Department patients. To assess appropriateness of artificial intelligence (AI)-generated, patient-facing summaries of these findings.

Radiology reports extracted from the electronic health record of a large academic medical center were manually reviewed to identify non-emergent, incidental findings with high likelihood of requiring follow-up, further sub-stratified as “definitely actionable” (DA) or “possibly actionable—clinical correlation” (PA-CC). Instruction prompts to GPT-4 were developed and iteratively optimized using a validation set of 50 reports. The optimized prompt was then applied to a test set of 430 unseen reports. GPT-4 performance was primarily graded on accuracy identifying either DA or PA-CC findings, then secondarily for DA findings alone. Outputs were reviewed for hallucinations. AI-generated patient-facing summaries were assessed for appropriateness via Likert scale.

For the primary outcome (DA or PA-CC), GPT-4 achieved 99.3% recall, 73.6% precision, and 84.5% F-1. For the secondary outcome (DA only), GPT-4 demonstrated 95.2% recall, 77.3% precision, and 85.3% F-1. No findings were “hallucinated” outright. However, 2.8% of cases included generated text about recommendations that were inferred without specific reference. The majority of True Positive AI-generated summaries required no or minor revision.

GPT-4 demonstrates proficiency in detecting actionable, incidental findings after refined instruction prompting. AI-generated patient instructions were most often appropriate, but rarely included inferred recommendations. While this technology shows promise to augment diagnostics, active clinician oversight via “human-in-the-loop” workflows remains critical for clinical implementation.

Incidental findings are abnormalities discovered on imaging tests performed for unrelated reasons. Amidst recent advances in imaging, the widespread adoption of high-resolution technologies has led to a corresponding rise in incidental findings. 1–7

The interpretation, communication, and management of incidental findings, especially when their clinical significance is uncertain, can be complex and resource-intensive. Challenges are accentuated in the emergency department (ED) where high patient throughput, care complexity, and frequent provider transitions can hinder appropriate acknowledgement and communication of incidental findings. Approximately 5%-30% of imaging studies include actionable incidental findings (AIFs). 1 , 3–5 Only 10%-49% of incidental findings may be documented within the discharge instructions for ED patients 8–11 and as few as 17% of patients discharged from the ED follow-up on their AIFs. 12 This gap in care results in adverse patient health outcomes, uneven quality of care delivery, missed opportunities for earlier diagnosis of potential malignancy, and medicolegal repercussions. 2 , 3 , 10–15

Best practice guidelines recommend that clinicians communicate the presence of AIFs to the patient at the time of the ED visit and document this information into the patient’s discharge instructions. 4 Although rates of verbal communication are not easily measured, specific documentation of AIFs into a patient’s discharge summary are associated with timely completion of follow-up. 16 In addition, artificial intelligence (AI)-generated summaries can improve the readability and accessibility of written instructions for patients, 17 , 18 thus providing the potential to improve compliance of follow-up.

In this context, advancements in machine learning (ML) and natural language processing (NLP) have shown potential for automated analysis of free-text radiology reports to detect incidental findings 10 , 19–22 and may assist with prompting adherence to best practices. Traditional ML and NLP models have long demonstrated their utility in the identification of incidental findings for ED patients. 10 , 19 , 21 However, these rule-based algorithms, often relying on key word stems, regular expressions, or extensive fine-tuning required for custom tasks, limit the generalizability of this technology to unseen examples across a broadened range of imaging modalities and across institutions. In contrast, large language models (LLMs) such as GPT-4 may offer an evolved approach benefitting from huge swaths of training data to understand human language and, in turn, the potential to outperform specialized, fine-tuned models with additional adaptability on domain-specific tasks. 23 Recently, GPT-4 performed well at identifying incidental adrenal nodules, pancreatic cysts, and vascular calcifications from abdominal CTs. 24 Building upon these applications for specific incidental pathologies and other narrow tasks, we sought to investigate GPT-4's performance on a broader scale.

In this quality improvement initiative, we describe our investigation using a HIPAA-compliant instance of GPT-4 to detect and classify AIFs in radiology reports from ED patients using best practice prompt engineering techniques. We also investigate the feasibility of using GPT-4 to assist clinicians in generating patient-facing instructions about these AIFs. This preliminary exploration serves as a foundational analysis to inform future deployment of an LLM-based tool into our live electronic health record environment. This offers the potential to devise timely reminders to ED clinicians about AIFs at critical transition points in the patient’s care (ie, discharge or handoff) and to draft patient-friendly instructions about AIFs that clinician can, after review, opt to incorporate into a patient’s written discharge instructions.

Classification of incidental findings

We developed a structured classification scheme to delineate and categorize incidental findings found within reports based on likelihood of requiring further action for follow-up testing, imaging, or evaluation after ED care. By this definition, we excluded the following:

Non-incidental findings directly related to the clinical indication for imaging.

Emergent findings requiring immediate attention during the patient’s ED visit.

Incidental findings described as chronic, stable, or unchanged.

Incidental findings were classified into 3 categories (see Figures 1 and 2 ):

Instruction prompt to identify and categorize non-emergent, actionable incidental findings (AIFs).

Sample radiology report with categorizations of incidental findings by actionability.

“Definitely Actionable” (DA): findings with explicit instructions from the radiologist for non-emergent follow-up, or any finding concerning for malignancy.

“Possibly Actionable—Clinical Correlation” (PA-CC): findings with radiologist recommendation for clinical correlation.

“Indeterminately Actionable” (IA) : findings not meeting DA or PA-CC criteria.

Derivation of radiology report datasets

We executed multiple structured queries to extract radiology reports of multiple imaging modalities (x-rays, computed tomography scans, and ultrasounds) from our health system’s electronic health record across 4 academic ED sites at NYU Langone Health.

To derive our validation set, we pulled 50 cases over a 1-month period (July to August 2021), half of which contained at least 1 DA or PA-CC finding. The dataset was searched to eliminate duplicate medical record numbers, manually labeled via expert review (see below), and used for the initial experimentation, development, and optimization of our prompt.

For our test dataset, we performed a second query over an extended period (July 2021 to August 2023) across the same locations and imaging modalities that used search terms of type (eg, node, nodule, cyst, lesion), measurement (eg, mm, cm), and instruction (eg, follow-up). We executed a third query without adding the requirement of the presence of the aforementioned search terms to obtain a baseline sample of cases over a 2-day period in November 2023, selecting dates that ensured no overlap in patient visits from prior data queries. After eliminating duplicate medical record numbers, 124 cases from the second query and 306 cases from the third query were randomly selected and combined to form our 430-case test dataset. To ensure a high case mix complexity to test GPT-4 performance while also more closely mirroring real-word incidence, we targeted a positivity rate of 20%-30% for DA or PA-CC findings in this enriched test set. Additionally, each case from our validation set was manually cross-referenced against the 430 cases in our test set to confirm there were no duplications, ensuring that cases from the test set against which our optimized prompt would be applied were unseen from the validation set.

All patient radiology reports were handled with strict confidentiality without de-identification to eliminate duplicate reports and allow for chart review of AIFs. This project met NYU Grossman School of Medicine’s Institutional Review Board criteria for quality improvement work not requiring IRB review.

Expert review/case labeling

Two board-certified emergency physician reviewers (Unblind: KW and GS, 13 and 6 years’ experience, respectively) independently applied the above classification scheme to cases for identification of DA and PA-CC findings due to their high likelihood of actionability. Labeling of IA findings was deferred, as individual reports could contain dozens of indeterminate or normal findings with unspecified clinical significance that were generally non-actionable. Inter-rater reliability of DA and PA-CC identification was assessed using Cohen’s Kappa for a sample of 30 independent cases randomly selected from the same derivation pool as the test cases, prior to randomized selection of the final test set. Agreement was substantial for AIF categorization of actionability, including the possibility of a lack of any AIF. Agreement of classification scheme ratings was 93.1% with K = 0.74, standard error = 0.13, P < .001. Subsequently, each of our 430 test cases were reviewed by 1 of the same 2 emergency physician reviewers. Cases with potentially uncertain categorizations by 1 of the 2 expert reviewers were flagged for further discussion and consensus review. A third rater (board-certified radiologist: SKK, 10 years’ experience) resolved any remaining disagreement.

Task specification

We prompted GPT-4 to perform 3 main functions (see Supplementary Appendix A ): (1) review the full text of each radiology report (including indication for imaging) to identify all non-emergent, AIFs and any associated recommendations; (2) categorize the findings by the classification scheme defined above and extract the associated radiologist recommendation, request for clinical correlation, or concern for malignancy associated with each finding; and (3) for DA findings, generate patient-centered instructions that summarize the radiologist’s recommendations for follow-up at a seventh-grade reading level.

Prompt optimization

We employed a HIPAA-compliant instance of GPT-4 v0613 via Microsoft Azure OpenAI’s API to conduct extensive experimentation on our 50-case validation set targeting iterative performance improvement. Multiple trials were performed with varying prompts, encompassing thousands of experimental runs on both targeted individual cases as well as across the dataset. Python scripting was used to perform frequent, automated sample runs, parse free-text GPT-4 responses, and format outputs into discrete data types for clinician review.

Analysis and refinements of the prompt were anchored in prompt engineering best practices, including but not limited to task decomposition, sequential tasking, error checking, chain of thought reasoning, and example-based learning techniques. Parameter settings were maintained throughout prompt iterations as follows: Max Response—1000; Temperature—0.0, Top P—0.95, Frequency Penalty—0; Presence Penalty—0.

Evaluation of GPT-4 outputs

Each DA or PA-CC classification by GPT-4 was graded for presence of a True Positive (TP), False Positive (FP), or False Negative (FN) instance against the manual labelings previously described. In the primary analysis, outputs were graded for accuracy of identifying DA or PA-CC findings, encompassing a broad range of findings with high likelihood of actionability. In the secondary analysis, outputs were graded for accuracy of classifying DA findings alone, limiting analysis to findings with only the highest likelihood of actionability. Consequently, a True Negative (TN) event was defined by the absence of any DA or PA-CC findings per report for the primary outcome, or the absence of any DA findings for the secondary outcome. Multiple AIFs within the same report were counted as separate instances, unless findings were inherently related (eg, 2 renal cysts). FP instances were evaluated for frequency and the type of misclassification error (eg, non-incidental, emergent, indeterminately actionable, or chronic, stable, or unchanged). The numbers of TPs, FPs, and FNs were then used to calculate recall, precision, and F-1 for the primary (DA or PA-CC) and secondary (DA only) analyses, respectively.

To further evaluate performance, all GPT-4 outputs were independently reviewed by 4 practicing ED physicians for the presence of hallucinations of any finding or inferences of any recommendation without explicit foundation. Subsequently, physician reviewers rated the appropriateness and usability of GPT-4-generated patient-facing instructions for all TP DA findings using a 5-point Likert scale (1: Needs COMPLETE revision; 2: Needs MAJOR revision; 3: Needs MODERATE revision; 4: Needs MINOR revision; 5: Needs NO revision). Ratings were reported by mean (standard deviation) and frequency of each response category.

Chart review of cases with DA findings

To evaluate the potential for improvement that GPT-4 could present, we manually reviewed the electronic health records of patients with DA findings present to determine if AIFs were communicated with the patient. Communication was defined as documentation in the chart of having discussed the finding with the patient or inclusion of the finding or recommendations in the patient’s discharge instructions or handoff to an admitting inpatient team.

Validation set

Our 50-case validation set contained 17 DA and 16 PA-CC findings. 50.0% (25/50) of cases were positive for 1 or more DA or PA-CC findings, while 28.0% (14/50) of cases were positive for 1 or more DA findings. Our prompt was iteratively optimized until it demonstrated 100% recall, 89% precision, and 94% F-1 for DA or PA-CC findings and 100% recall, 65% precision, and 79% F-1 for DA findings alone.

Manual review of our 430-case test set revealed 104 DA and 42 PA-CC findings across 108 unique reports. 25.1% of reports were positive for 1 or more DA or PACC findings, achieving our target positivity between 20% and 30% for our enriched dataset. 18.4% (79/430) of cases were positive for DA findings, with more than 1 per report in 19 cases. Our test dataset included a diverse mix of imaging modalities (42.6% x-rays, 47.0% computed tomography scans, and 10.5% ultrasounds). Each call using our prompt averaged 2735 prompt tokens for input and 227 completion tokens for output, comprising an average of 9.6 cents per report. Our cumulative costs accounting for all experiments conducted amounted to $438.00 (input) and $141.60 (output).

Primary outcome (classification of DA or PA-CC findings)

GPT-4 identification of DA or PA-CC findings performed with 99.3% [95% CI, 95.6%-100.0%] recall, 73.6% [95% CI, 66.8%-79.5%] precision, and 84.5% F-1 (see Table 1 ).

Contingency table of non-emergent AIFs, on a per-finding basis.

Only those IA findings or non-AIFs (ie, non-incidental, emergent, and/or chronic, stable, or unchanged findings) that contributed to FP or FN instances are displayed here.

Fifty-two FP categorizations (see Table 2 ) were observed over 42 cases (8 cases included 2-3 findings and accounted for 18 occurrences), accounting for a FP rate of 15.3%. FP findings were comprised of 51.9% (27/52) non-incidental findings, 38.5% (23/52) IA findings that were misclassified by GPT-4 as either PA-CC or DA, 13.5%% (7/52) emergent findings, and 1.9% (1/52) finding that were chronic, stable, or unchanged (in this case chronic). (Five non-AIFs were both non-incidental and emergent and a sixth was both non-incidental and chronic.) Among the 23 IA findings that were misclassified by GPT-4 as PA-CC or DA, 9 were linked to GPT-4's inferences beyond what the radiologist stated (see Inferences and Hallucinations section).

Sample false positives representing common types of error with relevant text additionally underlined.

The most common types of GPT-4 error were classifying as incidental when the finding was related to the indication, or when a finding did not have an accompanying recommendation in the report (IA) but GPT-4 categorized the finding as needing clinical follow-up (DA) or clinical correlation (PA-CC).

Only 1 FN categorization was observed, constituting a 0.7% FN rate (see Table 3 ).

False Negatives, with relevant text additionally underlined.

Secondary outcome (classification of DA findings only)

GPT-4 identification of DA findings only performed at 95.2% [95% CI, 88.6%-98.2%] recall, 77.3% [95% CI, 68.9%-84.1%] precision, and 85.3% F-1 (see Table 1 ).

Twenty-nine FP DA categorizations (see Table 2 ) were observed over 25 cases (3 cases accounted for 7 instances), accounting for a FP rate of 8.1%. FP findings were comprised of 41.4% (12/29) non-incidental, 31.0% (9/29) PA-CC, 20.7% (6/29) IA, 17.2% (5/29) emergent, and 3.4% (1/29) chronic, stable, or unchanged findings (in this instance, chronic). (Three findings were both non-incidental and emergent and a fourth was both non-incidental and chronic.)

Five FN DA categorizations were observed (each from different cases), constituting a 1.5% FN rate. In addition to the 1 FN DA finding previously identified above that was misclassified as IA, the remaining 4 FN results were DA findings that the model misclassified as PA-CC (see Table 3 ).

Inferences and hallucinations

Among our 430 test cases, there were no observed hallucinations of any incidental findings that did not exist in the radiology report. However, in the recommendations for follow-up, 2.8% (12/430) of outputs included text that was inferred by GPT-4 rather than directly summarized from the radiologist’s recommendations (see Table 4 ).

GPT-4 inferences from radiologist recommendations, with relevant text additionally underlined.

Evaluation of performance for patient instructions

The mean (SD) Likert score for GPT-generated patient instructions for TP DA findings was 4.5 (0.65) on a 5-point Likert scale, indicating that reviewers perceived most outputs would require only minor (if any) revisions prior to inclusion into the patient’s discharge instructions. The most frequent rating was a 5 (no revision required), for 55.5% of reports; the range of scores was 3-5 with only a small minority (8.4%) indicated as needing moderate revision ( Figure 3 ).

Distribution of likert ratings for true positive DA non-emergent incidental findings.

Seventy-nine out of 430 cases contained 1 or more DA findings. Per chart review, 55.7% (44 of 79) had documentation of communication with the patient about the finding at discharge or with the inpatient team upon admission.

Performance

GPT-4 performance was noteworthy, especially given the model’s lack of clinical pretraining and complexity of the task. Nearly all DA or PA-CC findings were successfully screened by GPT-4. In comparison, only about half of DA findings were documented as communicated with the patient upon discharge or with the inpatient team upon admission, highlighting a void that this LLM tool could successfully fill. Outputs did not always reliably distinguish non-emergent DA/PA-CC findings from those that were non-incidental, IA, or emergent. However, the clinical consequences of these FPs are expected to be minor for patients who would already be receiving active treatment for emergent or non-incidental findings.

That the majority of FP misattributions were non-incidental highlights the model’s limitations in fully evaluating a radiologic finding in the context of the study’s clinical indication, which requires deep relational knowledge of medicine and nuanced semantic understanding. Additionally, we observed the model occasionally misattribute a radiologist’s follow-up recommendation to multiple upstream findings in preceding sentences, rather than to the specific antecedent finding to which the recommendation was referring, highlighting challenges in contextual interpretation.

Notably, we observed no cases where a FP finding was “hallucinated” de novo without an associated referenceable finding in the radiology report. However, GPT-4 on occasion inferred actionability of the finding that was not explicitly stated, asserting (reasonably so in most instances) that a finding was benign, or that clinical correlation or repeat imaging was warranted. This highlights the importance of active clinician review and oversight of GPT-4’s outputs.

Clinicians felt that the majority of GPT-4 outputs for TP DA findings were appropriate. However, free-text feedback indicated specific areas for improvement. For example, GPT-4 tended to avoid medical terminology when instructed to simplify the instructions to a seventh-grade reading level, so words like “nodule” could become replaced with ambiguous terms such as “lump.” GPT-4 sometimes paraphrased words that were actual medical terminology, turning lung “opacities” into “shadows,” or “atypical pneumonia” into an “unusual type of pneumonia.” GPT-4 sometimes added overly descriptive language, ranging from unnecessary (eg, describing a thyroid as “shaped like a butterfly, smaller in the middle with 2 wide wings that extend around the side of your throat”) to potentially upsetting (eg, defining a uterus as “the organ where a baby grows during pregnancy,” which would be inappropriate for a patient suffering from possible miscarriage or infertility). GPT-4 also tended to be overly confident in stating what a finding most likely represented but was not yet confirmed (eg, cyst, benign mass), which requires clinician oversight.

Prompt experimentation

Our process of prompt optimization though iterative experimentation was critical for enhancing model performance to complete the designated task utilizing a zero shot approach. Consistent with best practices, our strategy involved observing AI model outputs to varying prompt adjustments, identifying errors or misattributions, and revising the prompt accordingly over time.

We found that task decomposition, the process of dividing a complex task into many smaller, specific tasks within the same prompt, improved our model’s performance significantly. Additionally, sequential tasking proved instrumental in guiding the model’s interpretative processes. We then coupled this with an internal review or error check to revise initial classifications prior to a final output, which further improved GPT-4 accuracy.

Notably, we encountered challenges related to the formatting of GPT-4’s output in early testing. Initial attempts to direct the model’s output towards a JSON structure included specific example-based instruction. We observed that detailed clinical examples often biased clinical classification with resultant degradation of performance, which we attributed to the model’s tendency to overfit to the detailed classifications and format in the provided example. By minimizing detail within examples but maintaining general format guidelines, we achieved 100% adherence to the desired JSON format with minimal degradation.

Limitations

We occasionally employed multiple runs testing on the same prompt to gauge the stability of GPT-4’s responses, an inherent challenge for generative AI models. Anecdotally, this approach revealed a degree of sensitivity to the phrasing of instructions and highlighted occasional instability in outputs. Integrating an internal self-check mechanism within our prompt and adjusting model parameters (eg, temperature) were observed to mitigate some of this inconsistency. However, extensive high-volume testing for further stability assessment was not feasible as a limitation in our methodology. Our work also highlights the inherent challenge in the interpretation of “actionability.” While the distinct categories of DA, PA-CC and IA were clinically reasoned and useful for our purposes here, this inherently introduces some degree of subjectivity into ground truth classification. Our criteria for “actionability” were anchored in explicit radiologist recommendations for follow-up, clinical correlation, or concerns of malignancy, which, in real-life scenarios, may not encompass all clinically relevant AIFs. The PA-CC classification proved to be complex due to ambiguous phrasings suggestive of correlation even to human readers. Differentiating “non-incidental” findings presented its own set of complexities, reflecting the nuanced and subjective nature of clinical decision-making. This underscores the difficulty in translating the subtleties of clinical assessments into a rigid AI-driven categorization framework.

In addition, the derivation of our datasets may harbor unanticipated sources of bias. The use of a key word search to yield a high capture of AIFs may have limited the types of AIFs rendered. To ensure mutual exclusivity of case visits, the baseline cases within our test set were derived from a brief, convenience sample of time (November 2023) several months after the timeframe used for the radiology reports derived via key word search (July 2021 to August 2023). Still, from July 2021 to November 2023, there were no significant differences in use of structured reporting templates by the radiologists that would be anticipated to affect GPT-4's task at hand.

Future directions

Anticipated future advancements in state-of-the-art general LLMs and the evolution of models specialized to the medical domain can also be expected to yield improved performance. At our institution, we plan to silently deploy this model in a controlled framework that enables integration of AIF classification into the provider workflow and monitoring of real-world performance. For findings with high likelihood of actionability (DA or PA-CC), we envision creating clinician-facing alerts at the point of discharge or inpatient handoff as a just-in-time reminder at critical transition points in a patient’s ED visit. For findings with the highest likelihood of actionability (DA only), we envision presenting AI-generated instructions for the clinician to review, edit as needed, and incorporate into the patient’s discharge instructions if desired. It should be noted, however, that the same pressures that may predispose clinicians to overlook AIFs can also expose clinicians to a different but equally important risk of automation bias leading to over-reliance on AI outputs without thorough scrutiny. Deployment of this technology necessitates accompanying safeguards to mitigate such biases and ensure that AI tools are used as aids, not substitutes, for clinical expertise. As AI technology advances, reevaluation of these risks will need to be weighed against the benefits of enhancing clinicians’ ability to deliver informed, personalized, and patient-centered care.

As a novel investigation into the evaluation of GPT-4 performance in a highly relevant medical context, our work demonstrates a general model’s reasonable proficiency in the highly specialized and complex task of discerning and translating radiologic AIFs. We also highlight the need for rigorous scrutiny of GPT use in medicine, reinforcing the notion that while AI can provide significant assistance, it cannot supplant the nuanced judgement of healthcare experts in current state. Through rigorous testing of AI methodologies in an academic setting with real patient data, we endeavor to advance future applications that augment patient care while maintaining the highest standards of patient safety. Our study aims lay the groundwork for future investigations into the downstream applications of this technology to assist with provider-facing clinical decision support and patient-facing health communication.

We thank Dr Catherine Jamin, Dr Soterios Gyftopoulos, Dr Sarah Spiegel, Dr Leland Chan, and Duo (Walter) Wang for their support. We also thank the NYU Langone Health Medical Center IT for supporting exploratory and API access to the secure, HIPAA-compliant instance of GPT-4 that enabled this work.

Kar-mun C. Woo and Gregory W. Simon developed the instruction prompt, manually labeled cases, tested individual radiology reports within GPT-4, and drafted the manuscript. Stella K. Kang provided expert guidance on project direction, aims, labeling schema, dataset design, and arbitrated on manual labeling of cases. Chloé S. Pariente and Vincent J. Major developed and executed the Python code to enable automated high volume GPT testing and format outputs for review. Yindalon Aphinyanaphongs, Jonathan S. Austrian, and Vincent J. Major served as expert AI project mentors, guided project direction, and provided institutional executive leadership. Chloé S. Pariente, Nicholas Genes, Vincent J. Major, Yindalon Aphinyanaphongs, Jonathan S. Austrian, and Edwin G. Pineda assisted with prompt and output optimization. Nicholas Genes, Chloé S. Pariente, and Gregory W. Simon performed data queries to extract ED radiology reports. Olumide Akindutire and Jacob A. Goldenring performed chart reviews of cases with DA findings. Jacob A. Goldenring, Olumide Akindutire, Gregory W. Simon, and Kar-mun C. Woo performed Likert ratings of GPT outputs and reviewed outputs for hallucinations. Jung G. Kim assisted with statistical calculations. All authors reviewed, revised, and approved the submitted version.

Supplementary material is available at Journal of the American Medical Informatics Association online.

This work was supported by the NYU Langone Health, MCIT. Y.A. and V.J.M. were partially supported by NIH 3UL1TR001445-05 and Y.A. was supported by the National Science Foundation award #1928614 and #2129076.

None declared.

Available upon reasonable request.

Berland LL. Overview of white papers of the ACR incidental findings committee II on adnexal, vascular, splenic, nodal, gallbladder, and biliary findings . J Am Coll Radiol . 2013 ; 10 ( 9 ): 672 - 674 .

Google Scholar

Evans CS , Arthur R , Kane M , et al. Incidental radiology findings on CT studies in the ED: a systematic review and meta-analysis . Ann Emerg Med . 2022 ; 80 ( 3 ): 243 - 256 .

Lumbreras B , Donat L , Hernández-Aguado I. Incidental findings in imaging diagnostic tests a systematic review . Br J Radiol . 2010 ; 83 ( 988 ): 276 - 289 .

Moore CL , Baskin A , Chang AM , et al. White paper: best practices in the communication and management of actionable incidental findings in emergency department imaging . J Am Coll Radiol . 2023 ; 20 ( 4 ): 422 - 430 .

O'Sullivan JW , Muntinga T , Grigg S , Ioannidis JPA. Prevalence and outcomes of incidental imaging findings: umbrella review . BMJ . 2018 ; 361 : k2387 .

Vernooij MW , Ikram MA , Tanghe HL , et al. Incidental findings on brain MRI in the general population . N Engl J Med . 2007 ; 357 ( 18 ): 1821 - 1828 .

Smith-Bindman R , Kwan ML , Marlow EC , et al. Trends in use of medical imaging in US health care systems and in Ontario, Canada, 2000-2016 . JAMA . 2019 ; 322 ( 9 ): 843 - 856 .

Munk M-D , Peitzman AB , Hostler DP , Wolfson AB. Frequency and follow-up of incidental findings on trauma computed tomography scans: experience at a level one trauma center . J Emerg Med . 2010 ; 38 ( 3 ): 346 - 350 .

Thompson RJ , Wojcik SM , Grant WD , Ko PY. Incidental findings on CT scans in the emergency department . Emerg Med Int . 2011 ; 2011 : 624847 .

Dutta S , Long WJ , Brown DFM , Reisner AT. Automated detection using NLP and radiologists' recommendations for additional imaging of incidental findings . Ann Emerg Med . 2013 ; 62 ( 2 ): 162 - 169 .

Berge P , Darsonval A , Nedelcu C , Paisant A , Aubé C. Incidental findings on emergency CT scans: predictive factors and medico-economic impact . Eur J Radiol . 2020 ; 129 : 109072 .

Liao GJ , Liao JM , Lalevic D , Zafar HM , Cook TS. Location, location, location: the association between imaging setting and follow-up of findings of indeterminate malignant potential . J Am Coll Radiol . 2019 ; 16 ( 6 ): 781 - 787 .

Barrett TW , Garland NM , Freeman CL , et al. Catching those who fall through the cracks: integrating a follow-up process for emergency department patients with incidental radiologic findings . Ann Emerg Med . 2022 ; 80 ( 3 ): 235 - 242 .

Hansra SS , Loehfelm TW , Wilson M , Corwin MT. Factors affecting adherence to recommendations for additional imaging of incidental findings in radiology reports . J Am Coll Radiol . 2021 ; 18 ( 2 ): 233 - 239 .

Moore CL , Kadom N , Seidenwurm D , et al. Incidental findings: a survey of radiologists and emergency physicians . J Am Coll Radiol . 2021 ; 18 ( 6 ): 853 - 856 .

Kwan JL , Yermak D , Markell L , Paul NS , Shojania KG , Cram P. Follow up of incidental high-risk pulmonary nodules on computed tomography pulmonary angiography at care transitions . J Hosp Med . 2019 ; 14 ( 6 ): 349 - 352 .

Lee T-C , Staller K , Botoman V , Pathipati MP , Varma S , Kuo B. ChatGPT answers common patient questions about colonoscopy . Gastroenterology . 2023 ; 165 ( 2 ): 509.e7 - 511.e7 . https://doi.org/10.1053/j.gastro.2023.04.033 .

Kirchner GJ , Kim RY , Weddle JB , Bible JE. Can artificial intelligence improve the readability of patient education materials? Clin Orthop Relat Res . 2023 ; 481 ( 11 ): 2260 - 2267 .

Evans CS , Dorris HD , Kane MT , et al. A NLP and ML approach to ID incidental radiology findings in trauma patients discharged from the ED . Ann Emerg Med . 2023 ; 81 ( 3 ): 262 - 269 .

Fu S , Leung LY , Wang Y , et al. Natural language processing for the identification of silent brain infarcts from neuroimaging reports . JMIR Med Inform . 2019 ; 7 ( 2 ): e12109 .

Kang SK , Garry K , Chung R , et al. Natural language processing for identification of incidental pulmonary nodules in radiology reports . J Am Coll Radiol . 2019 ; 16 ( 11 ): 1587 - 1594 .

Pham AD , Névéol A , Lavergne T , et al. Natural language processing of radiology reports for the detection of thromboembolic diseases and clinically relevant incidental findings . BMC Bioinformatics . 2014 ; 15 ( 1 ): 266 .

Nori H , Lee YT , Zhang S , et al. 2023 . Can generalist foundation models outcompete special-purpose tuning? A case study in medicine. arXiv , arXiv:2311.16452v1 [cs.CL].

Bhayana R , Elias G , Datta D , Bhambra N , Deng Y , Krishna S. Use of GPT-4 with single-shot learning to identify incidental findings in radiology reports . AJR Am J Roentgenol . 2024 ; 222 ( 3 ): e233065 .

Supplementary data

Email alerts, citing articles via.

Recommend to your Library

Affiliations

Online ISSN 1527-974X
Copyright © 2024 American Medical Informatics Association
About Oxford Academic
Publish journals with us
University press partners
What we publish
New features
Open access
Institutional account management
Rights and permissions
Get help with access
Accessibility
Advertising
Media enquiries
Oxford University Press
Oxford Languages
University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

Copyright © 2024 Oxford University Press
Cookie settings
Cookie policy
Privacy policy
Legal notice

This Feature Is Available To Subscribers Only

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Log in using your username and password

Search More Search for this keyword Advanced search
Latest content
Current issue
BMJ Journals More You are viewing from: Google Indexer

http://orcid.org/0000-0003-2891-0113 Christian Morgner 1 ,
Karen Harrison Dening 2 ,
Tom Dening 3 ,
Barry Gibson 4
1 Management School , The University of Sheffield , Sheffield , UK
2 Dementia UK , London , UK
3 Mental Health & Clinical Neurosciences, School of Medicine , University of Nottingham , Nottingham , UK
4 School of Clinical Dentistry & The Healthy Lifespan Institute , The University of Sheffield , Sheffield , UK
Correspondence to Dr Christian Morgner, The University of Sheffield, Sheffield, UK; c.morgner{at}sheffield.ac.uk

The theoretical approach presented in this paper describes a novel experimental-theoretical methodology to conceptualise pain in people with dementia. Existing procedures for assessment of pain rely on subjective self-report using pain questionnaires and rating scales that have proven to be highly problematic where a person has dementia. Consequently, pain in people with dementia can be undetected and/or undertreated. To address that, we have developed an alternative experimental approach that builds on theoretical and methodological precedents from the arts, humanities and social sciences, for instance, visual thinking strategies, creative thinking or two-step flow of communication. Based on this approach, we designed an experimental workshop setting to ingrate these methodologies to explore pain and its expression in people with dementia. This had led to a new definition of pain as an interruption of the socially mediated process of bodily meaning-making. Furthermore, our experimental methodology could equally well be applied as a training method, where professional staff can intervene into existing implicit meanings and understandings of medical issues. These results emphasise that the future of pain research needs to consider the relational aspects of pain more seriously.

pain management
Medical humanities

Data availability statement

No data are available. The consent obtained from the participants covered only access to and data sharing within the research team. In accordance with data protection policies, it did not authorise the sharing of data with third parties.

https://doi.org/10.1136/medhum-2023-012718

Statistics from Altmetric.com

Request permissions.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction: pain recognition in people with dementia

This paper describes a novel experimental-theoretical methodology to provide a better conceptualisation about pain in people with dementia. There are over 50 million people in the world with dementia with the numbers expected almost to double in the next 20 years ( Alzheimer’s Disease International 2021 ). Understanding and treating pain is a common problem among this large and growing population.

Physical and mental pain are among the most common symptoms experienced by people with dementia ( Lautenbacher and Gibson 2016 ; Scherder et al 2009 ). The majority of people living with dementia have at least three comorbid conditions ( Scrutton and Brancati 2016 ; Timmons et al 2016 ), which commonly include chronic obstructive pulmonary disease, cardiovascular disease, hypertension, diabetes, sensory impairment and musculoskeletal disorders ( Browne et al 2017 ; Bunn et al 2016 ). Many of these comorbid conditions can induce pain, but no existing medical procedure, for example, measuring electric signals in the nervous system, muscle contraction or electrical heatmaps of the brain, can objectively diagnose that pain. Detecting and responding to pain experienced by people with dementia is an important clinical issue, as pain may cause worsening of the person’s condition ( Scherder et al 2009 ), for example, increased cognitive impairment or changes in behaviour. As already stressed by Loeser (1991) , an important issue is the paucity of medical theory in conceptualising pain, which is often ineffective, but may be even one of the causes of this rapidly growing problem of pain diagnoses and management.

Clinical definitions of pain

An unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage. International Association for the Study of Pain (cited in Merskey 1979 ). An unpleasant emotional experience usually initiated by noxious stimulus and transmitted over a specialised neural network to the central nervous system where it is interpreted as such. Bennett (1984) . Pain is an unpleasant sensation localised to a part of the body. ( Rathmell and Fields 2013 ). Pain is a mutually recognisable somatic experience that reflects a person’s apprehension of threat to their bodily or existential integrity. Cohen et al (2018) . An unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage. Revised Definition of International Association for the Study of Pain ( Raja et al 2020 ).

These common clinical definitions of pain tend to conceptualise pain mainly as bodily experience, and variously describe it, for example, as a sensation or a somatic experience that is negative in nature. 1 These definitions of pain restrict the meaning of pain to neurophysiological activities, for example, as a somatic experience and neglect to consider the meaning of pain for a person, which requires a more sociological approach to pain ( Bendelow and Williams 1995 ). 2 This is perhaps unsurprising, when we shift the focus to their conceptualisation. Medical theories are typically challenged for their practical application ( Alderson 1998 ; Thagard 2005 ), but very little is known about the practice of theorisation in medicine.

The few studies that exist in this area suggest that medical definitions or theories emerge from ‘intuitive heuristics performed by individual professionals’ ( Malterud 1993 , 81); ‘reviews of reviews’ ( Smith et al 2011 ; see also Gates et al 2022 ) or summarising and reorganising existing text-book definitions ( Hucklenbroich 2017 ). Based on these existing practices of theorisation in medicine, we can understand why the above-mentioned definitions are very consistent, and very similar to each other. This is because they more or less derive from tautological processes that strongly build on each other. As stated by Corns (2020) , there is a need for complexity and to reject an orthodoxy of simplistic conceptualisations. For instance, Williams and Craig (2016) emphasised that only such a more complex definition can account for a wider range of sensory, emotional, cognitive and social factors that influence the experience of pain. The aim of this paper with be double-fold: (1) to intervene into this implicit process of theorisation in medicine to develop a new approach that will assist medical professionals in the creation of new medical concepts and (2) to develop a new definition of pain, which is able to integrate its communicative and social dimensions, that is, being able to talk about it, and the importance of where and when the pain is being experienced.

Diagnosing pain: from questionnaires to interaction

Existing procedures of assessing pain that build on these clinical definitions rely on pain questionnaires and rating scales, which employ large and disparate vocabularies to describe and categorise pain, ranging from 100 to 200 words in various languages ( Breivik et al 2008 ; De Silva et al 2020 ; Mathieson et al 2015 ). However, a growing body of research suggests that these tools are highly problematic, not just for people with dementia ( Herr et al 2017 ). Pain ratings give an illusion suggesting that pain can be assessed in an objective way, which reflects the above-mentioned clinical definitions. However, this ignores that pain ratings are part of a sequence of complex sociological interactions to establish the significance of the pain, which include a variety of subjective interpreters that include the patient, carer, nurse, pharmacists or general practitioner (GP) ( Heath 1989 ; Jenkins 2015 ). Consequently, pain scale ratings by the same person have proved inconsistent and unreliable from 1 day to the next. Relying on complex vocabularies that are not part of people’s active language competence is also a problem, as there is significant ambiguity about the meaning of words in the context of a given assessment ( Herr 2011 ; Schofield 2018 ). Ehlich (1985, 177) therefore concluded in his research on the language of pain: ‘Speaking about pain is thus one of the most difficult forms of linguistic activities’.

Consequently, pain in people with dementia can be undetected and/or undertreated ( Achterberg et al 2019 ; Dening 2020 ; Liao et al 2023 ). This may lead to undertreatment with analgesics or inappropriate treatment, for example, antipsychotics for agitated behaviour ( Brunkert et al 2019 ). In some older, but still relevant research, relatives and carers were asked about how pain was managed in the person they cared for. Many reported that the person experienced significant pain but that its recognition, and thus control, was often poor ( Horgas et al 2003 ; McCarthy et al 1997 ). Other research has found differences in the experience of trauma, such as hip fracture, where people with dementia were prescribed only a third as much pain relief as those people without dementia ( Armour and Murphy-Jones 2016 ; Morrison and Siu 2000 ).

As mentioned above, our more sociological approach to pain argues that pain diagnosis depends on interactive co-construction and disparate complex vocabularies, particular challenges arise in relation to people with dementia ( Dening 2020 ). More specifically, dementia leads to decline in linguistic skills (eg, expressing, interpreting and understanding meanings in oral and written form), and the ability to engage interactively in negotiating and identifying pain is severely challenged. As people with dementia have difficulty in communicating their pain, care staff may fail to understand pain-related behaviours and thus misinterpret pain signals ( Oosterman et al 2014 ). However, little has been done to address the challenges of understanding pain in people with dementia. As a step towards bridging this gap, the experimental approach proposed here builds on theoretical and methodological precedents from the arts, humanities and social sciences to unravel the complex processes of expressing and interpreting meaning.

Scholars in several disciplines have noted that meaning-making is a complex, creative, collaborative and selective process beyond the mere transmission of information from one person to another ( MacKay 1969 ; Alexander 2003 ; Morgner 2022 ). For instance, an everyday conversation between two close friends differs significantly from more formal interactions (eg, in a courtroom or between doctor and patient) in terms of both the challenges and freedoms of managing and negotiating meaning. Communication about pain is complex, as it is a sensitive topic with a limited lexicon, and is further complicated in people with dementia as communication has to rely more on non-verbal modes of expression. Consequently, our research was designed to address the limitations of current clinical definitions by developing a conceptual methodology that could lead to new ways to gain a better understanding of pain in people with dementia.

In order to overcome the limitations of current clinical definitions of pain and to devise new ways to assess pain, this research built on recent debates about the importance of involving people with dementia in research more generally, and sought to devise an accessible methodology that overcomes some of the existing issues identified in current research ( Sheth 2019 ; Wilkinson and Weitkamp 2020 ).

In recent research, several studies ( Caron-Flinterman et al 2005 ; Duffett 2017 ; Popay and Williams 1996 ) have noted obstacles to participation created by the uneven distribution of knowledge. Some forms of knowledge are valued as having greater relevance and so require less justification, in turn rendering other forms of knowledge more peripheral. These asymmetries make it easier for those in possession of recognised knowledge to be included in the debate while effectively silencing others. Other research highlights the role of power in facilitating, constraining or preventing social action ( Boardman 2014 ; Kelleher et al 1994 ; Kerr et al 2007 ). In medical and other institutional contexts, authority is typically perceived as a legitimate form of power, and those in authority can enable or prevent the participation of others. Claims of authority are difficult to challenge because they reflect institutional norms, so creating a conversational structure that favours those in authority.

Existing research also highlights the need to take account of variations in verbal communication and debating skills ( Jordan et al 2010 ; Elberse et al 2011 ). In particular, while professionals are more likely to have experience of public speaking or discussing issues with other professionals, lay people and people with dementia may feel less confident in voicing their opinions. A second communication issue mentioned in the literature is that dementia may hinder the ability to follow a fast-paced or complex debate that uses academic or professional jargon requiring a higher level of health literacy. This again relates to the role of power in designating the roles of speaker and listener, as varying conversation skills can determine participants’ ability to participate or share their meaning. In enabling innovative solutions to pain diagnosis in people with dementia, it is also important not to limit the discussion by predetermining the subject matter. Equally, the literature confirms the importance of avoiding unduly open discussion that makes it more difficult to identify and select appropriate innovations ( Morrison and Dearden 2013 ).

Finally, active and creative participation depends on a welcoming, stimulating and non-stigmatising atmosphere. When we refer to the term creative in the context of this workshop, we do not refer to creative activities, such as drawing or singing, but to creative thinking. With creative thinking, we refer to the ability to consider something in a new way, encompassing the generation of new ideas or concepts or finding fresh associations between existing ideas or concepts. This involves divergent thinking, which seeks multiple solutions to a problem rather than focusing on a single answer. This kind of thinking is characterised by its flexibility, originality and the willingness to explore unconventional solutions ( Mumford et al 2013 ). There is evidence that individuals will avoid participation if labelled as people with dementia ( Bond and Corner 2001 ; Garand et al 2009 ; Manafo et al 2018 ), as they may feel that this reduces their social standing to this single identifier. Based on this review of the literature, we devised an experimental-theoretical workshop that combined methodologies from performing arts, arts education and communication studies to enable a meaningful contribution by people with dementia.

Methodological approach

An experimental-theoretical workshop.

To plan the conduct of the workshop and manage the issues of equality, power and knowledge, we devised a conversational sociological interactive framework for interpreting the meaning of pain. The central idea of this interactive meaning-making was broadly envisaged in the form of a workshop, with ‘workshop’ taken from performing arts contexts. In this context, a workshop format combines focused group discussions with an openness to improvisation and testing of new ideas as a research method to explore new understandings and meanings before formalising any outcomes ( Tarr et al 2018 ). We aimed to avoid limiting conceptual constructions to a predefined clinical context and also reflected on the sociological importance of meaning-making with respect to pain, in opposition to an isolationist arm-chair construction. Additionally, this workshop discussion format makes no distinction between those who enable actions and those who respond. In other words, we did not group participants into ‘authorities’ (academics, professionals) and others, as might happen in a traditional lecture hall. On the same basis, authority to speak was not predelegated or unilateral; following Goffman (1981) , the traditional roles of (authorised) speaker and listener were disrupted enabling all to speak or to listen as they wished, and to improvise and think beyond any authorised roles. To that end, participants were arranged in a circle, with no overall authority, mirroring the collaborative communication models of deliberative democracy and so-called round-table talks ( Perkins 2003 ; Reykowski 2020 ).

While this set-up avoids any sense of hierarchy, it does not automatically ensure that workshop participation and contribution will be balanced. In particular, this face-to-face discussion format can impose significant pressure on participants because it becomes virtually impossible not to communicate ( Beavin et al 1967 ). Consequently, participants feel obliged to produce ideas quickly, and there is a risk that they may detach from the discussion to reflect and prepare their comments. As this undermines listening, contributions may display an erratic quality rather than building on each other ( Martin et al 2012 ; Muijs and Reynolds 2017 ). Clearly, this erratic approach must be avoided in order to deliver tangible results in terms of learning and innovation. Because detached opinions also tend to be presented quite rapidly ( Hawkins 1985 ; Savoy and Yunyongying 2013 ), this poses particular challenges for anyone who is unused to fast-paced discussion, including people with dementia. To mitigate these issues and to enable all participants to contribute in a meaningful way, we incorporated two further elements: paraphrasing as used in visual thinking strategies (VTS) and a two-step communication flow .

Paraphrasing

Paraphrasing forms part of the VTS approach developed at the New York Museum of Modern Art in the 1990s ( Housen 2001 ). Developed in an arts education context, this format was designed to improve visitors’ learning experience during guided tours of the museum. The approach was informed by evidence that, when discussing works of art, participants were unable to listen to or follow others’ comments because of the complexity of formulating their own response when the subsequent communication slot opened. As a result, visitors were unable to collaborate or complement each other to deepen their interpretation of the artworks. To overcome this issue, VTS introduced the key tool of advanced paraphrasing. Paraphrasing is defined as neutral or factual rephrasing by the facilitator of what the previous speaker has just said. The facilitator acts as a mirror that buys time, so that the participants can gain clarity on their own thinking. This crucial intervention slows communication and eases any anxiety about lack of preparation by repeating what others have to say in summary form before inviting subsequent statements. In this way, statements can build on what has already been said, facilitating more complex communication through open exploration of different ideas. By implication, the role of the ‘authorised’ speaker is also disrupted, as those assigned to provide answers are equally enabled to listen.

Two-step communication flow

While VTS can in principle deal with the challenges around pace and meaningful contribution, those with more experience of discussion nevertheless remain in a privileged position in terms of attributes such as eloquence or confidence. To build the confidence of less experienced participants, our experimental-theoretical methodology also incorporated two-step communication flow as first proposed by Katz and Lazarsfeld (1955) in the context of mass communication studies. They wanted to understand how arguments presented in the media are received by the audience when relayed through intermediaries. They discovered that intermediaries play a risk-taking role as pioneers, showcasing different models before formulating an opinion. This has a shielding effect, as their ideas can be adopted and reused by others to enhance their communication skills. However, it should also be noted that providing an example through the first group discussion needs to be carefully managed, so that it does not limit the creative thinking. We emphasised the experimental nature of this discussion to account for this risk.

On that basis, the workshop was divided into two distinct phases. The first phase was confined to speakers with more experience of debating as an inner circle surrounded by the others in an arrangement modelled on televised political debates ( Livingstone and Lunt 1994 ). Those in the outer circle were informed that, in a second step, the inner round table debate would be extended to everyone, using the same paraphrasing approach. In this way, the outer circle was able to gain a sense of the content, as well as how such issues are debated and how arguments are presented. That way all participants were equipped with a number of strategies to experiment with for their own purposes.

In summary, participants were helped to overcome any obstacles to exploring new ways of understanding pain in people with dementia by means of a novel experimental approach. This combined arts-based research methods like workshops and VTS with theoretical inputs from the field of communication studies, including round-table talks and two-step communication flow and resulted in a half-day workshop for a diverse group of stakeholders to explore the understanding of pain in dementia.

Specifically, the study explored two main research questions:

What new understandings about pain in dementia emerged from the workshop?

Is this an appropriate, accessible and effective methodology for exploring such topics?

Participants

Based on the considerations discussed above, the workshop participants were recruited in two stages. The first stage was to recruit a group of speakers to introduce the topic to the workshop. Invited speakers (n=8) were drawn from several relevant professional backgrounds, including nursing, therapy and research, and included four pain nurses from a local private hospice, a care home and a National Health Service Trust; two art therapists who treated pain; a psychologist and an academic researcher from the field of communication studies. The group of speakers was gender balanced (n=4/4) and included people (n=2) from black and minority ethnic groups. The aim was to avoid any one-sided preference for certain medical practices, and so to reveal any gaps, differences or conflicting approaches.

The second stage was an open call that invited people with a general interest in this topic to participate. The workshop promotional material reflected our intention for an open, collaborative approach to exploring a topic of high general relevance. Considerable effort was made to promote the workshop to a diverse audience through local dementia networks, Alzheimer’s Research UK East Midlands and personal contacts. The workshop attracted a wider circle of 22 additional participants which represented a diverse group in terms of gender, ethnicity and educational background, including people with dementia and their carers (n=10), general as well as specialist nurses and allied health professionals (n=5), GPs (n=2), academics (n=3) and people working in hospital management (n=2).

To address issues of equality, power and labelling, we devoted significant consideration to the location of the workshop and the supporting environment. For instance, care was taken to avoid any context that would reinforce the authority of any particular profession, such as a university building, care home or arts space. Instead, we sought to identify a space with a friendly and non-hierarchical atmosphere, and conviviality research directed our attention to pubs and restaurants as appropriate settings (see the classic study by Snyder 1966 ). It was equally important to select a space with acoustics that would support conversation and a relaxed ‘workshop’ atmosphere. We ruled out office spaces as somewhat sterile, potentially hindering creative thought. Finally, we needed a space that would be accessible and safe for people with dementia. Taking account of all these factors, we identified a function room at the Trent Bridge cricket ground as a suitable venue for the workshop. The room combined a relaxed social setting of a pub or bar with large windows overlooking the cricket field and an open workplace atmosphere. As the room has also been used by another dementia group, stadium staff had received appropriate training and were able to guide participants on the day. The stadium is also just a few minutes from the main rail station, with good public transport connections and easy-access parking.

All participants were offered tea, coffee and biscuits, both on arrival and ad lib throughout the workshop. An adjacent hospitality room was available to anyone who needed a rest or a time-out. Finally, all participants received name tags to ensure that everybody felt included and on an equal footing with others (on the process, see figure 1 ).

Download figure
Open in new tab
Download powerpoint

Process flow chart.

The workshop was led by one of the researchers (CM) and assisted by a postgraduate student (MN), who created a transcript of the discussions (workshop protocol). The conduct of the workshop reflected the theoretical basis of its design. As described in the ‘Introduction’ section, the two key aspects were paraphrasing and a two-step communication flow. Thus, the workshop was divided into two distinct phases. Phase I placed the eight invited speakers in an inner circle, facing each other and surrounded by the other participants, in an arrangement modelled on televised political debates ( Livingstone and Lunt 1994 ). Those in the outer circle were instructed simply to watch and listen the debate being conducted by the inner circle. They were informed that, in a second step, the discussion would be extended to everyone, using the same paraphrasing approach. In this way, the outer circle was able to gain a sense of the content, as well as how such issues are debated and how arguments are presented. The invited speakers had been requested to respond to an opening question: “How and where have you come across pain in your practice/research?” The issues raised by each speaker were debated by the members of the inner circle, with the outer circle observing silently. This discussion lasted about 45 min.

Following this debate, all participants were invited to write down on post-it notes any thoughts that they had, either from their own experiences or arising from the speakers’ discussion. The post-it notes were mounted on wall charts, and then moved into thematic groups according to their content.

In phase II of the workshop, which followed a break for refreshments, participants were divided into small groups of four or five people that were facilitated by the discussants from phase I. They were encouraged to discuss the issues raised by the inner circle debate and by the post-it notes, alongside their own personal experience and views about pain and dementia.

In round I part of phase II, participants were invited to discuss at their tables the question of what pain meant to them. After 15–20 min, they were asked to make any notes on the post-it notes provided, and to stick these on a large communal sheet of paper ( figure 2 ), which everyone was then asked to view.

Results of thematic analysis. Round 1: How would you describe pain and what it means to you? And round 2: issues encountered on engaging with people with dementia who are experiencing pain (source: Post-it Notes Workshop).

In the round II of phase II, participants were asked a more specific question that invited them to reflect on their understanding of pain on engaging with people with dementia. This was designed to explore their professional or lived experience in this context. As before, they discussed the question for an agreed length of time and again attached post-it notes to another large sheet of paper ( figure 2 ).

In addressing these questions, participants were encouraged to be imaginative and experimental in their responses. Altogether phase II lasted about 60 min, and, at the end of the groups, the post-it notes were added to the initial display on the wall charts. Where it appeared appropriate, a regrouping of themes was made. The workshop ended at this point with general reflection and discussion, and participants were thanked for their time and for their contributions.

Ethical considerations

This was an exploratory study with participants taking part either by invitation through word of mouth or having volunteered. All individuals, including those living with a diagnosis of dementia, had adequate mental capacity to contribute. They were all aware that their participation was entirely voluntary and that they were free to withdraw or leave the workshop at any time. Participants were reimbursed of expenses.

Data management

The conversation protocol and post-it notes from phases I and II of the workshop constituted the data for addressing our first research question, that is, if any new understandings about pain in dementia emerged from the workshop. The post-it notes were anonymous by their nature, so there was no potential to breach confidentiality. The data were taken away after the workshop for subsequent analysis.

The methodology of the workshop was also assessed in order to address our second research question regarding the appropriateness, acceptability and effectiveness of the workshop. No formal measures were employed, so this assessment comprised observation of the levels of participant involvement; informal estimation of their levels of comfort, interest and enjoyment; looking out for any signs of distress and also in the informal comments made during and at the end of the workshop, as well as any feedback received afterwards, for example, by email. In addition, written notes of the session were taken by a postgraduate assistant, which also comprised evidence about the success or otherwise of the workshop. These notes had been made for various purposes, including making a record of the inner circle debate in phase I and issues around the conduct of the workshop; for example, if there were any problems with participants feeling uncomfortable or any issues with the catering or location.

Data analysis

We followed the approach by ⇓ Braun and Clarke (2006, 2019) conducting a reflexive thematic analysis of the initial group comments and post-it notes created during the workshop. This methodology was employed for the identification and analysis of patterns and themes based on the dataset. The aim of this was to draw out the key themes about the nature of pain, its importance in dementia, how it may be detected and how it is best managed. The final array of themes was arrived at by means of discussion between the authors. This enabled us to address the first research question for the study. In addition to looking at the data as a whole, we compared the post-its from phase I of the workshop with those from phase II to examine whether there appeared to have been any shift in the responses of workshop participants during the course of the workshop.

Workshop results

New understandings about pain.

In relation to our first research question, we found that pain is best understood as a mediated experience involving multiple stakeholders, and that the quality of pain is unpleasant and undefined. Findings will be presented for each of the two phases of the data collection. 3

In phase I, two main themes emerged: pain as a clinical phenomenon, and pain as a social relationship. Much of the initial exploration elicited and experimented with understandings of current medical practices and how these related to pain in people with dementia. Topics of these exchanges included the drive to objectify pain through protocols and professional standards, and how this becomes an obstacle because pain as an emotive quality can only be captured by less exact or non-verbal forms of expression. Some important contributions are listed below:

… the medical/clinical model of pain that fails to take account of complexity (participant F). Current regimented institutional settings make it difficult for professionals to function as human beings because they discourage feelings, and quality measures do not accommodate gut feeling (participant K). Palliative care is a good example of the need to be instinctive as an element of ethical practice (participant V). Evidence-based practice is important but limited at individual level and does not take the broader picture into account (participant T). Medical practice discourages the use of instinct (participant J).

By establishing this tension between ‘intuition’ and professional protocols a common obstacle, the exploration opened itself up to a new turning point. One participant, who had previously remained in the background, seized the opportunity to speak at greater length. Suggesting that this tension could be handled differently, she explained that she works outside institutional care, and mostly on a one-to-one basis with her clients: ‘I am able to work as a team with the family of the client and the client themselves. Based on these quality interactions, I can develop a deep understanding of the individual, which is needed to understand pain’ (Participant E). From here, the exploration entered a second stage, in which pain was considered as a two-way relationship, and participants reflected on what such a relationship should look like to ensure that pain was understood. This shift involved testing out the notion of two-way relationships in broader terms of the need to listen and the ability to understand the other person. This in turn was linked to the idea of building a relationship or connection at human level, recognising the person as an individual rather than merely as a patient or as someone experiencing pain. This broader view was considered crucial in understanding what pain meant to that person.

Other people told me what would be best for my father [who had dementia], and no one listened to him (participant J). Reflecting on my professional practice and it’s important to have the courage to listen without fear of litigation (participant K). If we cannot be human together, we cannot recognise pain (participant F). We need to build ‘connections’ into broader frameworks (participant T).

As mentioned above, in phase II, participants were invited to discuss two specific questions in two rounds (round I and round II) of discussion. In response to the first question in round I (“How would you describe pain and what does it mean to you?”), there were 150 responses recorded. A sentiment analysis on polarity (positive, negative, neutral) of the responses was applied to illustrate some variations in the clustering of certain descriptors over others (on this methodology within a context of pain, see Tighe et al 2015 ) The analysis showed that about 99 (66%) were negative, using descriptors like hurt , sharp or unpleasant to characterise pain. Other words used included causes of pain like trauma, injury or disease, and impacts such as exhaustion, fear or loss of balance. The remaining responses were either neutral or technical descriptions (eg, complex , expression of circumstances or unusual n=40; 27%) or positive (eg, love , peace or pride n=11; 7%). Most of the responses were single-word descriptors, which perhaps indicated that, at this point, participants were taking quite a narrow view of the meaning of pain.

In the second round of phase II, participants were asked to reflect on their understanding of pain on engaging with people with dementia. There were 131 responses in total ( figure 2 ). Two findings in particular stood out from this session. First, people more often described pain in sentences or short phrases rather than in single keywords.

Second, the meanings attributed to pain were more evenly balanced and included positive descriptions of pain in childbirth, sport and sexual encounters (n=44; 34%). In a notable change, some comments referred to the role of pain in the interaction between persons with dementia and their carers; for example, in providing support or helping to interpret the pain, accepting their self-evaluation of pain or providing love and tenderness. Many comments also rejected any easy classification of pain, describing it instead as challenging or undefined; beyond simply being an unpleasant emotion, pain was seen as part of a larger process.

To summarise, in round II of phase II, discussion of the meaning of pain within the context of dementia shifted noticeably, starting to place more emphasis on how pain involves a social relationship and interactions between people, not merely a simple biological phenomenon.

Discussion: towards a new understanding of pain

The central aim of this research was to employ an experimental approach to develop new conceptualisation of pain in people with dementia. The discussion in both phases of our workshop indicated that a narrow definition, such as medical terminology that typically defines pain as an unpleasant bodily sensation, fails to grasp important aspects of the meaning of pain. In particular, medical definitions fail to acknowledge the relational nature of pain; in other words, because pain is seen as a kind of bodily experience, which can be accessed more or less directly and objectively by the affected person, it is viewed as unmediated. However, the present findings suggest that pain should instead be understood as mediated, because it depends jointly on our nervous system, our brain, how we feel about things and how we relate to others. By mediated, we mean that pain ‘exists’ and acquires its meaning within this network. As evidenced through our experimental theoretical method, we could show that pain is not owned or produced by the body speaking for itself, individual thoughts and feelings, but the making of the meaning of pain rests on a distributed agency ( Garud and Karnøe 2005 ). As such, we can conclude that pain in general, but especially in more advanced stages of dementia acquires a stronger relational emphasis as it requires witnessing and advocating—given the increasing communication difficulties as dementia progresses. Additionally, it can be said that what pain introduces more than just a negative framing into this network; it disrupts the process of meaning-making—the consequences of which are not easy grasped. The disruptive quality of pain can be enticing, because it presents something new and surprising, and may lead to frustration and fear if the making of meaning of pain is impaired for any length of time, because it remains unmediated. Research shows that such unresolved interruptions of meanings lead to a reaction on itself with a spiralling effect known as pain catastrophising ( Quartana et al 2009 ). We can conclude, then, that pain can be defined as an interruption of the socially mediated process of bodily meaning-making. Such a definition of pain suggests that future forms of pain assessment need to recognise that pain is mediated ( van Hout et al 2023 ). For instance, to one person the interruption of the socially mediated process of bodily meaning-making might imply notions of failure, for instance, failing to function as a provider for their family, whereas another person might consider this failure as a source to be a potential burden for others. Such different meanings of pain will impact on people’s awareness, and on their understanding of the severity of the pain. These considerations have two further implications. The emphasis on pain as a socially mediated process puts greater emphasis on its ‘intersubjective’ quality, which means that the diagnoses of pain involves the person expressing pain, and the person interpreting this expression co-producing pain as an epistemic object. Approaches like person-centred care, stress such a reflective quality in diagnosis and treatment (on such recent developments, see Wigert et al 2021 ). Furthermore, this revised definition of pain is helpful within a dementia context, and contributes a wider understanding about the changing importance of pain in contemporary society. For instance, if pain remains unmediated, it is likely to increase in importance, which is why some authors speak of a pain epidemic ( Rosa et al 2022 ; Sessle 2011, 2012 ). Zajacova et al (2021) could demonstrate in their research that the experience of prolonged pain has been increasing across all age groups and genders in the USA in the past 20 years with the highest increase among men between 20 and 40 years. Such findings reveal that pain cannot simply be associated with old age, but raises much more profound question about the mediation of pain in society. This would be an important area of research for medical practitioners as well as social scientists, because the evolution of pain, for example, the rise in awareness of bodily sensitiveness, the social acceptance (or lack of it) of interruptions to bodily meaning-making is in co-variation with changing societal contexts ( Morgner 2022 ).

Our second aim was that we wanted to test a novel method in terms of it being appropriate, accessible and effective to unravel implicit and tacit theoretical heuristics.

The workshop appeared to be successful in terms of appropriateness; it was not difficult to find speakers from diverse contexts, and all participants acknowledged the importance of the issue of pain in dementia. In relation, to accessibility, the workshop progressed smoothly throughout and active contributions were made by all participants. Any initial hesitancy among participants about the inner and outer circle methods in phase I was assuaged by reassuring them that they would have time to contribute their views later. Participants with a diagnosis of dementia, along with their relatives, clearly welcomed the opportunity to contribute. They would actively take notes of the discussion from phase I and although they sometimes needed encouragement to speak out, individuals with dementia made important contributions and appeared comfortable in doing so during phase II. While one or two of these participants occasionally needed a brief rest, no one withdrew from the workshop because of distress. There were numerous positive comments, including informal feedback received at the end of the event. Finally, its effectiveness lies in its ability to deepen our understanding of pain in dementia, and the findings suggest that assessment and treatment requires a more interdisciplinary approach involving neuro-psychological-sociological inputs. These findings suggest that this experimental-theoretical format has contributed to the development of a new understanding of pain through an active contribution of participants from different social context by devising an accessible methodology based on performing arts, art education and communication studies.

Strengths and limitations

Based on established theory from the communications literature and on previous work with analogous stakeholder groups, this novel study proved successful in attracting diverse participants, including people with dementia, and produced new theoretical insights into how pain and dementia can be understood. The methodology appears robust and may prove more effective than reviews of existing literature.

While the sample was small, reflecting the relatively small population of people with dementia, the participants brought a broad range of experience of research and dementia and end of life care to the study. Other limitations include the extensive preparation and the longer workshop duration as compared with the typical focus group, which means that some individuals may become fatigued before the end of the session, and the need for careful management of the environment for the participants. One participant left the session during the break due to a malfunctioning hearing device, exacerbated by the room’s acoustics.

Implications

The study is important for future research on pain in dementia and research on pain in general as it provides a methodology that could be used for further enquiry, for example, in exploring further what methods are tried and tested and may be useful in practice. The relational mediated aspects of pain require more attention in future research designs. We have also shown that people with dementia can contribute well to discussions about pain, despite its abstract nature and difficulties in discussing it. Therefore, future studies should facilitate further involvement of people with lived experience as much as seeking proxy views from carers and professionals.

Our research also noted that less exact or non-verbal forms of expression should be considered in the diagnosis of pain. As noted, existing measures heavily rely on verbal or written means; however, the is an emerging area of research exploring alternative modes of expression, such as using visual art or music ( Padfield et al 2018 ). While such art interventions have notable effects on pain treatment, they have not yet been considered for assessing a person’s understanding of pain.

Staff training on pain in dementia is somewhat limited and tends to focus on the use of pain rating scales, which as we have argued above have considerable limitations. Our experimental methodology could equally well be applied as a training method, where professional staff could learn much from the contributions of participants with dementia. This would help enhance their confidence in supporting people with dementia who may be in pain, and may be applied to topics beyond pain ( Agarwal et al 2020 ). This emphasis on staff training is also justified, because the recognition of pain is on an individual expressing pain, and adds weight to the caregiver to interpret verbal and non-verbal expressions appropriately or even to assist the person in expressing pain ( Weissman and Matson 1999 ).

Finally, our theoretical-experimental method also opens up broader questions about the state of theorisation in the medical field and the need to intervene in the mostly implicit practice of medical theory ( Neilson 2016 ). Our method will equip medical practitioners and other clinical professionals with a robust as well as creative methodological process to enhance the formulation of medical concepts. This is an area where medical theory can benefit from recent debates in the field of social theory ( Swedberg 2012 ).

Conclusions

This study has used methodological interventions derived from museum and communication studies to develop a new theoretical-experimental approach to explore pain and its expression in people with dementia. This has led to a new definition of pain as an interruption of the socially mediated process of bodily meaning-making. Furthermore, we have emphasised the social, relational quality of pain that sits alongside its biological origins, and from this arises the notion of witness and testimony on the part of others as an essential component of responding to and managing pain in dementia. This means that future research needs to consider the expression of pain, and its interpretation, for instance, to what extent professional staff are able to engage in this interpretation. This is in contrast to the recently revised definition of pain by the International Association for the Study of Pain, which stated that this relational part is ‘not an essential component of the definition’ of pain ( Raja et al 2020 , 3). Finally, our experimental intervention reflected on the process of generating theory by making the various steps and considerations explicit as well as critically reflecting on intervening into this process. We therefore contribute to a yet unexplored field of research, which considers the practice of medical theory, that is, how concepts and theories are generated in the medical field.

Ethics statements

Patient consent for publication.

Not applicable.

Ethics approval

This study was approved by Alzheimer Research UK 1000393. Participants gave informed consent to participate in the study before taking part.

Acknowledgments

Grateful thanks to all the participants in the workshop and thanks to Dr Marie Nugent who took notes and acted as assistant to CM as workshop leader.

1. This negative conception of pain also overlooks a broader spectrum of meanings that can have a more positive connotation, for instance, the pain of childbirth or sexually stimulated pain.

2. Others have noted that the overt medicalisation of pain neglects that pain is a meaningful construct that makes us human (see the chapter ‘The Killing of Pain’ in Illich 1982 ). Likewise, Jünger and Durst (2008, 3 ) noted in his essay On Pain its very foundational notion, but not as a mere bodily sensation, but as a relational construct: “Tell me your relation to pain, and I will tell you who you are!”

3. The notion of a mediated experience stresses the mutual involvement of the ‘knower’ and the ‘known’ within a context of great complexity ( Morgner 2019 ).

Bibliography

Achterberg W. ,
Lautenbacher S. ,
Husebo B. ,
Erdal A. , and
Agarwal G. ,
McNulty M. ,
Santiago K. M. ,
Torrents H. , and
Caban-Martinez A. J.
Alexander J. C
Alzheimer’s Disease International
Armour R. , and
Murphy-Jones B.
Beavin J. ,
Jackson D. , and
Watzlawick P.
Bendelow G. A. , and
Williams S. J.
Bennett C. R
Boardman F. K
Bond J. , and
Braun V. , and
Breivik H. ,
Borchgrevink P. C. ,
Allen S. M. ,
Rosseland L. A. ,
Romundstad L. ,
Hals E. K. B. ,
Kvarstein G. , et al
Browne J. ,
Edwards D. A. ,
Rhodes K. M. ,
Brimicombe D. J. , and
Payne R. A.
Brunkert T. ,
Ruppen W. , and
Burn A.-M. ,
Goodman C. ,
Robinson L. ,
Norton S. ,
Bennett H. , et al
Caron-Flinterman J. F. ,
Broerse J. E. W. , and
Bunders J. F. G.
Quintner J. , and
van Rysewyk S.
De Silva T. ,
Hodges P. W. ,
Costa N. , and
Setchell J.
Elberse J. E. ,
Caron-Flinterman J. F. , and
Broerse J. E. W.
Garand L. ,
Lingler J. H. ,
Conner K. O. , and
Garud R. , and
Pieper D. ,
Fernandes R. M. ,
Tricco A. C. ,
Brennan S. E. , et al
Zwakhalen S. , and
Swafford K.
Horgas A. L. ,
McLennon S. M. , and
Floetke A. L.
Housen A. C
Hucklenbroich P
Jordan J. E. ,
Buchbinder R. , and
Osborne R. H.
Jünger E. , and
Durst D. C.
Katz E. , and
Lazarsfeld P.
Kelleher D. ,
Gabe J. , and
Williams G.
Cunningham-Burley S. , and
Lautenbacher S. , and
Liao Y. J. ,
Jao Y. L. ,
Berish D. ,
Hin A. S. ,
Mogle J. , and
Livingstone S. , and
Loeser J. D
MacKay D. M
McCarthy M. ,
Addington-Hall J. , and
Morrison C. , and
Morrison R. S. , and
Muijs D. , and
Reynolds D.
Manafo E. ,
Petermann L. ,
Mason-Lai P. , and
Vandall-Walker V.
Martin A. J. ,
Anderson J. ,
Way J. , and
Mathieson S. ,
Maher C. G. ,
Terwee C. B. ,
Folly de Campos T. , and
Lin C.-W. C.
Mumford M. D. ,
Giorgini V. ,
Gibson C. , and
Oosterman J. M. ,
Hendriks H. ,
White N. , and
Sampson E. L.
Padfield D. ,
Semino E. ,
Williams A. C. de C. , and
Zakrzewska J. M.
Popay J. , and
Quartana P. J. ,
Campbell C. M. , and
Edwards R. R.
Raja S. N. ,
Carr D. B. ,
Finnerup N. B. ,
Gibson S. ,
Keefe F. J. , et al
Rathmell J. P. , and
Fields H. L.
Reykowski J
Rosa W. E. ,
Knaul F. M. ,
Pettus K. I. ,
Bruera E. , and
Rajagopal M. R.
Savoy M. L. , and
Yunyongying P.
Scherder E. ,
Pickering G. ,
Benedetti F. , and
Lautenbacher S.
Schofield P
Scrutton J. , and
Brancati C. U.
Sessle B. J
Devane D. ,
Begley C. M. , and
Snyder C. R
Gonzalez-Polledo E. , and
Tighe P. J. ,
Goldsmith R. C. ,
Gravenstein M. ,
Bernard H. R. , and
Fillingim R. B.
Timmons S. ,
O’Shea E. ,
O’Neill D. ,
Gallagher P. ,
de Siún A. ,
McArdle D. ,
Gibbons P. , and
Kennelly S.
van Hout F. ,
van Rooden A. , and
Weissman D. E. , and
Wigert H. ,
Nilsson S. ,
Dalenius K. , and
Golsäter M.
Wilkinson C. , and
Weitkamp E.
Williams A. C. C. , and
Craig K. D.
Zajacova A. ,
Grol-Prokopczyk H. , and

X @TomDening

Contributors This article is the result of a collaborative effort among several team members, each of whom has made substantial contributions to its completion: CM (lead author) was responsible for conceptualising the study, leading the research design and overseeing the project. He played a pivotal role in the analysis and interpretation of data and drafted the initial manuscript and is responsible for the overall content. KHD (co-author) contributed significantly to the data collection process and was involved in the data analysis. She assisted in drafting the manuscript and revising it critically for important intellectual content, focusing specifically on the qualitative analysis section. TD (co-author) was instrumental in the literature review process, contributing to the formulation of the research question and hypothesis. He also played a key role in interpreting the results in relation to existing literature and contributed to writing the discussion section of the manuscript. BG (co-author) provided expertise in the specialised analytical methods used in the study. He offered technical guidance during the project and contributed to the writing and revision of the methods and results sections. Each author has approved the final version of the manuscript and agrees to be accountable for all aspects of the work, ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding Supported by a grant to CM from Alzheimer’s Research UK East Midlands network (ref no. 1000393).

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

Read the full text or download the PDF:

The Project Gutenberg eBook of Fir-Flower Tablets: Poems Translated from the Chinese

Title : Fir-Flower Tablets: Poems Translated from the Chinese

Translator : Florence Wheelock Ayscough

Release date : February 9, 2015 [eBook #48222]

Language : English

Credits : Produced by David Edwards, Christopher Wright, some illustration images from TIA and the Online Distributed Proofreading Team at http://www.pgdp.net (This book was produced from scanned images of public domain material from the Google Print project.)

FIR-FLOWER TABLETS POEMS FROM THE CHINESE

MAP OF CHINA SHOWING In Black, the Present Divisions and names In Red, the Ancient Districts &c. referred to in the Poems (The transliteration used is that of the Post Office)

FIR-FLOWER TABLETS

Poems translated from the Chinese by

FLORENCE AYSCOUGH

Hon. Mem. North China Branch, Royal Asiatic Society

ENGLISH VERSIONS

BOSTON AND NEW YORK

HOUGHTON MIFFLIN COMPANY

The Riverside Press, Cambridge 1921

PRINTED IN THE U.S.A.

BY AMY LOWELL

Let me state at the outset that I know no Chinese. My duty in Mrs. Ayscough's and my joint collaboration has been to turn her literal translations into poems as near to the spirit of the originals as it was in my power to do. It has been a long and arduous task, but one which has amply repaid every hour spent upon it. To be suddenly introduced to a new and magnificent literature, not through the medium of the usual more or less accurate translation, but directly, as one might burrow it out for one's self with the aid of a dictionary, is an exciting and inspiring thing. The method we adopted made this possible, as I shall attempt to show. The study of Chinese is so difficult that it is a life-work in itself, so is the study of poetry. A sinologue has no time to learn how to write poetry; a poet has no time to learn how to read Chinese. Since neither of us pretended to any knowledge of the other's craft, our association has been a continually augmenting pleasure.

I was lucky indeed to approach Chinese poetry through such a medium. The translations I had previously read had given me nothing. Mrs. Ayscough has been to me the pathway to a new world. No one could be a more sympa [vi] thetic go-between for a poet and his translator, and Mrs. Ayscough was well-fitted for her task. She was born in Shanghai. Her father, who was engaged in business there, was a Canadian and her mother an American. She lived in China until she was eleven, when her parents returned to America in order that their children might finish their education in this country. It was then that I met her, so that our friendship is no new thing, but has persisted, in spite of distance, for more than thirty years, to ripen in the end into a partnership which is its culmination. Returning to China in her early twenties, she became engaged to an Englishman connected with a large British importing house in Shanghai, and on her marriage, which took place almost immediately, went back to China, where she has lived ever since. A diligent student of Chinese life and manners, she soon took up the difficult study of literary Chinese, and also accepted the position of honorary librarian of the library of the North China Branch of the Royal Asiatic Society. Of late years, she has delivered a number of lectures on Chinese subjects in China, Japan, America, and Canada, and has also found time to write various pamphlets on Chinese literature and customs.

In the Autumn of 1917, Mrs. Ayscough arrived in America on one of her periodic visits to this country. She brought with her a large collection of Chinese paintings for exhibition, and among these paintings were a number of [vii] examples of the "Written Pictures." Of these, she had made some rough translations which she intended to use to illustrate her lectures. She brought them to me with a request that I put them into poetic shape. I was fascinated by the poems, and, as we talked them over, we realized that here was a field in which we should like to work. When she returned to China, it was agreed that we should make a volume of translations from the classic Chinese writers. Such translations were in the line of her usual work, and I was anxious to read the Chinese poets as nearly in the original as it was possible for me to do. At first, we hardly considered publication. Mrs. Ayscough lives in Shanghai and I in Boston, and the war-time mails were anything but expeditious, but an enthusiastic publisher kept constantly before us our ultimate, if remote, goal. Four years have passed, and after many unavoidable delays the book is finished. We have not done it all by correspondence. Mrs. Ayscough has come back to America several times during its preparation; but, whether together or apart, the plan on which we have worked has always been the same.

Very early in our studies, we realized that the component parts of the Chinese written character counted for more in the composition of poetry than has generally been recognized; that the poet chose one character rather than another which meant practically the same thing, because [viii] of the descriptive allusion in the make-up of that particular character; that the poem was enriched precisely through this undercurrent of meaning in the structure of its characters. But not always—and here was the difficulty. Usually the character must be taken merely as the word it had been created to mean. It was a nice distinction, when to allow one's self the use of these character undercurrents, and when to leave them out of count entirely. But I would not have my readers suppose that I have changed or exaggerated the Chinese text. Such has not been the case. The analysis of characters has been employed very rarely, and only when the text seemed to lean on the allusion for an added vividness or zest. In only one case in the book have I permitted myself to use an adjective not inherent in the character with which I was dealing—and, in that case, the connotation was in the word itself, being descriptive of an architectural structure for which we have no equivalent—except in the "Written Pictures," where, as Mrs. Ayscough has stated in her Introduction, we allowed ourselves a somewhat freer treatment.

It has been necessary, of course, to acquire some knowledge of the laws of Chinese versification. But, equally of course, these rules could only serve to bring me into closer relations with the poems and the technical limits of the various forms. It was totally impossible to follow either the rhythms or the rhyme-schemes of the originals. All [ix] that could be done was to let the English words fall into their natural rhythm and not attempt to handicap the exact word by introducing rhyme at all. This is the method I followed in my translations of French poems in my book, "Six French Poets." I hold that it is more important to reproduce the perfume of a poem than its metrical form, and no translation can possibly reproduce both.

Our plan of procedure was as follows: Mrs. Ayscough would first write out the poem in Chinese. Not in the Chinese characters, of course, but in transliteration. Opposite every word she put the various meanings of it which accorded with its place in the text, since I could not use a Chinese dictionary. She also gave the analyses of whatever characters seemed to her to require it. The lines were carefully indicated, and to these lines I have, as a rule, strictly adhered; the lines of the translations usually corresponding, therefore, with the lines of the originals. In the few poems in which the ordering of the lines has been changed, this has been done solely in the interest of cadence.

I had, in fact, four different means of approach to a poem. The Chinese text, for rhyme-scheme and rhythm; the dictionary meanings of the words; the analyses of characters; and, for the fourth, a careful paraphrase by Mrs. Ayscough, to which she added copious notes to acquaint me with all the allusions, historical, mythological, geographical, and technical, that she deemed it necessary for me to know. [x] Having done what I could with these materials, I sent the result to her, when she and her Chinese teacher carefully compared it with the original, and it was returned to me, either passed or commented upon, as the case might be. Some poems crossed continent and ocean many times in their course toward completion; others, more fortunate, satisfied at once. On Mrs. Ayscough's return to America this year, all the poems were submitted to a farther meticulous scrutiny, and I can only say that they are as near the originals as we could make them, and I hope they may give one quarter of the pleasure to our readers that they have to us in preparing them.

Thanks are due to the editors of The North American Review , The Bookman , The Dial , The New York Evening Post , Poetry , and Asia , for permission to reprint poems which have already appeared in their magazines.

ILLUSTRATIONS

Introduction.

By FLORENCE AYSCOUGH

There has probably never been a people in whose life poetry has played such a large part as it has done, and does, among the Chinese. The unbroken continuity of their history, throughout the whole of which records have been carefully kept, has resulted in the accumulation of a vast amount of material; and this material, literary as well as historical, remains available to-day for any one who wishes to study that branch of art which is the most faithful index to the thoughts and feelings of the "black-haired race," and which, besides, constitutes one of the finest literatures produced by any race the world has known.

To the confusion of the foreigner, however, Chinese poetry is so made up of suggestion and allusion that, without a knowledge of the backgrounds (I use the plural advisedly) from which it sprang, much of its meaning and not a little of its beauty is necessarily lost. Mr. Arthur Waley, in the preface to his "A Hundred and Seventy Chinese Poems," says: "Classical allusion, always the vice of Chinese poetry, finally destroyed it altogether." Granting the unhappy truth of this statement, the poetry of China is nevertheless so human and appealing as to speak with great force even [xx] to us who live under such totally different conditions; it seems worth while, therefore, to acquire a minimum of knowledge in regard to it and so increase the enjoyment to be derived from it. In the present collection, I have purposely included only those poems in which this national vice is less in evidence; and this was not a difficult task. There is such an enormous body of Chinese poetry that the difficulty has been, not what to take, but what to leave out. I have been guided somewhat by existing translations, not wishing to duplicate what has already been adequately done, when so much still remains untouched. Not that all these poems appear in English for the first time, but many of them do; and, except for Mr. Waley's admirable work, English renderings have usually failed to convey the flavour of the originals.

Chinese scholars rank their principal poets in the following order: Tu Fu, Li T'ai-po, and Po Chü-i. Realizing that, naturally, in any literature, it is the great poets which another nation wishes to read, I have purposely kept chiefly to them, and among them to Li T'ai-po, since his poems are of a universal lyricism. Also, Mr. Waley has devoted his energies largely to Po Chü-i. Tu Fu is very difficult to translate, and probably for that reason his work is seldom given in English collections of Chinese poems. Some of his simpler poems are included here, however. A small section of the book is devoted to what the Chinese [xxi] call "written-on-the-wall-pictures." I shall come back to these later.

The great stumbling-block which confronts the translator at the outset is that the words he would naturally use often bring before the mind of the Occidental reader an entirely different scene to that actually described by the Oriental poet. The topography, the architecture, the fauna and flora, to say nothing of the social customs, are all alien to such a reader's own surroundings and cannot easily be visualized by him. Let me illustrate with a modern poem, for it is a curious fact that there has lately sprung up in America and England a type of poetry which is so closely allied to the Chinese in method and intention as to be very striking. This is the more remarkable since, at the time of its first appearance, there were practically no translations of Chinese poems which gave, except in a remote degree, the feeling of the originals. So exact, in fact, is this attitude toward the art of poetry among the particular group of poets to whom I have reference and the Chinese masters, that I have an almost perfect illustration of the complications of rendering which a translator runs up against by imagining this little poem of Miss Lowell's being suddenly presented to a Chinese scholar in his grass hut among the Seven Peaks:

By Amy Lowell

As we read this poem, instantly pictures of American travel start before our eyes: rushing trains with plush-covered seats, negro porters in dust-grey suits, weary ticket-collectors; or marble-floored hotel entrances, clanging elevator doors, and hurrying bell-boys, also the vivid suggestion of a beautiful American house. But our scholar would see none of this. To him, a journey is undertaken, according to the part of the country in which he must travel, either in a boat, the types of which are infinitely varied, from the large, slow-going travelling barge capable of carrying many passengers, to the swifter, smaller craft [xxiii] which hold only two or three people; in one of the several kinds of carriages; in a wheelbarrow, a sedan chair, a mule litter, or on the back of an animal—horse, mule, or donkey, as the case may be. Again, there is no English-speaking person to whom "Home, Sweet Home" is not familiar; in a mental flash, we conclude the stanza suggested by the first line, and know, even without the title, that the subject of the poem is homesickness. Our scholar, naturally, knows nothing of the kind; the reference is no reference to him. He is completely at sea, with no clue as to the emotion the poem is intended to convey, and no understanding of the conditions it portrays. Poem after poem in Chinese is as full of the intimate detail of daily life, as dependent upon common literary experience, as this. There is an old Chinese song called "The Snapped Willow." It, too, refers to homesickness and allusions to it are very frequent, but how can an Occidental guess at their meaning unless he has been told? In this Introduction, therefore, I have endeavoured to give as much of the background of this Chinese poetry as seems to me important, and, since introductions are made to be skipped, it need detain no one to whom the facts are already known.

The vast country of China, extending from the plains of Mongolia on the North to the Gulf of Tonquin on the South, a distance of somewhat over eighteen hundred miles, and from the mountains of Tibet on the West to the [xxiv] Yellow Sea on the East, another stretch of about thirteen hundred miles, comprises within its "Eighteen Provinces" practically every climate and condition under which human beings can exist with comfort. A glance at the map will show the approximate positions of the ancient States which form the poetic background of China, and it will be noticed that, with the exception of Yüeh, they all abut either on the Huang Ho, better known as the Yellow River, or on the Yangtze Kiang. These two great rivers form the main arteries of China, and to them is largely due the character of the people and the type of their mythology.

The Yellow River, which in the old mythology was said to have its source in the Milky Way (in the native idiom, "Cloudy" or "Silver River"), really rises in the K'un Lun Mountains of Central Asia; from thence its course lies through the country supposed to have been the cradle of the Chinese race. It is constantly referred to in poetry, as is also its one considerable tributary, the Wei River, or "Wei Water," its literal name. The Yellow River is not navigable for important craft, and running as it does through sandy loess constantly changes its course with the most disastrous consequences.

The Yangtze Kiang, "Son of the Sea," often referred to as the "Great River," is very different in character. Its source lies among the mountains of the Tibetan border, where it is known as the "River of Golden Sand." After [xxv] flowing due South for several hundred miles, it turns abruptly to the North and East, and, forcing its way through the immense wall of mountain which confronts it, "rushes with incredible speed" to the far-off Eastern Sea, forming in its course the Yangtze Gorges, of which the most famous are the San Hsia, or "Three Chasms." To these, the poets never tire of alluding, for, to quote Li T'ai-po, the cliffs rise to such a height that they seem to "press Green Heaven." The water is low during the Winter months, leaving many treacherous rocks and shoals uncovered, but rises to a seething flood during the Summer, when the Tibetan snows are melting. The river is then doubly dangerous, as even great pinnacles of rock are concealed by the whirling rapids. Near this point, the Serpent River, so-called from its tortuous configuration, winds its way through deep ravines and joins the main stream. As may be imagined, navigation on these stretches of the river is extremely perilous, and an ascent of the Upper Yangtze takes several months to perform since the boats must be hauled over the numerous rapids by men, called professionally "trackers," whose work is so strenuous that they are bent nearly double as they crawl along the tow-paths made against the cliffs. In spite of the precipitous nature of the banks, many towns and villages are built upon them and rise tier on tier up the mountain sides. Having run about two-thirds of its course and reached the [xxvi] modern city of Hankow, the Great River changes its mood and continues on its way, immense and placid, forming the chief means of communication between the sea and Central China. The remarkably fertile country on either side is intersected by water-ways, natural and artificial, used instead of roads, which latter do not exist in the Yangtze Valley, their place being taken by paths, some of which are paved with stone and wide enough to accommodate two or three people abreast.

As travel has always been very popular, every conceivable form of water-borne craft has sprung up, and these the poets constantly used as they went from the capital to take up their official posts, or from the house of one patron to another, the ancient custom being for the rich to entertain and support men of letters with whom they "drank wine and recited verses," the pastime most dear to their hearts. The innumerable poems of farewell found among the works of all Chinese poets were usually written as parting gifts from the authors to their hosts.

As it nears the sea, the river makes a great sweep round Nanking and flows through what was once the State of Wu, now Kiangsu. This and the neighbouring States of Yüeh and Ch'u (the modern Chêkiang and parts of Hunan, Kweichow, and Kiangsi) is the country painted in such lovely, peaceful pictures by Li T'ai-po and his brother poets. The climate being mild, the willows which grow on the [xxvii] banks of the rivers and canals are seldom bare and begin to show the faint colour of Spring by the middle of January; and, before many days, the soft bud-sheaths, called by the Chinese "willow-snow," lie thick on the surface of the water. Plum-trees flower even while the rare snow-falls turn the ground white, and soon after the New Year, the moment when, according to the Chinese calendar, Spring "opens," the fields are pink with peach-bloom, and gold with rape-blossom, while the air is sweetly scented by the flowers of the beans sown the Autumn before. Walls and fences are unknown, only low ridges divide the various properties, and the little houses of the farmers are built closely together in groups, as a rule to the South of a bamboo copse which acts as a screen against the Northeast winds prevailing during the Winter; the aspect of the rich plain, which produces three crops a year, is therefore that of an immense garden, and the low, grey houses, with their heavy roofs, melt into the picture as do the blue-coated people who live in them. Life is very intimate and communistic, and the affairs of every one in the village are known to every one else. The silk industry being most important, mulberry-trees are grown in great numbers to provide the silk-worms with the leaves upon which they subsist, and are kept closely pollarded in order that they may produce as much foliage as possible.

This smiling country on the river-banks, and to the [xxviii] South, provides a striking contrast to those provinces lying farther North and West. Shantung, the birthplace of Confucius, is arid and filled with rocky, barren hills, and the provinces of Chili, Shansi, Shensi, and Kansu, which extend Westward, skirting the Great Wall, are also sandy and often parched for lack of water, while Szechwan, lying on the Tibetan border, although rich and well irrigated, is barred from the rest of China by tremendous mountain ranges difficult to pass. One range, called the "Mountains of the Two-Edged Sword," was, and is, especially famous. It formed an almost impassable barrier, and the great Chu Ko-liang, therefore, ordered that a roadway, of the kind generally known in China as chan tao (a road made of logs laid on piers driven into the face of a cliff and kept secure by mortar) be built, so that travellers from Shensi might be able to cross into Szechwan. This road is described by Li T'ai-po in a very beautiful poem, "The Terraced Road of the Two-Edged Sword Mountains."

These varied scenes among which the poets lived differed again from those which flashed before their mental eyes when their thoughts followed the soldiers to the far Northwest, to the country where the Hsiung Nu and other Mongol tribes lived, those Barbarians, as the Chinese called them, who perpetually menaced China with invasion, who, in the picturesque phraseology of the time, desired that their horses should "drink of the streams of the South." [xxix] These Mongol hordes harassed the Chinese State from its earliest days; it was as a defence against them that the "First Emperor" erected the Great Wall, with a length of "ten thousand li " as Chinese hyperbole unblushingly states—its real length is fifteen hundred miles. This defence could, however, merely mitigate, not avert, the evil; only constant effort, constant fighting, could prevent the Mongol hordes from overrunning the country.

Beyond the Jade Pass in Kansu, through which the soldiers marched, lay the desert and the steppes stretching to the very "Edge of Heaven," and on this "edge" stood the "Heaven-high Hills"; while, on the way, surrounded by miles of sand, lay the Ch'ing Hai Lake (Green, or Inland, Sea), a dreary region at best, and peopled by the ghosts of countless soldiers who had fallen in battle on the "Yellow Sand Fields."

In addition to these backgrounds of reality, that of the Fertile Empire and that of the Barren Waste, there was another—that of the "Western Paradise" inhabited by the Hsi Wang Mu (Western Empress Mother) and those countless beings who, after a life in this world, had attained Immortality and dwelt among the Hsien , supernatural creatures living in this region of perfect happiness supposed to lie among the K'un Lun Mountains in Central Asia. From the spontaneous manner in which they constantly refer to it, and from the vividness of the pictures [xxx] suggested by their references to it, one can almost question whether this Fairy World, the World of Imagination, with its inhabitants, were not as real to the writers of the early days as was the World of Actuality. Thus the topography of Chinese poetry may be said to fall into three main divisions, and allusions are made to

1. The beautiful scenes in the Eighteen Provinces.
2. The desolate region beyond the Jade Pass.
3. The glorious "Western Paradise."

Ideals determine government, and government determines social life, and social life, with all that the term connotes, is the essence of every literature.

The theory upon which the Chinese State was established is exceedingly interesting, and although the ideal was seldom reached, the system proved enduring and brought happiness to the people who lived under it.

The Emperor was regarded as the Son of the Celestial Ruler, as Father of his people, and was supposed to direct his Empire as a father should direct his children, never by the strong arm of force, but by loving precept and example. In theory, he held office only so long as peace and prosperity lasted, this beneficent state of things being considered a proof that the ruler's actions were in accordance with the decree of Heaven. Rebellion and disorder were an equal proof that the Son of Heaven had failed in his great [xxxi] mission; and, if wide-spread discontent continued, it was his duty to abdicate. The "divine right of kings" has never existed in China; its place has been taken by the people's right to rebellion.

This system created a very real democracy, which so struck the Dutchman, Van Braam, when he conducted a commercial embassy to the Court of Ch'ien Lung in 1794, that he dedicated his account of the embassy to "His Excellency George Washington, President of the United States," in the following remarkable manner:

Sir, Travels among the most ancient people which now inhabits this globe, and which owes its long existence to the system which makes its chief the Father of the National Family, cannot appear under better auspices than those of the Great Man who was elected, by the universal suffrage of a new nation, to preside at the conquest of liberty, and in the establishment of a government in which everything bespeaks the love of the First Magistrate for the people. Permit me thus to address the homage of my veneration to the virtues, which in your Excellency, afford so striking a resemblance between Asia, and America. I cannot shew myself more worthy of the title of Citizen of the United States, which is become my adopted country, than by paying a just tribute to the Chief, whose principles and sentiments, are calculated to procure them a duration equal to that of the Chinese Empire.

The semi-divine person of the Emperor was also regarded as the "Sun" of the Empire, whose light should shine on high and low alike. His intelligence was compared to the penetrating rays of the sun, while that of the Empress found its counterpart in the soft, suffusing brilliance [xxxii] of the moon. In reading Chinese poetry, it is important to keep these similes in mind, as the poets constantly employ them; evil counsellors, for instance, are often referred to as "clouds which obscure the sun."

The Son of Heaven was assisted in the government of the country by a large body of officials, drawn from all classes of the people. How these officials were chosen, and what were their functions, will be stated presently. At the moment, we must take a cursory glance at Chinese history, since it is an ever-present subject of allusion in poetry.

Two favourite, and probably mythical, heroes, the Emperors Yao and Shun, who are supposed to have lived in the semi-legendary period two or three thousand years before the birth of Christ, have been held up ever since as shining examples of perfection. Shun chose as his successor a man who had shown such great engineering talent in draining the country, always in danger of floods from the swollen rivers, that the Chinese still say: "Without Yü, we should all have been fishes." Yü founded the first hereditary dynasty, called the Hsia Dynasty, and, since then, every time the family of the Emperor has changed, a new dynasty has been inaugurated, the name being chosen by its first Emperor. With Yü's accession to the throne in 2205 B.C., authentic Chinese history begins.

Several centuries later, when Yü's descendants had deteriorated and become effete, a virtuous noble named [xxxiii] T'ang organized the first of those rebellions against bad government so characteristic of Chinese history. He was successful, and in his "Announcement to the Ten Thousand Districts," set forth what we should call his platform in these words: "The way of Heaven is to bless the good and punish the wicked. It sent down calamities upon the house of Hsia to make manifest its crimes. Therefore I, the little child, charged with the decree of Heaven and its bright terrors, did not dare forgive the criminal.... It is given to me, the one man, to ensure harmony and tranquillity to your State and families; and now I know not whether I may not offend the Powers above and below. I am fearful and trembling lest I should fall into a deep abyss." The doctrine that Heaven sends calamity as a punishment for man's sin is referred to again and again in the ancient "Book of History" and "Book of Odes." It is a belief common to all primitive peoples, but in China it persisted until the present republic demolished the last of the long line of dynastic empires.

T'ang made a great and wise ruler. The Dynasty of Shang, which he founded, lasted until 1122 B.C. , and was succeeded by that of Chou, the longest in the annals of Chinese history—so long, indeed, that historians divide it into three distinct periods. The first of these, "The Rise," ran from 1122 B.C. to 770 B.C. ; the second, "The Age of Feudalism," endured until 500 B.C. ; the third, "The [xxxiv] Age of the Seven States," until 255 B.C. Starting under wise rulers, it gradually sank through others less competent until by 770 B.C. it was little more than a name. During the "Age of Feudalism," the numerous States were constantly at war, but eventually the strongest of them united in a group called the "Seven Masculine Powers" under the shadowy suzerainty of Chou. Although, from the political point of view, this period was full of unrest and gloom, from the intellectual it was exceedingly brilliant and is known as the "Age of Philosophers." The most famous names among the many teachers of the time are those of Lao Tzŭ, the founder of Taoism, and Confucius. To these men, China owes the two great schools of thought upon which her social system rests.

The "Age of the Seven States" (Masculine Powers) ended when Ch'in, one of their number, overcame and absorbed the rest. Its prince adopted the title of Shih Huang Ti, or "First Supreme Ruler," thus placing himself on an equality with Heaven. Is it to be wondered at that the scholars demurred? The literary class were in perpetual opposition to the Emperor, who finally lost patience with them altogether and decreed that all books relating to the past should be burnt, and that history should begin with him. This edict was executed with great severity, and many hundreds of the literati were buried alive. It is scarcely surprising, therefore, that the name of Shih Huang [xxxv] Ti is execrated, even to-day, by a nation whose love for the written word amounts to veneration.

Although he held learning of small account, this "First Emperor," to give him his bombastic title, was an enthusiastic promoter of public works, the most important of these being the Great Wall, which has served as an age-long bulwark against the nomadic tribes of Mongolia and Central Asia. These tribes were a terror to China for centuries. They were always raiding the border country, and threatening a descent on the fertile fields beyond the mountains. The history of China is one long struggle to keep from being overrun by these tribes. There is an exact analogy to this state of affairs in the case of Roman Britain, and the perpetual vigilance it was obliged to exercise to keep out the Picts.

Shih Huang Ti based his power on fear, and it is a curious commentary upon the fact that the Ch'in Dynasty came to an end in 206 B.C ., shortly after his death, and only a scant half-century after he had founded it.

A few years of struggle, during which no Son of Heaven occupied the Dragon Throne, succeeded the fall of the Ch'in Dynasty; then a certain Liu Pang, an inconsiderable town officer, proved strong enough to seize what was no one's possession and made himself Emperor, thereby founding the Han Dynasty.

The Han is one of the most famous dynasties in Chinese [xxxvi] history. An extraordinary revival of learning took place under the successive Emperors of Han. Tho greatest of them, Wu Ti (140-87 B.C. ), is frequently mentioned by the poets. Learning always follows trade, as has often been demonstrated. During the Han Dynasty, which lasted until A.D. 221, intercourse with all the countries of the Near East became more general than ever before, and innumerable caravans wended their slow way across the trade routes of Central Asia. Expeditions against the harassing barbarians were undertaken, and for a time their power was scotched. It was under the Han that Buddhism was introduced from India, but deeply as this has influenced the life and thought of the Middle Kingdom, I am inclined to think that the importance of this influence has been exaggerated.

This period, and those immediately preceding it, form the poetic background of China. The ancient States, constantly referred to in the poems, do not correspond to the modern provinces. In order, therefore, to make their geographical positions clear, a map has been appended to this volume in which the modern names of the provinces and cities are printed in black ink and the ancient names in red. As these States did not all exist at the same moment, it is impossible to define their exact boundaries, but how strongly they were impressed upon the popular mind can be seen by the fact that, although they were merged into [xxxvii] the Chinese Empire during the reign of Shih Huang Ti, literature continued to speak of them by their old names and, even to-day, writers often refer to them as though they were still separate entities. There were many States, but only those are given in the map which are alluded to in the poems published in this book. The names of a few of the old cities are also given, such as Chin Ling, the "Golden Mound" or "Sepulchre," and Ch'ang An, "Eternal Peace," for so many centuries the capital. Its present name is Hsi An-fu, and it was here that the Manchu Court took refuge during the Boxer madness of 1900.

Little more of Chinese history need be told. Following the Han, several dynasties held sway; there were divisions between the North and South and much shifting of power. At length, in A.D. 618, Li Shih-min established the T'ang Dynasty by placing his father on the throne, and the T'ang brought law and order to the suffering country.

This period is often called the Golden Age of Chinese Learning. The literary examinations introduced under the Han were perfected, poets and painters were encouraged, and strangers flocked to the Court at Ch'ang An. The reign of Ming Huang ( A.D. 712-756), the "Brilliant Emperor," was the culmination of this remarkable era. China's three greatest poets, Li T'ai-po, Tu Fu, and Po Chü-i, all lived during his long reign of forty-five years. Auspiciously as this reign had begun, however, it ended sadly. The Em [xxxviii] peror, more amiable than perspicacious, fell into the toils of his favourite concubine, the lovely Yang Kuei-fei, to whom he was slavishly devoted. The account of their love story—a theme celebrated by poets, painters, and playwrights—will be found in the note to "Songs to the Peonies." A rebellion which broke out was crushed, but the soldiers refused to defend the cause of the Emperor until he had issued an order for the execution of Yang Kuei-fei, whom they believed to be responsible for the trouble. Broken-hearted, the Emperor complied, but from this date the glory of the dynasty was dimmed. Throughout its waning years, the shadow of the dreaded Tartars grew blacker and blacker, and finally, in A.D. 907, the T'ang Dynasty fell.

Later history need not concern us here, since most of the poems in this book were written during the T'ang period. Though these poems deal largely with what I have called the historical background, they deal still more largely with the social background and it is, above all, this social background which must be understood.

If the Emperor were the "Son of Heaven," he administered his Empire with the help of very human persons, the various officials, and these officials owed their positions, great and small, partly to the Emperor's attitude, it is true, but in far greater degree to their prowess in the literary examinations. An official of the first rank might owe his [xxxix] preferment to the Emperor's beneficence; but to reach an altitude where this beneficence could operate, he had to climb through all the lower grades, and this could only be done by successfully passing all the examinations, one after the other. The curious thing is that these examinations were purely literary. They consisted not only in knowing thoroughly the classics of the past, but in being able to recite long passages from them by heart, and with this was included the ability to write one's self, not merely in prose, but in poetry. Every one in office had to be, perforce, a poet. No one could hope to be the mayor of a town or the governor of a province unless he had attained a high proficiency in the art of poetry. This is brought strikingly home to us by the fact that one of the chief pastimes of educated men was to meet together for the purpose of playing various games all of which turned on the writing of verse.

The examinations which brought about this strange state of things were four. The first, which conferred the degree of Hsiu Ts'ai , "Flowering Talent," could be competed for only by those who had already passed two minor examinations, one in their district, and one in the department in which this district was situated. The Hsiu Ts'ai examinations were held twice every three years in the provincial capitals. There were various grades of the "Flowering Talent" degree, which is often translated as Bachelor of [xl] Arts, some of which could be bestowed through favour or acquired by purchase. The holders of it were entitled to wear a dress of blue silk, and in Chinese novels the hero is often spoken of as wearing this colour, by which readers are to understand that he is a clever young man already on the way to preferment.

The second degree, that of Ch'ü Jên , "Promoted Man," was obtained by passing the examinations which took place every third year in all the provincial capitals simultaneously. This degree enabled its recipients to hold office, but positions were not always to hand, and frequently "Promoted Men" had to wait long before being appointed to a post; also, the offices open to them were of the lesser grades, those who aspired to a higher rank had a farther road to travel. The dress which went with this degree was also of silk, but of a darker shade than that worn by "bachelors."

The third examination for the Chin Shih , or "Entered Scholar," degree was also held triennially, but at the national capital, and only those among the Ch'ü Jên who had not already taken office were eligible. The men so fortunate as to pass were allowed to place a tablet over the doors of their houses, and their particular dress was of violet silk.

The fourth, which really conferred an office rather than a degree, was bestowed on men who competed in a special examination held once in three years in the Emperor's Palace. Those who were successful in this last examination [xli] became automatically Han Lin , or members of the Imperial Academy, which, in the picturesque phraseology of China, was called the "Forest of Pencils." A member of the Academy held his position, a salaried one, for life, and the highest officials of the Empire were chosen from these Academicians.

This elaboration of degrees was only arrived at gradually. During the T'ang Dynasty, all the examinations were held at Ch'ang An. These four degrees of learning have often been translated as Bachelor of Arts, Master of Arts, Doctor of Literature, and Academician. The analogy is so far from close, however, that most modern sinologues prefer to render them indiscriminately, according to context, as student, scholar, and official.

By means of this remarkable system, which threw open the road to advancement to every man in the country capable of availing himself of it, new blood was continually brought to the top, as all who passed the various degrees became officials, expectant or in being, and of higher or lower grade according to the Chinese measure of ability. Military degrees corresponding to the civil were given; but, as these called for merely physical display, they were not highly esteemed.

Since only a few of the candidates for office passed the examinations successfully, a small army of highly educated men was dispersed throughout the country every three [xlii] years. In the towns and villages they were regarded with the reverence universally paid to learning by the Chinese, and many became teachers to the rising generation in whom they cultivated a great respect for literature in general and poetry in particular.

The holders of degrees, on the other hand, entered at once upon a career as administrators. Prevented by an inexorable law—a law designed to make nepotism impossible—from holding office in their own province, they were constantly shifted from one part of the country to another, and this is a chief reason for the many poems of farewell that were written. The great desire of all officials was to remain at, or near, the Court, where the most brilliant brains of the Empire were assembled. As may be easily imagined, the intrigues and machinations employed to attain this end were many, with the result that deserving men often found themselves banished to posts on the desolate outskirts of the country where, far from congenial intercourse, they suffered a mental exile of the most complete description. Innumerable poems dealing with this sad state are found in all Chinese anthologies.

There were nine ranks of nobility. The higher officials took the rank of their various and succeeding offices, others were ennobled for signal services performed. These titles were not hereditary in the ordinary sense, but backwards, if I can so express it. The dead ancestors of a nobleman [xliii] were accorded his rank, whatever had been theirs in life, but his sons and their descendants had only such titles as they themselves might earn.

The desire to bask in the rays of the Imperial Sun was shared by ambitious fathers who longed to have their daughters appear before the Emperor, and possibly make the fortune of the family by captivating the Imperial glance. This led to the most beautiful and talented young girls being sent to the Palace, where they often lived and died without ever being summoned before the Son of Heaven. Although numberless tragic poems have been written by these unfortunate ladies, many charming romances did actually take place, made possible by the custom of periodically dispersing the superfluous Palace women and marrying them to suitable husbands.

In striking contrast to the unfortunates who dragged out a purposeless life of idleness, was the lot of the beauty who had the good fortune to capture the Imperial fancy, and who, through her influence over the Dragon Throne, virtually ruled the Middle Kingdom. No extravagancies were too great for these exquisite creatures, and many dynasties have fallen through popular revolt against the excesses of Imperial concubines.

It would be quite erroneous to suppose, however, that the Emperor's life was entirely given up to pleasure and gaiety, or that it was chiefly passed in the beautiful seclu [xliv] sion of the Imperial gardens. The poems, it is true, generally allude to these moments, but the cares of state were many, and every day, at sunrise, officials assembled in the Audience Hall to make their reports to the Emperor. Moreover, Court ceremonials were extremely solemn occasions, carried out with the utmost dignity.

As life at Court centred about the persons of the Emperor and Empress, so life in the homes of the people centred about the elders of the family. The men of wealthy families were usually of official rank, and led a life in touch with the outer world, a life of social intercourse with other men in which friendship played an all-engrossing part. This characteristic of Chinese life is one of the most striking features of the poetic background. Love poems from men to women are so rare as to be almost non-existent (striking exceptions do occur, however, several of which are translated here), but poems of grief written at parting from "the man one loves" are innumerable, and to sit with one's friends, drinking wine and reciting verses, making music or playing chess, were favourite amusements throughout the T'ang period.

Wine-drinking was general, no pleasure gathering being complete without it. The wine of China was usually made from fermented grains, but wines from grapes, plums, pears, and other fruits were also manufactured. It was carefully heated and served in tall flagons somewhat resembling our [xlv] coffee-pots, and was drunk out of tiny little cups no bigger than liqueur glasses. These cups, which were never of glass, were made of various metals, of lacquered or carved wood, of semi-precious stones such as jade, or agate, or carnelian; porcelain, the usual material for wine-cups to-day, not having yet been invented. Custom demanded that each thimbleful be tossed off at a gulp, and many were consumed before a feeling of exhilaration could be experienced. That there was a good deal of real drunkenness, we cannot doubt, but not to the extent that is generally supposed. From the character of the men and the lives they led, it is fairly clear that most of the drinking kept within reasonable bounds. Unfortunately, in translation, the quantity imbibed at these wine-parties becomes greatly exaggerated. That wine was drunk, not merely for its taste, but as a heightener of sensation, is evident; but the "three hundred cups" so often mentioned bear no such significance as might at first appear when the size of the cups is taken into account. Undoubtedly, also, we must regard this exact number as a genial hyperbole.

If husbands and sons could enjoy the excitement of travel, the spur of famous scenery, the gaieties of Court, and the pleasures of social intercourse, wives and daughters were obliged to find their occupations within the Kuei or "Women's Apartments," which included the gardens set apart for their use. The ruling spirit of the Kuei was the [xlvi] mother-in-law; and the wife of the master of the house, although she was the mother of his sons and the director of the daughters-in-law, did not reach the fulness of her power until her husband's mother had died.

The chief duty of a young wife was attendance upon her mother-in-law. With the first grey streak of daylight, she rose from her immense lacquer bed, so large as to be almost an anteroom, and, having dressed, took the old lady her tea. She then returned to her own apartment to breakfast with her husband and await the summons to attend her mother-in-law's toilet, a most solemn function, and the breakfast which followed. These duties accomplished, she was free to occupy herself as she pleased. Calligraphy, painting, writing poems and essays, were popular pursuits, and many hours were spent at the embroidery frame or in making music.

Chinese poetry is full of references to the toilet, to the intricate hair-dressing, the "moth-antennæ eyebrows," the painting of faces, and all this was done in front of a mirror standing on a little rack placed on the toilet-table. A lady, writing to her absent husband, mourns that she has no heart to "make the cloud head-dress," or writes, "looking down upon my mirror in order to apply the powder and paint, I desire to keep back the tears. I fear that the people in the house will know my grief. I am ashamed."

In spite of the fact that they had never laid eyes on the [xlvii] men they were to marry before the wedding-day, these young women seem to have depended upon the companionship of their husbands to a most touching extent. The occupations of the day were carried on in the Kuei ; but, when evening came, the husband and wife often read and studied the classics together. A line from a well-known poem says, "The red sleeve replenishes the incense, at night, studying books," and the picture it calls up is that of a young man and woman in the typical surroundings of a Chinese home of the educated class. Red was the colour worn by very young women, whether married or not; as the years advanced, this was changed for soft blues and mauves, and later still for blacks, greys, or dull greens. A line such as "tears soak my dress of coarse, red silk" instantly suggests a young woman in deep grief.

The children studied every day with teachers; the sons and daughters of old servants who had, according to custom, taken the family surname, receiving the same advantages as those of the master. These last were, in all respects, brought up as children of the house, the only distinction being that whereas the master's own children sat "above" the table, facing South, the children of the servants sat "below," facing North. A more forcible reminder of their real status appeared later in life, since they were debarred from competing in the official examinations unless they left the household in which they had grown up and relinquished the family surname taken by their fathers. A curious habit among families, which extended even to groups of friends, was the designation by numbers according to age, a man being familiarly known as Yung Seven or T'sui Fifteen. It will be noticed that such designations often occur in the poems.

Only four classes of persons were recognized as being of importance to society and these were rated in the following order: scholars, agriculturalists, labourers, and traders—officials, of course, coming under the generic name of scholars. Soldiers, actors, barbers, etc., were considered a lower order of beings entirely and, as such, properly despised.

China, essentially an agricultural country, was economically self-sufficient, producing everything needed by her population. The agriculturalist was, therefore, the very backbone of the state.

In rendering Chinese poetry, the translator must constantly keep in mind the fact that the architectural background differs from that of every other country, and that our language does not possess terms which adequately describe it.

Apart from the humble cottages of the very poor, all dwelling-houses, or chia , are constructed on the same general plan. They consist of a series of one-story buildings divided by courtyards, which, in the houses of the [xlix] well-to-do, are connected by covered passages running along the sides of each court. A house is cut up into chien , or divisions, the number, within limits, being determined by the wealth and position of the owners. The homes of the people, both rich and poor, are arranged in three or five chien ; official residences are of seven chien ; Imperial palaces of nine. Each of these chien consists of several buildings, the number of which vary considerably, more buildings being added as the family grows by the marriage of the sons who, with their wives and children, are supposed to live in patriarchal fashion in their father's house. If officials sometimes carried their families with them to the towns where they were stationed, there were other posts so distant or so desolate as to make it practically impossible to take women to them. In these cases, the families remained behind under the paternal roof.

How a house was arranged can be seen in the plan at the end of this book. Doors lead to the garden from the study, the guest-room, and the Women's Apartments. These are made in an endless diversity of shapes and add greatly to the picturesqueness of house and grounds. Those through which a number of people are to pass to and fro are often large circles, while smaller and more intimate doors are cut to the outlines of fans, leaves, or flower vases. In addition to the doors, blank spaces of wall are often broken by openings at the height of a window, such openings being [l] most fantastic and filled with intricately designed latticework.

I have already spoken of the Kuei or Women's Apartments. In poetry, this part of the chia is alluded to in a highly figurative manner. The windows are "gold" or "jade" windows; the door by which it is approached is the Lan Kuei , or "Orchid Door." Indeed, the sweet-scented little epidendrum called by the Chinese, lan , is continually used to suggest the Kuei and its inmates.

Besides the house proper, there are numerous structures erected in gardens, for the Chinese spend much of their time in their gardens. No nation is more passionately fond of nature, whether in its grander aspects, or in the charming arrangements of potted flowers which take the place of our borders in their pleasure grounds. Among these outdoor buildings none is more difficult to describe than the lou , since we have nothing which exactly corresponds to it. Lous appear again and again in Chinese poetry, but just what to call them in English is a puzzle. They are neither summer-houses, nor pavilions, nor cupolas, but a little of all three. Always of more than one story, they are employed for differing purposes; for instance, the fo lou on the plan is an upper chamber where Buddhist images are kept. The lou generally referred to in poetry, however, is really a "pleasure-house-in-the-air," used as the Italians use their belvederes. Here the inmates of the house sit and look [li] down upon the garden or over the surrounding country, or watch "the sun disappear in the long grass at the edge of the horizon" or "the moon rise like a golden hook."

Another erection foreign to Western architecture is the t'ai , or terrace. In early days, there were many kinds of t'ai , ranging from the small, square, uncovered stage still seen in private gardens and called yüeh t'ai , "moon terrace," to immense structures like high, long, open platforms, built by Emperors and officials for various reasons. Many of these last were famous; I have given the histories of several of them in the notes illustrating the poems, at the end of the book.

It will be observed that I have said practically nothing about religion. The reason is partly that the three principal religions practised by the Chinese are either so well known, as Buddhism, for example, or so difficult to describe, as Taoism and the ancient religion of China now merged in the teachings of Confucius; partly that none of them could be profitably compressed into the scope of this introduction; but chiefly because the subject of religion, in the poems here translated, is generally referred to in its superstitious aspects alone. The superstitions which have grown up about Taoism particularly are innumerable. I have dealt with a number of these in the notes to the poems in which they appear. Certain supernatural personages, without a [lii] knowledge of whom much of the poetry would be unintelligible, I have set down in the following list:

Immortals who live in the Taoist Paradises. Human beings may attain " Hsien-ship ," or Immortality, by living a life of contemplation in the hills. In translating the term, we have used the word "Immortals."

Beneficent beings who inhabit the higher regions. They are kept extremely busy attending to their duties as tutelary deities of the roads, hills, rivers, etc., and it is also their function to intervene and rescue deserving people from the attacks of their enemies.

A proportion of the souls of the departed who inhabit the "World of Shades," a region resembling this world, which is the "World of Light," in every particular, with the important exception that it has no sunshine. Kindly kuei are known, but the influence generally suggested is an evil one. They may only return to the World of Light between sunset and sunrise, except upon the fifth [liii] day of the Fifth Month (June), when they are free to come during the time known as the "hour of the horse," from eleven A.M. to one P.M.

A class of fierce demons who live in the wild regions of the Southwest and delight in eating the flesh of human beings.

There are also supernatural creatures whose names carry a symbolical meaning. A few of them are:

A composite animal, somewhat resembling the fabulous unicorn, whose arrival is a good omen. He appears when sages are born.

A symbol of the forces of Heaven, also the emblem of Imperial power. Continually referred to in poetry as the steed which transports a philosopher who has attained Immortality to his home in the Western Paradise.

Fêng Huang.

A glorious bird, symbol of the Empress, therefore often associated with the dragon. The conception of this bird is probably based on the Argus pheasant. It is described as possessing every grace [liv] and beauty. A Chinese author, quoted by F. W. Williams in "The Middle Kingdom," writes: "It resembles a wild swan before and a unicorn behind; it has the throat of a swallow, the bill of a cock, the neck of a snake, the tail of a fish, the forehead of a crane, the crown of a mandarin drake, the stripes of a dragon, and the vaulted back of a tortoise. The feathers have five colours which are named after the five cardinal virtues, and it is five cubits in height; the tail is graduated like the pipes of a gourd-organ, and its song resembles the music of the instrument, having five modulations." Properly speaking, the female is Fêng , the male Huang , but the two words are usually given in combination to denote the species. Some one, probably in desperation, once translated the combined words as "phœnix," and this term has been employed ever since. It conveys, however, an entirely wrong impression of the creature. To Western readers, the word "phœnix" suggests a bird which, being consumed by fire, [lv] rises in a new birth from its own ashes. The Fêng Huang has no such power, it is no symbol of hope or resurrection, but suggests friendship and affection of all sorts. Miss Lowell and I have translated the name as "crested love-pheasant," which seems to us to convey a better idea of the beautiful Fêng Huang , the bird which brings happiness.

A supernatural bird sometimes confused with the above. It is a sacred creature, connected with fire, and a symbol of love and passion, of the relation between men and women.

The "paired-wings bird," described in Chinese books as having but one wing and one eye, for which reason two must unite for either of them to fly. It is often referred to as suggesting undying affection.

Real birds and animals also have symbolical attributes. I give only three:

Represents longevity, and is employed, as is the dragon, to transport those who [lvi] have attained to Immortality to the Heavens.

The exquisite little mandarin ducks, an unvarying symbol of conjugal fidelity. Li T'ai-po often alludes to them and declares that, rather than be separated, they would "prefer to die ten thousand deaths, and have their gauze-like wings torn to fragments."

Wild Geese.

Symbols of direct purpose, their flight being always in a straight line. As they follow the sun's course, allusions to their departure suggest Spring, to their arrival, Autumn.

A complete list of the trees and plants endowed with symbolical meanings would be almost endless. Those most commonly employed in poetry in a suggestive sense are:

Ch'ang P'u.

A plant growing in the Taoist Paradise and much admired by the Immortals, who are the only beings able to see its purple blossoms. On earth, it is known as the sweet flag, and has the peculiarity of never blossoming. It is hung on the lintels of doors on the fifth day of the [lvii] Fifth Month to ward off the evil influences which may be brought by the kuei on their return to this world during the "hour of the horse."

Riches and prosperity.

Purity. Although it rises from the mud, it is bright and spotless.

Plum-blossom.

Literally "the first," it being the first of the "hundred flowers" to open. It suggests the beginnings of things, and is also one of the "three friends" who do not fear the Winter cold, the other two being the pine and the bamboo.

A small epidendrum, translated in this book as "spear-orchid." It is a symbol for noble men and beautiful, refined women. Confucius compared the Chün Tzŭ , Princely or Superior Man, to this little orchid with its delightful scent. In poetry, it is also used in reference to the Women's Apartments and everything connected with them, suggesting, as it does, the extreme of refinement.

Chrysanthemum.

Fidelity and constancy. In spite of frost, its flowers continue to bloom.

Longevity. This fungus, which grows at the roots of trees, is very durable when dried.

Longevity, immutability, steadfastness.

This plant has as many virtues as it has uses, the principal ones are modesty, protection from defilement, unchangeableness.

A tree whose botanical name is sterculia platanifolia . Its only English name seems to be "umbrella-tree," which has proved so unattractive in its context in the poems that we have left it untranslated. It is a symbol for integrity, high principles, great sensibility. When "Autumn stands," on August seventh, although it is still to all intents and purposes Summer, the wu-t'ung tree drops one leaf. Its wood, which is white, easy to cut, and very light, is the only kind suitable for making that intimate instrument which quickly betrays the least emotion of the person playing upon it—the ch'in , or table-lute.

A prostitute, or any very frivolous person. Concubines writing to their lords often refer to themselves under this figure, in the same spirit of self-depreciation which prompts them to employ the euphemism, "Unworthy One," instead of the personal pronoun. Because of its lightness and pliability, it conveys also the idea of extreme vitality.

Peach-blossom.

Beautiful women and ill-success in life. The first suggestion, on account of the exquisite colour of the flower; the second, because of its perishability.

Peach-tree.

Longevity. This fruit is supposed to ripen once every three thousand years on the trees of Paradise, and those who eat of this celestial species never die.

Utility. Also suggests a peaceful hamlet. Its wood is used in the making of bows and the kind of temple-drums called mo yü —wooden fish. Its leaves feed the silk-worms.

Sadness and grief. It is symbolical of a heart which is not "flat" or "level," as [lx] the Chinese say, not open or care-free, but of one which is "tightly rolled." The sound of rain on its leaves is very mournful, therefore an allusion to the plantain always means sorrow. Planted outside windows already glazed with silk, its heavy green leaves soften the glaring light of Summer, and it is often used for this purpose.

Nothing has been more of a stumbling-block to translators than the fact that the Chinese year—which is strictly lunar, with an intercalary month added at certain intervals—begins a month later than ours; or, to be more exact, it is calculated from the first new moon after the sun enters Aquarius, which brings the New Year at varying times from the end of January to the middle of February. For translation purposes, however, it is safe to count the Chinese months as always one later by our calendar than the number given would seem to imply. By this calculation the "First Month" is February, and so on throughout the year.

The day is divided into twelve periods of two hours each beginning at eleven P.M. and each of these periods is called by the name of an animal—horse, deer, snake, bat, etc. As these names are not duplicated, the use of them tells at [lxi] once whether the hour is day or night. Ancient China's method of telling time was by means of slow and evenly burning sticks made of a composition of clay and sawdust, or by the clepsydra, or water-clock. Water-clocks are mentioned several times in these poems.

So much for what I have called the backgrounds of Chinese poetry. I must now speak of that poetry itself, and of Miss Lowell's and my method of translating it.

Chinese prosody is a very difficult thing for an Occidental to understand. Chinese is a monosyllabic language, and this reduces the word-sounds so considerably that speech would be almost impossible were it not for the invention of tones by which the same sound can be made to do the duty of four in the Mandarin dialect, five in the Nankingese, eight in the Cantonese, etc., a different tone inflection totally changing the meaning of a word. Only two chief tones are used in poetry, the "level" and the "oblique," but the oblique tone is subdivided into three, which makes four different inflections possible to every sound. Of course, like English and other languages, the same word may have several meanings, and in Chinese these meanings are bewilderingly many; the only possible way of determining which one is correct is by its context. These tones constitute, at the outset, the principal difference which divides the technique of Chinese poetry from our own. Another is to be found in the fact that nothing approaching our metri [lxii] cal foot is possible in a tongue which knows only single syllables. Rhyme does exist, but there are only a little over a hundred rhymes, as tone inflection does not change a word in that particular. Such a paucity of rhyme would seriously affect the richness of any poetry, if again the Chinese had not overcome this lingual defect by the employment of a juxtaposing pattern made up of their four poetic tones. And these tones come to the rescue once more when we consider the question of rhythm. Monosyllables in themselves always produce a staccato effect, which tends to make all rhythm composed of them monotonous, if, indeed, it does not destroy it altogether. The tones cause what I may call a psychological change in the time-length of these monosyllables, which change not only makes true rhythm possible, but allows marked varieties of the basic beat.

One of the chief differences between poetry and prose is that poetry must have a more evident pattern. The pattern of Chinese poetry is formed out of three elements: line, rhyme, and tone.

The Chinese attitude toward line is almost identical with that of the French. French prosody counts every syllable as a foot, and a line is made up of so many counted feet. If any of my readers has ever read French alexandrines aloud to a Frenchman, read them as we should read English poetry, seeking to bring out the musical stress, he will remember the look of sad surprise which crept over his [lxiii] hearer's face. Not so was this verse constructed; not so is it to be read. The number of syllables to a line is counted, that is the secret of French classic poetry; the number of syllables is counted in Chinese. But—and we come to a divergence—this method of counting does, in French practice, often do away with the rhythm so delightful to an English ear; in Chinese, no such violence occurs, as each syllable is a word and no collection of such words can fall into a metric pulse as French words can, and, in their Chansons , are permitted to do.

The Chinese line pattern is, then, one of counted words, and these counted words are never less than three, nor more than seven, in regular verse; irregular is a different matter, as I shall explain shortly. Five and seven word lines are cut by a cæsura, which comes after the second word in a five-word line, and after the fourth in a seven-word line.

Rhyme is used exactly as we use it, at the ends of lines. Internal rhyming is common, however, in a type of poem called a " fu ," which I shall deal with when I come to the particular kinds of verse.

Tone is everywhere, obviously, and is employed, not arbitrarily, but woven into a pattern of its own which again is in a more or less loose relation to rhyme. By itself, the tone-pattern alternates in a peculiar manner in each line, the last line of a stanza conforming to the order of tones in the first, the intervening lines varying methodi [lxiv] cally. I have before me a poem in which the tone-pattern is alike in lines one, four, and eight, of an eight-line stanza, as are lines two and six, and lines three and seven, while line five is the exact opposite of lines two and six. In the second stanza of the same poem, the pattern is kept, but adversely; the tones do not follow the same order, but conform in similarity of grouping. I use this example merely to show what is meant by tone-pattern. It will serve to illustrate how much diversity and richness this tone-chiming is capable of bringing to Chinese poetry.

Words which rhyme must be in the same tone in regular verse, and unrhymed lines must end on an oblique tone if the rhyme-tone is level, and vice versa . The level tone is preferred for rhyme.

In the early Chinese poetry, called Ku-shih (Old Poems), the tones were practically disregarded. But in the Lü-shih (Regulated Poems) the rules regarding them are very strict. The lü-shih are supposed to date from the beginning of the T'ang Dynasty. A lü-shih poem proper should be of eight lines, though this is often extended to sixteen, but it must be in either the five-word line, or the seven-word line, metre. The poets of the T'ang Dynasty, however, were by no means the slaves of lü-shih ; they went their own way, as good poets always do, conforming when it pleased them and disregarding when they chose. It depended on the character of the poet. Tu Fu was renowned for his careful [lxv] versification; Li T'ai-po, on the other hand, not infrequently rebelled and made his own rules. In his "Drinking Song," which is in seven-word lines, he suddenly dashes in two three-word lines, a proceeding which must have been greatly upsetting to the purists. It is amusing to note that his "Taking Leave of Tu Fu" is in the strictest possible form, which is at once a tribute and a poking of fun at his great friend and contemporary.

Regular poems of more than sixteen lines are called p'ai lu , and these may run to any length; Tu Fu carried them to forty, eighty, and even to two hundred lines. Another form, always translated as "short-stop," cuts the eight-line poem in two. In theory, the short-stop holds the same relation to the eight-line poem that the Japanese hokku does to the tanka , although of course it preceded the hokku by many centuries. It is supposed to suggest rather than to state, being considered as an eight-line poem with its end in the air. In suggestion, however, the later Japanese form far outdoes it.

So called "irregular verse" follows the writer's inclination within the natural limits of all Chinese prosody.

A tzŭ may be taken to mean a lyric, if we use that term, not in its dictionary sense, but as all modern poets employ it. It may vary its line length, but must keep the same variation in all the stanzas.

Perhaps the most interesting form to modern students is [lxvi] the fu , in which the construction is almost identical with that of "polyphonic prose." The lines are so irregular in length that the poem might be mistaken for prose, had we not a corresponding form to guide us. The rhymes appear when and where they will, in the middle of the lines or at the end, and sometimes there are two or more together. I have been told that Persia has, or had, an analogous form, and if so modern an invention as "polyphonic prose" derives, however unconsciously, from two such ancient countries as China and Persia, the fact is, at least, interesting.

The earliest examples of Chinese poetry which have come down to us are a collection of rhymed ballads in various metres, of which the most usual is four words to a line. They are simple, straightforward pieces, often of a strange poignance, and always reflecting the quiet, peaceful habits of a people engaged in agriculture. The oldest were probably composed about 2000 B.C. and the others at varying times from then until the Sixth Century B.C. , when Confucius gathered them into the volume known as the "Book of Odes." Two of these odes are translated in this book. The next epoch in the advance of poetry-making was introduced by Ch'ü Yüan (312-295 B.C. ), a famous statesman and poet, who wrote an excitable, irregular style in which the primitive technical rules were disregarded, their place being taken by exigencies of emotion and idea. We are wont to regard a poetical technique determined by [lxvii] feeling alone as a very modern innovation, and it is interesting to note that the method is, on the contrary, as old as the hills. These rhapsodical allegories culminated in a poem entitled "Li Sao," or "Falling into Trouble," which is one of the most famous of ancient Chinese poems. A further development took place under the Western Han (206 B.C. - A.D. 25), when Su Wu invented the five-character poem, ku fêng ; these poems were in Old Style, but had five words to a line. It is during this same period that poems with seven words to a line appeared. Legend has it that they were first composed by the Emperor Wu of Han, and that he hit upon the form on an occasion when he and his Ministers were drinking wine and capping verses at a feast on the White Beam Terrace. Finally, under the Empress Wu Hou, early in the T'ang Dynasty, the lü-shih , or "poems according to law," became the standard. It will be seen that the lü-shih found the five and seven word lines already in being and had merely to standardize them. The important gift which the lü-shih brought to Chinese prosody was its insistence on tone.

The great period of Chinese poetry was during the T'ang Dynasty. Then lived the three famous poets, Li T'ai-po, Tu Fu, and Po Chü-i. Space forbids me to give the biographies of all the poets whose work is included in this volume, but as Li T'ai-po and Tu Fu, between them, take up more than half the book, a short account of the princi [lxviii] pal events of their lives seems necessary. I shall take them in the order of the number of their poems printed in this collection, which also, as a matter of fact, happens to be chronological.

I have already stated in the first part of this Introduction the reasons which determined me to give so large a space to Li T'ai-po. English writers on Chinese literature are fond of announcing that Li T'ai-po is China's greatest poet; the Chinese themselves, however, award this place to Tu Fu. We may put it that Li T'ai-po was the people's poet, and Tu Fu the poet of scholars. As Po Chü-i is represented here by only one poem, no account of his life has been given. A short biography of him may be found in Mr. Waley's "A Hundred and Seventy Chinese Poems."

It is permitted to very few to live in the hearts of their countrymen as Li T'ai-po has lived in the hearts of the Chinese. To-day, twelve hundred and twenty years after his birth, his memory and his fame are fresh, his poems are universally recited, his personality is familiar on the stage: in fact, to use the words of a Chinese scholar, "It may be said that there is no one in the People's Country who does not know the name of Li T'ai-po." Many legends are told of his birth, his life, his death, and he is now numbered among the Hsien (Immortals) who inhabit the Western Paradise.

Li T'ai-po was born A.D. 701, of well-to-do parents named Li, who lived in the Village of the Green Lotus in Szechwan. [lxix] He is reported to have been far more brilliant than ordinary children. When he was only five years old, he read books that other boys read at ten; at ten, he could recite the "Classics" aloud and had read the "Book of the Hundred Sages." Doubtless this precocity was due to the fact that his birth was presided over by the "Metal Star," which we know as Venus. His mother dreamt that she had conceived him under the influence of this luminary, and called him T'ai-po, "Great Whiteness," a popular name for the planet.

In spite of his learning, he was no Shu Tai Tzŭ (Book Idiot) as the Chinese say, but, on the contrary, grew up a strong young fellow, impetuous to a fault, with a lively, enthusiastic nature. He was extremely fond of sword-play, and constantly made use of his skill in it to right the wrongs of his friends. However worthy his causes may have been, this propensity got him into a serious scrape. In the excitement of one of these encounters, he killed several people, and was forthwith obliged to fly from his native village. The situation was an awkward one, but the young man disguised himself as a servant and entered the employ of a minor official. This gentleman was possessed of literary ambitions and a somewhat halting talent; still we can hardly wonder that he was not pleased when his servant ended a poem in which he was hopelessly floundering with lines far better than he could make. After this, and one or two similar experiences, Li T'ai-po found it advisa [lxx] ble to relinquish his job and depart from his master's house.

His next step was to join a scholar who disguised his real name under the pseudonym of "Stern Son of the East." The couple travelled together to the beautiful Min Mountains, where they lived in retirement for five years as teacher and pupil. This period, passed in reading, writing, discussing literature, and soaking in the really marvellous scenery, greatly influenced the poet's future life, and imbued him with that passionate love for nature so apparent in his work.

At the age of twenty-five, he separated from his teacher and left the mountains, going home to his native village for a time. But the love of travel was inherent in him, nowhere could hold him for long, and he soon started off on a sight-seeing trip to all those places in the Empire famous for their beauty. This time he travelled as the position of his parents warranted, and even a little beyond it. He had a retinue of servants, and spent money lavishly. This open-handedness is one of the fine traits of his character. Needy scholars and men of talent never appealed to him in vain; during a year at Yangchow, he is reported to have spent three hundred thousand ounces of silver in charity.

From Yangchow he journeyed to the province of Hupeh ("North of the Lake") where, in the district of the "Dreary Clouds," he stayed at the house of a family named [lxxi] Hsü, which visit resulted in his marriage with one of the daughters. Li T'ai-po lived in Hupeh for some years—he himself says three—then his hunger for travel reasserted itself and he was off again. After some years of wandering, while visiting a magistrate in Shantung, an incident occurred which had far-reaching consequences. A prisoner was about to be flogged. Li T'ai-po, who was passing, glanced at the man, and, happening to be possessed of a shrewd insight into character, realized at once that here was an unusual person. He secured the man's release, and twenty-five years later this action bore fruit as the sequel will show. The freed prisoner was Kuo Tzŭ-i, who became one of China's most powerful generals and the saviour of the T'ang Dynasty.

It will be noticed that nothing has been said of the poet taking any examinations, and for the excellent reason that he never thought it worth while to present himself as a candidate. The simple fact appears to be that geniuses often do not seem to find necessary what other men consider of supreme importance. Presumably, also, he had no particular desire for an official life. The gifts of Heaven go by favour and the gifts of man are strangely apt to do the same thing, in spite of the excellent rules devised to order them. Li T'ai-po's career owed nothing to either the lack of official degrees or official interest. What he achieved, he owed to himself; what he failed in came from the same source.

About this time, the poet and a few congenial friends formed the coterie of "The Six Idlers of the Bamboo Brook." They retired to the Ch'u Lai Mountain and spent their time in drinking, reciting poems, writing beautiful characters, and playing on the table-lute. It must be admitted that Li T'ai-po was an inveterate and inordinate drinker, and far more often than was wise in the state called by his countrymen "great drunk." To this propensity he was indebted for all his ill fortune, as it was to his poetic genius that he owed all his good.

So the years passed until, when he was forty-two, he met the Taoist priest, Wu Yün. They immediately became intimate, and on Wu Yün's being called to the capital, Li T'ai-po accompanied him. Wu Yün took occasion to tell the Emperor of his friend's extraordinary talent. The Emperor was interested, the poet was sent for, and, introduced by Ho Chih-chang, was received by the Son of Heaven in the Golden Bells Hall.

The native accounts of this meeting state that "in his discourses upon the affairs of the Empire, the words rushed from his mouth like a mountain torrent." Ming Huang, who was enchanted, ordered food to be brought and helped the poet himself.

So Li T'ai-po became attached to the Court and was made an honorary member of the "Forest of Pencils." He was practically the Emperor's secretary and wrote the [lxxiii] Emperor's edicts, but this was by the way—his real duty was simply to write what he chose and when, and recite these poems at any moment that it pleased the Emperor to call upon him to do so.

Li T'ai-po, with his love of wine and good-fellowship, was well suited for the life of the gay and dissipated Court of Ming Huang, then completely under the influence of the beautiful concubine, Yang Kuei-fei. Conspicuous among the Emperor's entourage was Ho Chih-chang, a famous statesman, poet, and calligraphist, who, on reading Li T'ai-po's poetry, is said to have sighed deeply and exclaimed: "This is not the work of a human being, but of a Tsê Hsien (Banished Immortal)." To understand fully the significance of this epithet, it must be realized that mortals who have already attained Immortality, but who have committed some fault, may be banished from Paradise to expiate their sin on earth.

For about two years, Li T'ai-po led the life of supreme favourite in the most brilliant Court in the world. The fact that when sent for to compose or recite verses he was not unapt to be drunk was of no particular importance since, after being summarily revived with a dash of cold water, he could always write or chant with his accustomed verve and dexterity. His influence over the Emperor became so great that it roused the jealousy, and eventually the hatred, of Kao Li-shih, the Chief Eunuch, who, until [lxxiv] then, had virtually ruled his Imperial master. On one occasion, when Li T'ai-po was more than usually incapacitated, the Emperor ordered Kao to take off the poet's shoes. This was too much, and from that moment the eunuch's malignity became an active intriguing to bring about his rival's downfall. He found the opportunity he needed in the vanity of Yang Kuei-fei. Persuading this lady that Li T'ai-po's "Songs to the Peonies" contained a veiled insult directed at her, he enlisted her anger against the poet and so gained an important ally to his cause. On three separate occasions when Ming Huang wished to confer official rank upon the poet, Yang Kuei-fei interfered and persuaded the Emperor to forego his intention. Li T'ai-po was of too independent a character, and too little of a courtier, to lift a finger to placate his enemies. But the situation became so acute that at last he begged leave to retire from the Court altogether. His request granted, he immediately formed a new group of seven congenial souls and with them departed once more to the mountains. This new association called itself "The Eight Immortals of the Wine-cup."

Although Li T'ai-po had asked for his own dismissal, he had really been forced to ask it, and his banishment from the "Imperial Sun," with all that "Sun" implied, was a blow from which he never recovered. His later poems are full of more or less veiled allusions to his unhappy state.

The next ten years were spent in his favourite occupation of travelling, especially in the provinces of Szechwan, Hunan, and Hupeh.

Meanwhile, political conditions were growing steadily worse. Popular discontent at the excesses of Yang Kuei-fei and her satellite An Lu-shan were increasing, and finally, in A.D. 755, rebellion broke out. I have dealt with this rebellion earlier in this Introduction, and a more detailed account is given in the Notes; I shall, therefore, do no more than mention it here. Sometime during the preceding unrest, Li T'ai-po, weary of moving from place to place, had taken the position of adviser to Li Ling, Prince of Yung. In the wide-spread disorder caused by the rebellion, Li Ling conceived the bold idea of establishing himself South of the Yangtze as Emperor on his own account. Pursuing his purpose, he started at the head of his troops for Nanking. Li T'ai-po strongly disapproved of the Prince's course, a disapproval which affected that headstrong person not at all, and the poet was forced to accompany his master on the march to Nanking.

At Nanking, the Prince's army was defeated by the Imperial troops, and immediately after the disaster Li T'ai-po fled, but was caught, imprisoned, and condemned to death. Now came the sequel to the incident which had taken place long before at Shantung. The Commander of the Imperial forces was no other than Kuo Tzŭ-i, the [lxxvi] former prisoner whose life Li T'ai-po had saved. On learning the sentence passed upon the poet, Kuo Tzŭ-i intervened and threatened to resign his command unless his benefactor were spared. Accordingly Li T'ai-po's sentence was changed to exile and he was released, charged to depart immediately for some great distance where he could do no harm. He set out for Yeh Lang, a desolate spot beyond the "Five Streams," in Kueichow. This was the country of the yao kuai , the man-eating demons; and whether he believed in them or not, the thought of existence in such a gloomy solitude must have filled him with desperation.

He had not gone far, luckily, when a general amnesty was declared, and he was permitted to return and live with his friend and disciple, Lu Yang-ping, in the Lu Mountains near Kiukiang, a place which he dearly loved. Here, in A.D. 762, at the age of sixty-one, he died, bequeathing all his manuscripts to Lu Yang-ping.

The tale of his drowning, repeated by Giles and others, is pure legend, as an authoritative statement of Lu Yang-ping proves. The manuscripts left to his care, and all others he could collect from friends, Lu Yang-ping published in an edition of ten volumes. This edition appeared in the year of the poet's death, and contained the following preface by Lu Yang-ping:

Li T'ai-po's poetry is full of dash and surprise. At his best, there is an extraordinary exhilaration in his work; at his worst, he is merely repetitive. Chinese critics have complained that his subjects are all too apt to be trivial, and that his range is narrow. This is quite true; poems of farewell, deserted ladies sighing for their absent lords, officials consumed by homesickness, pæans of praise for wine—in the aggregate there are too many of these. But how fine they often are! "The Lonely Wife," "Poignant Grief During a Sunny Spring," "After being Separated for a Long Time," such poems are the truth of emotion. Take again his inimitable humour in the two "Drinking Alone in the Moonlight" poems, or "Statement of Resolutions after being Drunk on a Spring Day." Then there are the poems of hyperbolical description such as "The Perils of the Shu Road," "The Northern Flight," and "The Terraced Road of the Two-Edged Sword Mountains." Mountains seem to be in his very blood. Of the sea, on the other hand, he has no such intimate knowledge; he sees it afar, from some height, but always as a thing apart, a distant view. The sea he gazes at; the mountains he treads under foot, their creepers scratch his face, the jutting rocks beside the path bruise his hands. He knows the [lxxviii] straight-up, cutting-into-the-sky look of mountain peaks just above him, and feels, almost bodily, the sheer drop into the angry river tearing its way through a narrow gully below, a river he can see only by leaning dangerously far over the cliff upon which he is standing. There is a curious sense of perpendicularity about these mountain rhapsodies. The vision is strained up for miles, and shot suddenly down for hundreds of feet. The tactile effect of them is astounding; they are not to be read, but experienced. And yet I am loth to say that Li T'ai-po is at his greatest in description, with poems so full of human passion and longing as "The Lonely Wife," and "Poignant Grief During a Sunny Spring," before me. There is no doubt at all that in Li T'ai-po we have one of the world's greatest lyrists.

Great though he was, it cannot be denied that he had serious weaknesses. One was his tendency to write when the mood was not there, and at these moments he was not ashamed to repeat a fancy conceived before on some other occasion. Much of his style he crystallized into a convention, and brought it out unblushingly whenever he was at a loss for something to say. Sustained effort evidently wearied him. He will begin a poem with the utmost spirit, but his energy is apt to flag and lead to a close so weak as to annoy the reader. His short poems are always admirably built, the endings complete and unexpected; the architectonics of his long poems leave much to be desired. He seems [lxxix] to be ridden by his own emotion, but without the power to draw it up and up to a climax; it bursts upon us in the first line, sustains itself at the same level for a series of lines, and then seems to faint exhausted, reducing the poet to the necessity of stopping as quickly as he can and with as little jar as possible. Illustrations of this tendency to a weak ending can be seen in "The Lonely Wife," "The Perils of the Shu Road," and "The Terraced Road of the Two-Edged Sword Mountains," but that he could keep his inspiration to the end on occasion, "The Northern Flight" proves.

Finally, there are his poems of battle: "Songs of the Marches," "Battle to the South of the City," and "Fighting to the South of the City." Nothing can be said of these except that they are superb. If there is a hint of let-down in the concluding lines of "Fighting to the South of the City," it is due to the frantic Chinese desire to quote from older authors, and this is an excellent example of the chief vice of Chinese poetry, since these two lines are taken from the "Tao Tê Ching," the sacred book of Taoism; the others, even the long "Songs of the Marches," are admirably sustained.

In Mr. Waley's excellent monograph on Li T'ai-po, appears the following paragraph: "Wang An-shih ( A.D. 1021-1086), the great reformer of the Eleventh Century, observes: 'Li Po's style is swift, yet never careless; lively, yet [lxxx] never informal. But his intellectual outlook was low and sordid. In nine poems out of ten he deals with nothing but wine and women.'" A somewhat splenetic criticism truly, but great reformers have seldom either the acumen or the sympathy necessary for the judgment of poetry. Women and wine there are in abundance, but how treated? In no mean or sordid manner certainly. Li T'ai-po was not a didactic poet, and we of the Twentieth Century may well thank fortune for that. Peradventure the Twenty-first will dote again upon the didactic, but we must follow our particular inclination which is, it must be admitted, quite counter to anything of the sort. No low or mean attitude indeed, but a rather restricted one we may, if we please, charge against Li T'ai-po. He was a sensuous realist, representing the world as he saw it, with beauty as his guiding star. Conditions to him were static; he wasted none of his force in speculating on what they should be. A scene or an emotion was , and it was his business to reproduce it, not to analyze how it had come about or what would best make its recurrence impossible. Here he is at sharp variance with Tu Fu, who probes to the roots of events even when he appears to be merely describing them. One has but to compare the "Songs of the Marches" and "Battle to the South of the City" with "The Recruiting Officers" and "Crossing the Frontier" to see the difference.

Tu Fu was born in Tu Ling, in the province of Shensi, [lxxxi] in A.D. 713. His family was extremely poor, but his talent was so marked that at seven years old he had begun to write poetry; at nine, he could write large characters; and at fifteen, his essays and poems were the admiration of his small circle. When he was twenty-four, he went up to Ch'ang An, the capital, for his first examination—it will be remembered that, in the T'ang period, all the examinations took place at Ch'ang An. Tu Fu was perfectly qualified to pass, as every one was very well aware, but the opinions he expressed in his examination papers were so radical that the degree was withheld. There was nothing to be done, and Tu Fu took to wandering about the country, observing and writing, but with little hope of anything save poverty to come. On one of his journeys, he met Li T'ai-po on the "Lute Terrace" in Ching Hsien. The two poets, who sincerely admired each other, became the closest friends. Several poems in this collection are addressed by one to the other.

When Tu Fu was thirty-six, it happened that the Emperor sent out invitations to all the scholars in the Empire to come to the capital and compete in an examination. Tu Fu was, of course, known to the Emperor as a man who would have been promoted but for the opinions aired in his papers. Of his learning, there could be no shadow of doubt. So Tu Fu went to Ch'ang An and waited there as an "expectant official." He waited for four years, when it oc [lxxxii] curred to him to offer three fu to the Emperor. The event justified his temerity, and the poet was given a post as one of the officials in the Chih Hsien library. This post he held for four years, when he was appointed to a slightly better one at Fêng-hsien. But, a year later, the An Lu-shan rebellion broke out, which put a summary end to Tu Fu's position, whereupon he left Fêng-hsien and went to live with a relative at the Village of White Waters. He was still living there when the Emperor Ming Huang abdicated in favour of his son, Su Tsung. If the old Emperor had given him an office, perhaps the new one would; at any rate it was worth an attempt, for Tu Fu was in dire poverty. Having no money to hire any kind of conveyance, he started to walk to his destination, but fell in with brigands who captured him. He stayed with these brigands for over a year, but finally escaped, and at length reached Fêng Chiang, where the Emperor was in residence.

His appearance on his arrival was miserable in the extreme. Haggard and thin, his shoulders sticking out of his coat, his rags literally tied together, he was indeed a spectacle to inspire pity, and the Emperor at once appointed him to the post of Censor. But this did not last long. He had the imprudence to remonstrate with the Emperor anent the sentence of banishment passed upon the general Tan Kuan. Considering that this clever and extremely learned soldier had so far relaxed the discipline of his army [lxxxiii] during one of the Northern campaigns that, one night, when his troops were all peacefully sleeping in their chariots, the camp was surrounded and burnt and his forces utterly routed, the punishment seems deserved. But Tu Fu thought otherwise, and so unwisely urged his opinion that the Emperor lost patience and ordered an investigation of Tu Fu's conduct. His friends, however, rallied to his defence and the investigation was quashed, but he was deprived of the censorship and sent to a minor position in Shensi. This he chose to regard as a punishment, as indeed it was. He proceeded to Shensi, but, on arriving there, dramatically refused to assume his office; having performed which act of bravado, he joined his family in Kansu. He found them in the greatest distress from famine, and although he did his best to keep them alive by going to the hills and gathering fire-wood to sell, and by digging up roots and various growing things for them to eat, several of his children died of starvation.

Another six months of minor officialdom in Hua Chou, and he retired to Ch'êngtu in Szechwan, where he lived in a grass-roofed house, engaged in study and the endeavour to make the two ends of nothing meet. At length, a friend of his arrived in Szechwan as Governor-General, and this friend appointed him a State Counsellor. But the grass-house was more to his taste than state councils, and after a year and a half he returned to [lxxxiv] it, and the multifarious wanderings which always punctuated his life.

Five years later, when he was fifty-five, he set off on one of his journeys, but was caught by floods and obliged to take refuge in a ruined temple at Hu Kuang, where he nearly starved before help could reach him. After ten days, he was rescued through the efforts of the local magistrate, but eating again after so long a fast was fatal and he died within an hour.

Innumerable essays have been written comparing the styles of Li T'ai-po and Tu Fu. Yüan Chên, a poet of the T'ang period, says that Tu Fu's poems have perfect balance; that, if he wrote a thousand lines, the last would have as much vigour as the first and that no one can equal him in this, his poems make a "perfect circle." He goes on: "In my opinion, the great living wave of poetry and song in which Li T'ai-po excelled is surpassed in Tu Fu's work, he is shoulder higher than Li Po." Again: "The poems of Li T'ai-po are like Spring flowers, those of Tu Fu are like the pine-trees, they are eternal and fear neither snow nor cold."

Shên Ming-chên says: "Li Po is like the Spring grass, like Autumn waves, not a person but must love him. Tu Fu is like a great hill, a high peak, a long river, the broad sea, like fine grass and bright-coloured flowers, like a pine or an ancient fir, like moving wind and gentle waves, like [lxxxv] heavy hoar-frost, like burning heat—not a quality is missing."

Hu Yu-ling uses a metaphor referring to casting dice and says that Li T'ai-po would owe Tu Fu "an ivory"; and Han Yü, speaking of both Li T'ai-po and Tu Fu, declares that "the flaming light of their essays would rise ten thousand feet."

Poetic as these criticisms are, it is their penetration which is so astonishing; but I think the most striking comparison made of Tu Fu's work is that by Tao Kai-yu: "Tu Fu's poems are like pictures, like the branches of trees reflected in water—the branches of still trees. Like a large group of houses seen through clouds or mist, they appear and disappear."

Sometime ago, in a review of a volume of translations of Chinese poetry in the London "Times," I came across this remarkable statement: "The Chinese poet starts talking in the most ordinary language and voices the most ordinary things, and his poetry seems to happen suddenly out of the commonplace as if it were some beautiful action happening in the routine of actual life."

The critic could have had no knowledge of the Chinese language, as nothing can be farther from the truth than his observation. It is largely a fact that the Oriental poet finds his themes in the ordinary affairs of everyday life, but he describes them in a very special, carefully chosen, [lxxxvi] medium. The simplest child's primer is written in a language never used in speaking, while the most highly educated scholar would never dream of employing the same phrases in conversation which he would make use of were he writing an essay, a poem, or a state document. Each language—the spoken, the poetic, the literary, the documentary—has its own construction, its own class of characters, and its own symbolism. A translator must therefore make a special study of whichever he wishes to render.

Although several great sinologues have written on the subject of Chinese poetry, none, so far as I am aware, has devoted his exclusive attention to the poetic style, nor has any translator availed himself of the assistance, so essential to success, of a poet—that is, one trained in the art of seizing the poetic values in fine shades of meaning. Without this power, which amounts to an instinct, no one can hope to reproduce any poetry in another tongue, and how much truer this is of Chinese poetry can only be realized by those who have some knowledge of the language. Such poets, on the other hand, as have been moved to make beautiful renditions of Chinese originals have been hampered by inadequate translations. It is impossible to expect that even a scholar thoroughly versed in the philological aspects of Chinese literature can, at the same time, be endowed with enough of the poetic flair to convey, uninjured, the thoughts [lxxxvii] of one poet to another. A second personality obtrudes between poet and poet, and the contact, which must be established between the two minds if any adequate translation is to result, is broken. How Miss Lowell and I have endeavoured to obviate this rupture of the poetic current, I shall explain presently. But, to understand it, another factor in the case must first be understood.

It cannot be too firmly insisted upon that the Chinese character itself plays a considerable part in Chinese poetic composition. Calligraphy and poetry are mixed up together in the Chinese mind. How close this intermingling may be, will appear when we come to speak of the "Written Pictures," but even without following the interdependence of these arts to the point where they merge into one, it must not be forgotten that Chinese is an ideographic, or picture, language. These marvellous collections of brushstrokes which we call Chinese characters are really separate pictographic representations of complete thoughts. Complex characters are not spontaneously composed, but are built up of simple characters, each having its own peculiar meaning and usage; these, when used in combination, each play their part in modifying either the sense or the sound of the complex. Now it must not be thought that these separate entities make an over-loud noise in the harmony of the whole character. They are each subdued to the total result, the final meaning, but they do produce a qualifying [lxxxviii] effect upon the word itself. Since Chinese characters are complete ideas, it is convenient to be able to express the various degrees of these ideas by special characters which shall have those exact meanings; it is, therefore, clear that to grasp a poet's full intention in a poem there must be a knowledge of the analysis of characters.

This might seem bizarre, were it not for a striking proof to the contrary. It is a fact that many of the Chinese characters have become greatly altered during the centuries since they were invented. So long ago as A.D. 200, a scholar named Hsü Shih, realizing that this alteration was taking place, wrote the dictionary known as "Shuo Wên Chieh Tzŭ," or "Speech and Writing: Characters Untied," containing about ten thousand characters in their primitive and final forms. This work is on the desk of every scholar in the Far East and is studied with the greatest reverence. Many editions have appeared since it was written, and by its aid one can trace the genealogy of characters in the most complete manner. Other volumes of the same kind have followed in its wake, showing the importance of the subject in Chinese estimation. While translators are apt to ignore this matter of character genealogy, it is ever present to the mind of the Chinese poet or scholar who is familiar with the original forms; indeed, he may be said to find his overtones in the actual composition of the character he is using.

All words have their connotations, but this is connotation and more; it is a pictorial representation of something implied, and, lacking which, an effect would be lost. It may be objected that poems were heard as well as read, and that, when heard, the composition of the character must be lost. But I think this is to misunderstand the situation. Recollect, for a moment, the literary examinations, and consider that educated men had these characters literally ground into them. Merely to pronounce a word must be, in such a case, to see it and realize, half-unconsciously perhaps, its various parts. Even if half-unconscious, the nuances of meaning conveyed by them must have hung about the spoken word and given it a distinct flavour which, without them, would be absent.

Now what is a translator to do? Shall he render the word in the flat, dictionary sense, or shall he permit himself to add to it what it conveys to an educated Chinese? Clearly neither the one nor the other in all cases; but one or the other, which the context must determine. In description, for instance, where it is evident that the Chinese poet used every means at his command to achieve a vivid representation, I believe the original poem is more nearly reproduced by availing one's self of a minimum of these "split-ups"; where, on the other hand, the original carefully confines itself to simple and direct expression, the word as it is, without overtones, must certainly be preferred. The [xc] "split-ups" in these translations are few, but could our readers compare the original Chinese with Miss Lowell's rendition of it, in these instances, I think they would feel with me that in no other way could the translation have been made really "literal," could the poem be "brought over" in its entirety. If a translation of a poem is not poetry in its new tongue, the original has been shorn of its chief reason for being. Something is always lost in a translation, but that something had better be the trappings than the essence.

I must, however, make it quite clear how seldom these "split-ups" occur in the principal parts of the book; in the "Written Pictures," where the poems were not, most of them, classics, we felt justified in making a fuller use of these analytical suggestions; but I believe I am correct in saying that no translations from the Chinese that I have read are so near to the originals as these. Bear in mind, then, that there are not, I suppose, more than a baker's dozen of these "split-ups" throughout the book, and the way they were managed can be seen by this literal translation of a line in "The Terraced Road of the Two-Edged Sword Mountains." The Chinese words are on the left, the English words on the right, the analyses of the characters enclosed in brackets:

Miss Lowell's rendering of the line was:

"On their heights, the wind whistles awesomely in the pines; it booms in great, long gusts; it clashes like the strings of a jade-stone psaltery; it shouts on the clearness of a gale."

Can any one doubt that this was just the effect that the Chinese poet wished to achieve, and did achieve by means of the overtones given in his characters?

Another, simpler, example is in a case where the Chinese poet speaks of a rising sun. There are many characters which denote sunrise, and each has some shade of difference from every other. In one, the analysis is the sunrise light seen from a boat through mist; in another, it is the sun just above the horizon; still another is made up of a period of time and a mortar, meaning that it is dawn, when people begin to work. But the poet chose none of these; instead, he chose a character which analyzes into the sun at the height of a helmeted man, and so Miss Lowell speaks of [xcii] the sun as "head-high," and we have the very picture the poet wanted us to see.

Miss Lowell has told in the Preface the manner in which we worked. The papers sent to Miss Lowell were in exactly the form of the above, and with them I also sent a paraphrase, and notes such as those at the end of this book. Far from making the slightest attempt at literary form in these paraphrases, I deliberately made them as bald as possible, and strove to keep my personality from intruding between Miss Lowell and the Chinese poet with whose mood she must be in perfect sympathy. Her remarkable gift for entering into the feeling of the poet she is translating was first shown in "Six French Poets," but there she approached her authors at first hand. It was my object to enable her to approach these Chinese authors as nearly at first hand as I could. That my method has been justified by the event, the book shows; not merely are these translations extraordinarily exact, they are poetry, and would be so though no Chinese poet had conceived them fourteen hundred years ago. It is as if I had handed her the warp and the woof, the silver threads and the gold, and from these she has woven a brocade as nearly alike in pattern to that designed by the Chinese poet as the differences in the looms permit. I believe that this is the first time that English translations of Chinese poetry have been made by a student of Chinese and a poet working together. Our ex [xciii] perience of the partnership has taught us both much; if we are pioneers in such a collaboration, we only hope that others will follow our lead.

The second section of the book, "Written Pictures," consists of illustrations, or half illustrations, of an art which the Chinese consider the most perfect medium in which a man can express himself. These Tzŭ Hua , "Hanging-on-the-Wall Poems," are less known and understood than any other form of Oriental art. A beautiful thought perpetuated in beautiful handwriting and hung upon the wall to suggest a mental picture—that is what it amounts to.

In China, the arts of poetry and calligraphy are united in the ideographs which form the written language. There are several different styles in which these ideographs, or characters, may be written. The earliest are pictograms known as the "ancient pictorial script," they were superseded in the Eighth Century B.C. by the "great seal" characters and later by the "lesser seal." These, which had been executed with the "knife pen," were practically given up when the invention of the writing-brush, which is usually translated as "pencil," revolutionized calligraphy ( circa 215 B.C. ). Their place was taken by a type of character known as " li " or "official script," a simplified form of the "seal," and this, being an improvement upon all previous styles, soon became popular. It created almost a new character in which the pictorial element had largely [xciv] disappeared, and, with certain modifications, holds good to-day. The "model hand," the "running hand," and the famous "grass hand," so popular with poets and painters, are merely adaptations of the li ; all three of these, together with the li itself, are used in the composition of written pictures.

The written pictures here translated were formerly in the possession of a Chinese gentleman of keenly æsthetic taste, and are excellent examples of the art. A photograph of one of the originals will be found opposite the translation made from it on page 170. The names which follow the poems are not those of the authors, but of the calligraphists. In the case of two poems, the authors' names are also given. These written pictures had no titles, those given here were added simply for convenience; but the titles to the poems in the body of the book are those of the poets themselves, except in one or two instances where the Chinese title conveyed so little to an Occidental mind that its meaning had to be paraphrased.

The Notes at the end of the book are intended for the general reader. For which reason, I have purposely excluded the type of note which consists in cataloguing literary cross-allusions. To know that certain lines in a poem are quoted from some earlier author, is one of a class of facts which deeply interest scholars, but are of no importance whatever to the rest of the world.

A word as to the title of this book: There lived at Ch'êng-tu, the capital of Szechwan, early in the Ninth Century, a courtesan named Hsieh T'ao, who was famous for her wit and verse-writing. Hsieh T'ao made a paper of ten colours, which she dipped in a stream, and on it wrote her poems. Now, some years before, a woman had taken the stole of a Buddhist priest to this stream in order to wash it. No sooner had the stole touched the water than the stream became filled with flowers. In an old Chinese book, "The Treasury of Pleasant Records," it is told that, later in life, Hsieh T'ao gave up the "fir-flower tablets" and made paper of a smaller size. Presumably this fir-flower paper was the paper of ten colours. The mountain stream which ran near Hsieh T'ao's house is called the "Hundred Flower Stream."

I cannot close this Introduction without expressing my gratitude to my teacher, Mr. Nung Chu. It is his unflagging interest and never-failing patience that have kept me spurred on to my task. Speaking no word of English, Mr. Nung must often have found my explanations of what would, and what would not, be comprehensible to Occidental readers very difficult to understand, and my only regret is that he cannot read the book now that it is done.

SONGS OF THE MARCHES

BY LI T'AI-PO

THE BATTLE TO THE SOUTH OF THE CITY

THE PERILS OF THE SHU ROAD

LOOKING AT THE MOON AFTER RAIN

THE LONELY WIFE

THE PLEASURES WITHIN THE PALACE

THE YOUNG GIRLS OF YÜEH

Written in the character of a beautiful woman grieving before her mirror, songs to the peonies sung to the air: "peaceful brightness", spring grief and resentment by li t'ai-po, the cast-off palace woman of ch'in and the dragon robes, the poet is detained in a nanking wine-shop on the eve of starting on a journey.

FÊNG HUANG T'AI

ASCENDING THE TERRACE OF THE SILVER-CRESTED LOVE-PHEASANTS AT THE CITY OF THE GOLDEN MOUND

THE NORTHERN FLIGHT

FIGHTING TO THE SOUTH OF THE CITY

THE CROSSWISE RIVER

ON HEARING THE BUDDHIST PRIEST OF SHU PLAY HIS TABLE-LUTE

SORROW DURING A CLEAR AUTUMN

POIGNANT GRIEF DURING A SUNNY SPRING

TWO POEMS WRITTEN AS PARTING GIFTS TO TS'UI (THE OFFICIAL) OF CH'IU PU

Sent as a parting gift to the second official of ch'iu pu, the song of the white clouds.

SAYING GOOD-BYE TO LIU SIXTEEN ON HIS RETURN TO THE HILLS

WIND-BOUND AT THE NEW FOREST REACH. A LETTER SENT TO A FRIEND

In the province of lu, at the ancestral shrine of king yao. saying farewell to wu five on his departure for lang ya.

DRINKING ALONE IN THE MOONLIGHT

A STATEMENT OF RESOLUTIONS AFTER BEING DRUNK ON A SPRING DAY

RIVER CHANT

SEPARATED BY IMPERIAL SUMMONS FROM HER WHO LIVES WITHIN

A WOMAN SINGS TO THE AIR: "SITTING AT NIGHT"

THE PALACE WOMAN OF HAN TAN BECOMES THE WIFE OF THE SOLDIERS' COOK

The sorrel horse, a poem given to a beautiful woman encountered on a field-path, saying good-bye to a friend, descending the extreme south mountain; passing the house of hu ssŭ, lover of hills; spending the night in the preparation of wine, the terraced road of the two-edged sword mountains, hearing a bamboo flute on a spring night in the city of lo yang.

THE RETREAT OF HSIEH KUNG

A TRAVELLER COMES TO THE OLD TERRACE OF SU

THEME OF THE REST-HOUSE ON THE CLEAR WAN RIVER

Drinking song.

ANSWER TO AN AFFECTIONATE INVITATION FROM TS'UI FIFTEEN

PARROT ISLAND

THE HONOURABLE LADY CHAO

THINKING OF THE FRONTIER

A SONG OF RESENTMENT

Picking willow, autumn river song.

ON THE BROAD REACH

VISITING THE TAOIST PRIEST ON THE MOUNTAIN WHICH UPHOLDS HEAVEN. HE IS ABSENT

Reply to an unrefined person encountered in the hills, reciting verses by moonlight in a western upper chamber in the city of the golden mound.

PASSING THE NIGHT AT THE WHITE HERON ISLAND

ASCENDING THE THREE CHASMS

PARTING FROM YANG, A HILL MAN WHO IS RETURNING TO THE HIGH MOUNTAIN

Night thoughts.

THE SERPENT MOUND

SENT AS A PRESENT TO CHIA THE SECRETARY

ON THE SUBJECT OF OLD TAI'S WINE-SHOP

DRINKING IN THE T'AO PAVILION

A SONG FOR THE HOUR WHEN THE CROWS ROOST

POEM SENT TO THE OFFICIAL WANG OF HAN YANG

DRINKING ALONE ON THE ROCK IN THE RIVER OF THE CLEAR STREAM

A FAREWELL BANQUET TO MY FATHER'S YOUNGER BROTHER YÜN, THE IMPERIAL LIBRARIAN

In the province of lu, to the east of the stone gate mountain, taking leave of tu fu, the moon over the mountain pass, the taking-up of arms, a song of the rest-house of deep trouble.

THE "LOOKING-FOR-HUSBAND" ROCK

AFTER BEING SEPARATED FOR A LONG TIME

BITTER JEALOUSY IN THE PALACE OF THE HIGH GATE

ETERNALLY THINKING OF EACH OTHER

PASSIONATE GRIEF

SUNG TO THE AIR: "THE MANTZŬ LIKE AN IDOL"

AT THE YELLOW CRANE TOWER, TAKING LEAVE OF MÊNG HAO JAN ON HIS DEPARTURE TO KUANG LING

In deep thought, gazing at the moon.

THOUGHTS FROM A THOUSAND LI

WORD-PATTERN

The heaven's gate mountains, poem sent on hearing that wang ch'ang-ling had been exiled to lung piao, a parting gift to wang lun, saying good-bye to a friend who is going on an excursion to the plum-flower lake, a poem sent to tu fu from sha ch'iu ch'êng, bidding good-bye to yin shu.

A DESULTORY VISIT TO THE FÊNG HSIEN TEMPLE AT THE DRAGON'S GATE

THE THATCHED HOUSE UNROOFED BY AN AUTUMN GALE

The river village, the excursion.

A NUMBER OF YOUNG GENTLEMEN OF RANK, ACCOMPANIED BY SINGING-GIRLS, GO OUT TO ENJOY THE COOL OF EVENING. THEY ENCOUNTER A SHOWER OF RAIN

THE RECRUITING OFFICERS AT THE VILLAGE OF THE STONE MOAT

Crossing the frontier, the sorceress gorge, thinking of li po on a spring day.

AT THE EDGE OF HEAVEN. THINKING OF LI T'AI-PO

SENT TO LI PO AS A GIFT

A TOAST FOR MÊNG YÜN-CH'ING

Hearing the early oriole (written in exile).

BY PO CHÜ-I

THE CITY OF STONES. (NANKING)

BY LIU YÜ-HSI

SUNG TO THE TUNE OF "THE UNRIPE HAWTHORN BERRY"

BY NIU HSI-CHI

WRITTEN BY WANG WEI, IN THE MANNER OF CHIA, THE (PALACE) SECRETARY, AFTER AN IMPERIAL AUDIENCE AT DAWN IN THE "PALACE OF GREAT BRILLIANCE"

The blue-green stream.

BY WANG WEI

FARM HOUSE ON THE WEI STREAM

Seeking for the hermit of the west hill; not meeting him.

BY CH'IU WEI

FLOATING ON THE POOL OF JO YA. SPRING

BY CHI WU-CH'IEN

SUNG TO THE AIR: "THE WANDERER"

(COMPOSED BY SU WU IN THE TIME OF THE EMPEROR WU OF HAN)

BY MÊNG CHIAO

FAREWELL WORDS TO THE DAUGHTER OF THE HOUSE OF YANG

BY WEI YING-WU

SUNG TO THE AIR: "LOOKING SOUTH OVER THE RIVER AND DREAMING"

BY WÊN T'ING-YÜN

TOGETHER WE KNOW HAPPINESS

WRITTEN BY A DESCENDANT OF THE FOUNDER OF THE SOUTHERN T'ANG DYNASTY

ONCE MORE FIELDS AND GARDENS

BY T'AO YÜAN-MING

SONG OF THE SNAPPED WILLOW

WRITTEN DURING THE LIANG DYNASTY

THE CLOUDY RIVER

(FROM THE "BOOK OF ODES")

TO THE AIR: "THE FALLEN LEAVES AND THE PLAINTIVE CICADA"

BY THE EMPEROR WU OF HAN

WRITTEN IN EARLY AUTUMN AT THE POOL OF SPRINKLING WATER

BY CHAO TI OF HAN, THE "BRIGHT EMPEROR"

PROCLAIMING THE JOY OF CERTAIN HOURS

BY THE EMPEROR LING OF (LATER) HAN

A SONG OF GRIEF

BY PAN CHIEH-YÜ

A LETTER OF THANKS FOR PRECIOUS PEARLS BESTOWED BY ONE ABOVE

BY CHIANG TS'AI-P'IN

(THE "PLUM-BLOSSOM" CONCUBINE OF THE EMPEROR MING HUANG)

BY YANG KUEI-FEI

(THE "WHITE POPLAR" IMPERIAL CONCUBINE OF THE EMPEROR MING HUANG)

SONGS OF THE COURTESANS

(WRITTEN DURING THE LIANG DYNASTY)

ONE OF THE "SONGS OF THE TEN REQUESTS"

BY TING LIU NIANG

AI AI THINKS OF THE MAN SHE LOVES

SENT TO HER LOVER YÜAN AT HO NAN (SOUTH OF THE RIVER) BY CHANG PI LAN (JADE-GREEN ORCHID) FROM HU PEI (NORTH OF THE LAKE)

CH'IN, THE "FIRE-BIRD WITH PLUMAGE WHITE AS JADE," LONGS FOR HER LOVER

THE GREAT HO RIVER

BY THE MOTHER OF THE LORD OF SUNG

(FROM "THE BOOK OF ODES")

WRITTEN PICTURES

AN EVENING MEETING

Li Hai-ku , 19th Century

THE EMPEROR'S RETURN FROM A JOURNEY TO THE SOUTH

Wên Chêng-ming , 16th Century

ON SEEING THE PORTRAIT OF A BEAUTIFUL CONCUBINE

Ch'en Hung-Shou , 19th Century

CALLIGRAPHY

Liang T'ung-shu , 18th Century

THE PALACE BLOSSOMS

Tai Ta-mien , 18th Century

ONE GOES A JOURNEY

Liu Shih-an , 18th Century

FROM THE STRAW HUT AMONG THE SEVEN PEAKS

Lu Kun , 19th Century

ON THE CLASSIC OF THE HILLS AND SEA

Written by Li Hai-ku , 19th Century

Composed by T'ao Ch'ien

Composed by Wang Ch'ang-Ling

AFTER HOW MANY YEARS

The inn at the mountain pass.

Wang Ching-ts'êng , 19th Century

LI T'AI-PO MEDITATES

Ho Ping-shou , 19th Century

PAIR OF SCROLLS

Ho Shao-chi , 19th Century

Liang T'ung-shu , 19th Century

EVENING CALM

Kao Shih-chi , 19th Century

FISHING PICTURE

Ta Chung-kuang , 19th Century

FACSIMILE OF ORIGINAL "HANGING-ON-THE-WALL POEM" ENTITLED "FISHING PICTURE"

SPRING. SUMMER. AUTUMN

[174] [175]

It is the Fifth Month, But still the Heaven-high hills Shine with snow.

The Fifth Month corresponds to June. (See Introduction.) The Heaven-high hills are the T'ien Shan Mountains, which run across the Northern part of Central Asia and in places attain a height of 20,000 feet. (See map.)

Playing "The Snapped Willow."

The name of an old song suggesting homesickness; it is translated in this volume. It was written during the Liang Dynasty ( A.D. 502-557). References to it are very common in Chinese poetry.

So that they may be able in an instant to rush upon the Barbarians.

The Chinese regarded the tribes of Central Asia, known by the generic name of Hsiung Nu, as Barbarians, and often spoke of them as such. It was during the reign of Shih Huang Ti (221-206 B.C. ) that these tribes first seriously threatened China, and it was to resist their incursions that the Great Wall was built. They were a nomadic people, moving from place to place in search of fresh pasture for their herds. They were famous for their horsemanship and always fought on horseback.

And the portrait of Ho P'iao Yao Hangs magnificently in the Lin Pavilion.

Ho P'iao Yao was a famous leader whose surname was [176] Ho. He was given the pseudonym of P'iao Yao, meaning "to whirl with great speed to the extreme limit," because of his energy in fighting. His lust for war was so terrible that the soldiers under him always expected to be killed. After his death, the Emperor Wu of Han erected a tomb in his honour. It was covered with blocks of stone in order that it might resemble the Ch'i Lien Mountains, where Ho P'iao Yao's most successful battles had been fought.

The Lin Pavilion was a Hall where the portraits of distinguished men were hung.

The Heavenly soldiers arise .

The Chinese soldiers were called the "Heavenly Soldiers" because they fought for the Emperor, who was the Son of Heaven.

Divides the tiger tally .

A disk broken in half, worn as a proof of identity and authority. The General was given one half, the Emperor kept the other.

The Jade Pass has not yet been forced .

In order to reach the Central Asian battle-fields, the soldiers were obliged to go out through the Jade Pass, or Barrier, which lay in the curious bottle-neck of land between the mountain ranges which occupy the centre of the continent. (See map.)

They seized the snow of the Inland Sea .

The Inland, or Green Sea, is the Chinese name for the Kokonor Lake lying West of the Kansu border. (See map.)

They lay on the sand at the top of the Dragon Mound .

The Dragon Mound is a high ridge of land on the Western border of Shensi, now comprising part of the Eastern [177] boundary of Kansu. The native accounts say that the road encircles the mountains nine times, and that it takes seven days to make the ascent. "Its height is not known. From its summit, one can see five hundred li . To the East, lie the homes of men; to the West, wild wastes. The sound of a stone thrown over the precipice is heard for several li ."

All this they bore that the Moon Clan.

Name of one of the Hsiung Nu tribes. It was this tribe, known to Europeans under name of Huns, who overran Europe in the Fifth Century.

During the reign of the T'ang Emperor, Hsüan Tsung ( A.D. 712-756), better known as Ming Huang, a rebellion broke out under An Lu-shan, an official who had for many years enjoyed the Emperor's supreme favour. Opinions among the advisers to the throne differed as to whether or not the Emperor had better fly from his capital and take refuge in the province of Szechwan, the ancient Shu. Li T'ai-po strongly disapproved of the step, but as he was no longer in office could only express his opinion under the guise of a poem. This poem, which the Chinese read in a metaphorical sense, describes the actual perils of the road leading across the Mountains of the Two-Edged Sword, the only thoroughfare into Szechwan. Li T'ai-po's counsel did not prevail, however, and the Emperor did actually flee, but not until after the poem was written.

No greater undertaking than this has been since Ts'an Ts'ung and Yü Fu ruled the land.

These were early Rulers. Ts'an Ts'ung was the first [178] King of Shu, the modern Szechwan. He was supposed to be a descendant of the semi-legendary Yellow Emperor.

But the earth of the mountain fell and overwhelmed the Heroes so that they perished.

An historical allusion to five strong men sent by the King of Shu to obtain the daughters of the King of Ch'in.

Above, the soaring tips of the high mountains hold back the six dragons of the sun.

The sun is supposed to drive round the Heavens once every day in a chariot drawn by six dragons and driven by a charioteer named Hsi Ho.

The gibbons climb and climb.

Gibbons, which are very common in this part of China, are a small species of tailless ape, thoroughly arboreal in their habits. They make the woods sound with unearthly cries at night, and are unsurpassed in agility and so swift in movement as to be able to catch flying birds with their paws.

This is what the Two-Edged Sword Mountains are like!

In this range, the mountains are so high, the cliffs so precipitous, and the passes so few, that it was almost impossible to devise a means of crossing them. The Chinese, however, have invented an ingenious kind of pathway called a "terraced" or "flying" road. Holes are cut in the face of the cliffs, and wooden piles are mortised into them at an angle. Tree trunks are then laid across the space between the tops of the piles and the cliff wall, making a corduroy road, the whole being finally covered with earth. These roads are so solidly built that not only people, but horses and even small carts, can pass over them. As there are no railings, however, travel upon them is always fraught with more or less danger.

Half of the moon-toad is already up.

In Chinese mythology, the ch'an , a three-legged toad, lives in the moon and is supposed to swallow it during an eclipse. The toad is very long-lived and grows horns at the age of three thousand years. It was originally a woman named Ch'ang O, who stole the drug of Immortality and fled to the moon to escape her husband's wrath. The moon is often referred to as ch'an , as in the poem.

The glimmer of it is like smooth hoar-frost spreading over ten thousand li .

A li is a Chinese land measurement, equal to about one third of a mile.

There is only the moon, shining through the clouds of a hard, jade-green sky.

The term "jade," in Chinese literature, includes both the jadeites and nephrites. These semi-transparent stones are found in a great variety of colours. There are black jades; pure white jades, described by the Chinese as "mutton fat"; jades with brown and red veins; yellow jades tinged with green; grey jades with white or brown lines running through them; and, most usual of all, green jades, of which there are an infinite number of shades.

These green jades vary from the dark, opaque moss-green, very much like the New Zealand "green-stone," to the jewel jade called by the Chinese fei ts'ui , or "kingfisher feather," which, in perfect examples, is the brilliant green of an emerald. As a result of this range of [180] colouring, the Chinese use the term "jade" to describe the tints seen in Nature. The colours of the sky, the hills, the sea, can all be found in the jades, which are considered by the Chinese as the most desirable of precious stones. In addition to its employment in actual comparison, the word "jade" is very often used in a figurative sense to denote anything especially desirable.

Beneath the quilt of the Fire-Bird, on the bed of the Silver-Crested Love-Pheasant.

The Fire-Bird is the Luan , and the Love-Pheasant the Fêng Huang ; both are fully described in the table of mythical animals in the Introduction.

As the tears of your so Unworthy One escape and continue constantly to flow.

The term "Unworthy One" is constantly used by wives and concubines in speaking of themselves to their husbands or to the men they love.

As I toss on my pillow, I hear the cold, nostalgic sound of the water-clock.

The clepsydra, or water-clock, has been used by the Chinese for many centuries, one can still be seen in the North Worshipping Tower in Canton, and another in the "Forbidden" portion of the Peking Palace, where the dethroned Manchu Emperor lives. The following account of the one in Canton is taken from the "Chinese Repository," Volume XX, Page 430: "The clepsydra is called the 'copper-jar water-dropper.' There are four covered jars standing on a brickwork stairway, the top of each of which is level with the bottom of the one above it. The largest measures twenty-three inches high and broad and contains seventy catties or ninety-seven and a half pints of water; the second is twenty-two inches high and twenty-one inches broad; the third, [181] twenty-one inches high and twenty broad; and the lowest, twenty-three inches high and nineteen inches broad. Each is connected with the other by an open trough along which the water trickles. The wooden index in the lowest jar is set every morning and afternoon at five o'clock, by placing the mark on it for these hours even with the cover through which it rises and indicates the time. The water is dipped out and poured back into the top jar when the index shows the completion of the half day, and the water is renewed every quarter."

From little, little girls, they have lived in the Golden House.

The "Golden House" is an allusion to a remark made by the Emperor Wu of Han who, when still a boy, exclaimed that if he could marry his lovely cousin A-chiao he would build a golden house for her to live in.

Palaces were often given most picturesque names, and different parts of the precincts were described as being of "jade" or some other precious material, the use of the word "golden" is, of course, in this case, purely figurative.

The organization of the Imperial seraglio, which contained many thousands of women, was most complicated, and the ladies belonged to different classes or ranks.

There was only one Empress, whose title was Hou , and, if the wife of the preceding monarch were still alive, she was called T'ai Hou , or Greater Empress. These ladies had each their own palace. Next in rank came the principal Imperial concubines or secondary wives called Fei . As a rule, there were two of them, and they had each their palace and household. After them came the P'in described as "Imperial concubines of first rank," or maids of honour, who lived together in a large palace [182] and who, once they had attained this rank, could never be dispersed. (See Note 69.) The ladies of the Court are often spoken of as Fei-P'in . Of lower rank than these were the innumerable Palace women called Ch'ieh , concubines or handmaids. The use of the word is not confined to the inmates of the Palace, as ordinary people may have ch'ieh . Little girls who were especially pretty, or who showed unusual promise, were often sent to the Palace when quite young, that they might become accustomed to the surroundings while still children. (See Introduction.)

They are lovely, lovely, in the Purple Hall.

The Ruler of Heaven lives in a circumpolar constellation called the Tzŭ Wei, Purple Enclosure; therefore the Palace of his Son, the Ruler of Earth, is called "Purple."

Their only sorrow, that the songs and wu dances are over.

The wu dance is a posturing dance for which special, very elaborately embroidered dresses with long streamers are worn. As the arms move, these scarves float rhythmically in the air.

Changed into the five-coloured clouds and flown away.

The allusion to the five-coloured clouds is to the beautifully variegated clouds, bright with the five colours of happiness, upon which the Immortals ride.

WRITTEN IN THE CHARACTER OF A BEAUTIFUL WOMAN

Bright, bright, the gilded magpie mirror.

Magpies are the birds of happiness. There is an old story of the Gold Magpie which tells that, ages ago, a husband and wife, at parting, divided a round mirror between them, each keeping a half as a guarantee of fidelity. Unhappily, the wife forgot her marriage vows, [183] and to her horror the half circle she had kept turned into a magpie and flew away. Since then, magpies are often carved on mirror backs as reminders and warnings.

I sit at my dressing-stand, and I am like the Green Fire-Bird who, thinking of its mate, died alone.

The Green Fire-Bird is a fabulous creature who is regarded as the embodiment of every grace and beauty. It is the essence of the Fire God, and references to it in stories of love and marriage are frequent. One of the most popular of these tales is that of a King of India who caught a beautiful bird with green plumage of an extraordinary brilliance. He valued it greatly, and had an exquisite gold cage made for it. For three years it lived in captivity, and not a sound came from it in all that time. At last, the King, who was much puzzled at its silence, consulted his wife, saying: "Is the creature dumb?" She replied: "No, but every creature is the same, when it meets one of its own species it will speak." Not knowing how to obtain a mate for the Green Fire-Bird, the King placed a large mirror in its cage. The Luan danced with joy, uttered strange cries, and then, with all its strength, hurled itself against its own reflection and fell dead.

My tears, like white jade chop-sticks, fall in a single piece.

It was said of the Empress Ch'ên of Wei (403-241 B.C. ) that her tears fell so fast they formed connected lines like jade chop-sticks.

SONGS TO THE PEONIES

The "Songs to the Peonies" were written on a Spring morning when Ming Huang, accompanied by Yang Kuei-fei, his favourite concubine, and his Court, had gone to see the blooms for which he had a passion. As he [184] sat, admiring the flowers and listening to the singing of the Palace maidens, he suddenly exclaimed: "I am tired of these old songs, call Li Po." The poet was found, but unfortunately in a state best described by the Chinese expression of "great drunk." Supported by attendants on either side of him, he appeared at the pavilion, and while Yang Kuei-fei held his ink-slab, dashed off the "Songs." She then sang them to the air, "Peaceful Brightness," while the Emperor beat time.

The "Songs" compare Yang Kuei-fei to the Immortals and to Li Fu-jên, a famous beauty of whom it was said that "one glance would overthrow a city, a second would overthrow the State." But, unluckily, Li T'ai-po also brought in the name of the "Flying Swallow," a concubine of the Han Emperor Ch'êng, who caused the downfall of the noble Pan Chieh-yü (see Note 155) and is looked upon as a despicable character. Kao Li-shih, the Chief Eunuch of the Court, induced Yang Kuei-fei to take this mention as an insult, and it finally cost Li T'ai-po his place at Court.

In the third "Song," there is an allusion to the Emperor under the figure of the sun. When his presence is removed, the unhappy, jealous flowers feel as if they were growing on the North side of the pavilion.

Yang Kuei-fei, the most famous Imperial concubine in Chinese history, was a young girl of the Yang (White Poplar) family, named Yü Huan, or Jade Armlet; she is generally referred to as Yang Kuei-fei or simply Kuei-fei—Exalted Imperial Concubine.

The Chief Eunuch brought her before the T'ang Emperor, Ming Huang, at a time when the old man was inconsolable from the double deaths of his beloved Empress and his favourite mistress.

The story goes that the Emperor first saw Yang Yü Huan, then fifteen years old, as she was bathing in the [185] pool made of stone, white as jade, in the pleasure palace he had built on the slopes of the Li Mountains. As the young girl left the water, she wrapped herself in a cloak of open-work gauze through which her skin shone with a wonderful light. The Emperor immediately fell desperately in love with her, and she soon became chief of the Palace ladies wearing "half the garments of an Empress."

Yang Kuei-fei rose to such heights of power that her word was law; she had her own palace, her own dancing-girls, and was even allowed by the doting monarch to adopt the great An Lu-shan, for whom she had a passion, as her son. Her follies and extravagancies were innumerable, and her ill-fame spread about the country to such an extent that, when the rebellion broke out (see Note 37), the soldiers refused to fight until she had been given over to them for execution.

After her death, Ming Huang spent three inconsolable years as an exile in Szechwan, and his first act upon his return to the Empire, which he had ceded to his son, was to open her grave. It was empty. Even the gold hair-ornaments, and the half of a round gold box shared with the Emperor as an emblem of conjugal unity, had gone; the only trace of the dead beauty was the scent-bag in which she had kept these treasures. "Ah," cried the unhappy monarch, "may I not see even the bones of my beloved?" In despair, he sent for a Taoist magician and begged him to search the Worlds for Yang Kuei-fei. The Taoist burnt charms to enlist the help of the beneficent spirits, but these were unsuccessful in their search. He finally sat in contemplation until the "vital essence" issued from his body and descended to the World of Shades. Here the names of all the spirits who have passed from the World of Light are entered in classified books, but that of Yang Kuei-fei was not among them. [186] The demon in charge insisted that if the name were not entered, the spirit had not arrived, and the Taoist left, sad and crest-fallen.

He then reflected that if she really were not at the Yellow Springs below, she must be among the Immortals above. He therefore ascended to Paradise, and asked the first person he met, who happened to be the Weaving Maiden who lives in the sky, for news of the lost lady. The Weaving Maiden was most uncommunicative, and found much difficulty in believing that Ming Huang, who had consented to the execution of Yang Kuei-fei, really mourned her death, but finally admitted that she was living among the Immortals on the island of P'êng Lai in the Jade-grey Sea, and even assisted the Taoist to find her. She then told Yang Kuei-fei that, if she still loved the Emperor, the Moon Mother might be induced to allow a meeting at the full moon on the fifteenth day of the Eighth Month. Yang Kuei-fei eagerly assented, and giving the Taoist a gold hairpin and her half of the round box as a proof of her existence, begged that he hasten back to the World of Light and make all arrangements with her lover.

Accordingly, at the appointed time, the Taoist threw his fly-whip into the air, creating a bridge of light between this world and the moon, and over this Ming Huang passed. Yang Kuei-fei was waiting for him. She stood under the great cassia-tree which grows in the moon, and was surrounded by fairies.

The story, which is often sung to the air "Rainbow Skirts and Feather Collar," goes on to relate that the Weaving Maiden was moved to deep pity by their joy at meeting and arranged with the Jade Emperor, Chief Ruler of the Heavens, that the pair, immortalized by their great love, should live forever in the Tao Li Heaven.

THE PALACE WOMAN AND THE DRAGON ROBES

I ponder his regard, not mine the love Enjoyed by those within the Purple Palace.

The Palace woman of Ch'in was evidently one of the lower ranks of concubines who lived in the Women's Apartments and only appeared when sent for, not in one of the palaces given to ladies of higher rank.

If floods should come, I also would not leave. A bear might come and still I could protect.

Now that she is no longer needed, she reflects sadly on the stories of two heroines whose behaviour she would gladly have emulated. These are Fên Chieh-yü, a favourite of the Han Emperor, Yüan, who once protected her master with her own body from the attack of a bear which had broken out of its cage; and Liu Fu-jên, concubine of King Chao of Ch'u. It is told of Liu Fu-jên that one day she went with the King to the "Terrace by the Stream," where he told her to wait for him until he returned from the capital. While she waited, the river rose, but she refused to leave unless by Imperial command. By the time this arrived she was drowned.

Of serving Sun and Moon.

The "Sun and Moon" are the Emperor and Empress.

THE NANKING WINE-SHOP

In the wine-shops of Wu, women are pressing the wine.

Wine made from grain is fermented for several weeks in tubs and then strained or "pressed" through cloths. It is not red, like wine from grapes, but either a shade of yellow or pure white. Wines made from grapes, plums, apples, pears, lichis, and roses, are sometimes used, but are not nearly so strong as the decoctions from grains.

The silver-crested love-pheasants strutted upon the Pheasant Terrace.

About A.D. 493, three strange and beautiful birds were noticed inside the city walls of Nanking, then called the "City of the Golden Mound." At first, the people did not suspect the identity of the birds, but when they saw that all the other birds assembled and appeared to be paying homage to the strangers, they realized that the visitors were the famous Fêng Huang . (See table of mythical birds in Introduction.) The terrace was built to commemorate the occasion.

Here also, drifting clouds may blind the Sun.

The drifting clouds are supposed to be the evil courtiers who have poisoned the mind of the Emperor, i.e. the Sun, against Li T'ai-po.

The An Lu-shan rebellion, which broke out during the reign of the T'ang Emperor, Ming Huang, was very nearly successful, and, if the leader had not been assassinated in A.D. 757 by his son, might have caused the overthrow of the dynasty. As it was, the Emperor, having fled to Szechwan—a step strongly deprecated by Li Tai-po in the poem, "The Perils of the Shu Road" (see Note 11)—abdicated in favour of his son, Su Tsung, who crushed the rebellion. The poem refers to the time when it was at its height, and the Emperor's forces were flying to the North.

The rushing whale squeezes the Yellow River; The man-eating beasts with long tusks assemble at Lo Yang.

During the rebellion, both sides of the Yellow River were [189] lined with rebels, the population was obliged to fly, and the country was devastated as if a whale had rushed up the river and caused it to overflow its banks.

The "beasts" are fabulous creatures called tso chih , with tusks three feet long, who delight in eating the flesh of men. Li T'ai-po uses them metaphorically for the rebels who are threatening the capital.

When, before our glad faces, shall we see the Glory of Heaven?

The Emperor, under the usual figure of the Sun.

I say the Crosswise River is terrible. The savage wind blows as if it would overturn the Heaven's Gate Mountains.

The "Crosswise River" is that section of the Yangtze which flows past steep cliffs in Anhwei. The "Heaven's Gate Mountains" tower above, making a sharp defile.

From the beginning of things, the Ox Ledge has been more dangerous than the Standing Horse Hill.

A very swift current runs past the Ox Lodge, and boats are obliged to wait for daylight before attempting to breast it. The Standing Horse Hill, so called from its resemblance to a standing horse, is above a reach of the Yangtze where the river is comparatively tranquil.

Is the Eighth Month tide-bore of Chêkiang equal to this?

The T'ien River in Chêkiang is famous for its bore, or tidal wave. During the Autumnal equinox, this bore sometimes attains a height of twenty feet and more.

I could not yet lay aside my face of shame; I hung my head, facing the dark wall.

In China, little girls are supposed to hide their faces at the suggestion of marriage.

I often thought that you were the faithful man who clung to the bridge-post.

A certain Wei Shêng had a great reputation for sincerity and reliability, which was put to proof on an occasion when he had an appointment with a lady to meet on a bridge. The lady did not come. But, in spite of the fact that the waters rose to a flood, Wei Shêng would not leave. Finally, as he stood there clinging to the bridge-post to keep himself firm, the waves engulfed him and he was never seen again.

That I should never be obliged to ascend the Looking-for-Husband Ledge.

A hill on the banks of the Yangtze, so called because of a legend that, many centuries ago, a wife, whose husband had been away for several years, went daily to watch for his returning sail. In the end, she was turned to stone on the spot where she had kept her vigil.

To the Ch'ü Tang Chasm and the Whirling Water Rock of the Yü River Which, during the Fifth Month, must not be collided with; Where the wailing of the gibbons seems to come from the sky.

The Ch'ü T'ang is the first of the three noted chasms in the upper reaches of the Yangtze. At the point where the River Yü empties into the Yangtze, there is a great rock which, when uncovered, is more than two hundred feet high. In the Fifth Month (June) the water from the melting snows of the Tibetan mountains causes the river to rise to such an extent that the rock is covered, which makes it especially dangerous to navigation. The height of the cliffs on either side of the gorge is so tremendous that the wailing of the gibbons (see Note 15) in the woods above sounds as though it came from the sky.

I will not go far on the road to meet you, I will go straight until I reach the Long Wind Sands.

The Long Wind Sands are many a day's journey from the village of Ch'ang Kan, which stands just outside the South Gate of Nanking. What the lady implies is that she will go to "the ends of the earth" to meet her returning husband.

I climb the hills of Chiu I.

The Chiu I, or "Nine Peaks," lie to the South of the Tung T'ing Lake (see map) into which the three divisions of the Hsiang River debouch after having united.

I go by the "Bird's Path."

A term very often used for steep mountain paths.

I think much of fishing for a leviathan from the Island of the Cold Sea.

The legend referred to at the end of the poem is as follows: A group of five islands in the Pi Hai, the Jade-grey Sea, were inhabited by the Immortals, who found themselves very uncomfortable as these islands, instead of standing firmly, rose and fell in the most disconcerting manner. The Immortals therefore applied to the Jade Emperor for assistance, and he commanded fifteen leviathans, three to each island, to raise their heads and support the islands, thus keeping them from rocking. All was well until a man from the Elder Dragon Country appeared and with one cast of his line caught six of the monsters, the result being that two of the islands toppled over and sank in the sea. The three which remain are known as the "Three Hills of the Immortals." This tale has become proverbial, and people who are disappointed in their ambition say "I have no rod with which to catch a leviathan."

I feel as one feels listening to the sound of the waters of the Dragon Mound in Ch'in.

(See Note 9.)

The gibbons wailing by the Serpent River.

(See Note 15.)

I feel as the "Shining One" felt when she passed the Jade Frontier, As the exile of Ch'u in the Maple Forest.

Two allusions which suggest homesickness. The "Shining One" is Chao Chün. (See Note 79.) The exile of Ch'u is Ch'ü Yüan, the famous statesman. (See Note 62.)

TWO POEMS WRITTEN TO TS'UI (THE OFFICIAL)

In both these poems, Ts'ui is compared to T'ao Yüan-ming, author of "Once More Fields and Gardens," published in this volume. T'ao is the ideal of the educated scholar, who prefers a life in the fields to any official post. Many stories are told of him. He planted five willows in front of his house, and is therefore often spoken of as the "Teacher of the Five Willows." He was so fond of music that he declared he could imagine the sweet sounds of the ch'in , and often carried about a stringless instrument over which he moved his hands. The ch'in , or table-lute, is fully described in Note 114.

WIND-BOUND AT THE NEW FOREST REACH

To-day, at dawn, see the willows beyond the White Gate.

The White Gate is the Western Gate. The points of the compass are governed by colours, elements, mythological beasts, and seasons, thus:

East: Green. Wood. The Blue-green Dragon. Spring. South: Red. Fire. The Vermilion Bird. Summer. West: White. Metal. The White Tiger. Autumn. North: Black. Water. The Black Warrior. Winter. Centre: Yellow. Earth.

But we will keep our appointment by the far-off Cloudy River.

The Cloudy River is the Chinese name for the Milky Way.

There would be no Wine Star in Heaven.

The Wine Star is a constellation composed of three stars, to the North of the Dipper.

There should be no Wine Springs on Earth.

The Wine Springs lie, one in Kansu, and one in Shansi. (See map.) The water of the one in Kansu is supposed to taste like wine, that of the one in Shansi is used in the making of wine.

Jade flageolets and pipes of gold.

The Chinese flageolet is a tube measuring a little more than a foot in length. It has five holes above, one below, and one at the end through which it is played. They are now made of bamboo, but formerly were made of copper, jadestone, or marble, as such materials were considered less liable to be affected by the weather.

The Immortal waited, Then mounted and rode the yellow crane.

Tou Tzŭ-an, who had attained Immortality by living a life of contemplation, was transported to the Taoist Paradise by a crane so old that it had turned yellow.

Rather would he be followed by the white gulls.

This line refers to a story from a book treating of Taoist subjects long supposed to have been written by a philosopher called Lieh Tzŭ, but this is now known to have been a Second Century forgery. A translation of the story reads: "The man who lived by the sea loved the sea-gulls. Every day, as the sun rose above the horizon, the birds from the sea assembled in hundreds and flew about. His father said: 'I hear the sea-gulls follow you and fly round you. Catch some in your hands and bring them to me that I too may enjoy them.' The next day the birds from the sea all performed the posturing dance in the air, but did not descend."

The tzŭ and fu of Ch'ü P'ing hang suspended like the sun and moon.

The tzŭ and fu are two irregular forms of verse, they are referred to in the Introduction in the part dealing with versification. Ch'ü P'ing is another name for Ch'ü Yüan, a famous poet and statesman who lived 332-295 B.C. (See Introduction.)

I could move the Five Peaks.

The sacred mountains of the "four quarters" and the nadir (or the four points of the compass and the centre of the earth). They are the T'ai Shan in the East, the Hua Shan in the West, the Hêng Shan in the North, the Hêng Shan in the South, and the Sung Shan in the centre.

SEPARATED BY IMPERIAL SUMMONS

The Emperor commands; three times the summons. He who left has not yet returned.

The official has not responded quickly to the summons from the capital, so the messenger has been obliged to come three times. Upon the third occasion, the official realizes that the matter is urgent and prepares to [195] depart the next day at sunrise, before the messenger can have reached the Palace on his return journey.

Our thoughts will be with each other. I must ascend the Looking-for-Husband Hill.

(See Note 45.)

You must not imitate Su Ch'in's wife and not leave your loom.

Su Ch'in, who lived in the Fourth Century B.C. , was away from home many years; when he returned, his wife took no notice whatever, and did not even leave the loom at which she sat weaving cloth.

I sit, sit in the North Hall.

The "North Hall" is a term for the Women's Apartments, which always lie farthest from the Great Gate placed in the South wall of the house. (See Plan of House.)

Then, though my Lord sang ten thousand verses which should cause even the dust on the beams to fly, to me it would be nothing.

It is said that when Yü Kung, a man of the State of Lu who lived during the Han Dynasty, sang, the sounds were so exquisite that even the dust on the beams flew. "To cause the dust on the beams to fly" has therefore become a current saying.

THE PALACE WOMAN AND THE SOLDIERS' COOK

Once the Unworthy One was a maiden of the Ts'ung Terrace.

The Ts'ung Terrace referred to by the sad lady who, in the dispersal of the Palace women (see Introduction), had fallen to such a low degree, stood in the Palace of King Chao, who lived at the time of the "Spring and Autumn Annals," many centuries before our era.

A BEAUTIFUL WOMAN ENCOUNTERED ON A FIELD-PATH

Down comes the riding-whip, straight down—it strikes the Five Cloud Cart.

The Immortals used Five Coloured Clouds to ride upon, therefore the term, "Five Cloud Cart," has become a complimentary expression for a cart or carriage in which a beautiful young woman is travelling.

HEARING A BAMBOO FLUTE IN THE CITY OF LO YANG

I hear "The Snapped Willow."

An allusion to the old song suggesting homesickness. (See Note 2.)

Hsieh Kung is the honorary title of the poet, Hsieh T'iao, who lived in the Fifth Century A.D. Li T'ai-po, who greatly admired him, constantly quoted his poems, and expressed a wish to be buried on the Spring-green Mountain where Hsieh Kung had lived. Some accounts say that he was first buried elsewhere, but that afterwards his body was removed and put where he desired.

The old Imperial Park—the ruined Terrace—the young willows.

Early in the Fifth Century B.C. , Fu Ch'ai, King of Wu, built the Ku Su Terrace to please Hsi Shin, one of the most famous beauties in history. It was nearly two miles long, and took three years to build. Its foundations can still be traced on the hills near Soochow, which was the capital of Wu.

THE REST-HOUSE ON THE CLEAR WAN RIVER

I love the beauty of the Wan River.

A little river near Ning Kuo-fu in Anhwei. (See map.)

Really, one cannot help laughing to think that, until now, the rapid current celebrated by Yen Has usurped all the fame.

The philosopher Yen Kuang ( circa A.D. 25) is better known as Yen Tzŭ-ling. The river in which he loved to fish was the Hsin An.

A party of friends who are in the habit of meeting each other constantly are called by numbers according to age. The same custom is used to distinguish members of a family. (See Introduction.)

You have the "bird's foot-print" characters.

Writing is supposed by the Chinese to have been invented by Ts'ang Chieh, a minister of the Yellow Emperor (2698-2598 B.C. ) who, having "observed the shapes of things in the heavens and the forms of things on earth, also the foot-prints of birds and beasts on the sand and mud," suddenly conceived the idea of pictographic writing. It is highly complimentary to speak of a person's writing as being like the "bird's foot-prints."

You suggest that we drink together at the Lute Stream.

The Ch'in Ch'i T'ai (Table-lute Stream Terrace) was a stone terrace where a famous player of the table-lute, who is said to have attained Immortality, lived. The legend is that he took a small dragon in the form of a carp from the Ch'in stream and kept it for a month, when it changed its shape into that of a dragon and ascended to Heaven.

Moon over the houses of Han, over the site of Ch'in.

Ch'in was the name of the State which overcame all the others and welded China into a homogeneous Empire instead of a loose federation. (See Introduction.) The lady Chao lived during the Han Dynasty.

Wang Ch'iang, known to posterity as Chao Chün, the "Brilliant-and-Perfect," lived in the First Century B.C. The daughter of educated parents, she was brought up in the strictest Confucian principles; in the words of the Chinese, she "did not speak loudly nor did she look beyond the doors, indeed, even within the house, she only walked the path which led to her mother's room. Her ears were closed to all distracting sounds, therefore her heart and mind were pure like those of the Immortals." Her father regarded her as a precious jewel, and although many suitors presented themselves, he refused to listen to their proposals, and finally, when she was seventeen, sent her to the capital as an offering to the Han Emperor Yüan.

Upon arriving at the Palace, the young girl was housed in the inner rooms, among the innumerable Palace women who lived there in constant hope of a summons to the Imperial presence. As the Son of Heaven never went into this part of his Palace, it was customary to catalogue the inmates and submit their portraits to him, a form of procedure which led to much bribery of the Court painters. The rigid principles of the daughter of the Wang clan forbade her to comply with this Palace custom, and the portrait which appeared in the catalogue was such a travesty of her exquisite features that it roused no desire in the Imperial breast.

Five or six dreary years passed, and the young girl remained secluded in the Women's Apartments. Shortly [199] before this time, one of the Hsiung Nu tribes (see Note 3) had surrendered to the Chinese soldiers, and as a proof of good faith on both sides had received permission to serve as a frontier guard. Soon after, the head of the tribe sent to ask that one of Yüan Ti's ladies be sent him as Queen. The catalogue was consulted, and the decision fell upon the daughter of Wang as being the one among the Palace women who had the fewest charms. She was therefore told to prepare herself for a journey to the desert wastes where she would reign over a savage Central Asian tribe, a prospect terrifying to one brought up in strict seclusion among people of refinement.

Custom demanded that, on the point of departure, she should appear before the Son of Heaven in order to thank her Imperial Master for his kind thoughtfulness in thus providing for her future, and then be formally handed over to the envoys. The audience was held in one of the secondary halls, the Court was assembled, the envoys stood ready, and the lady entered. At the sight of her unusual beauty, every one was thunderstruck, even the Emperor could hardly refrain from springing off the Dragon Throne and speaking to her. But it was too late; there was nothing to be done. The most beautiful of all the Palace women was pledged to the Hsiung Nu Khan, the escort which was to convey her over the Jade Pass waited, and soon the broken-hearted girl set off.

Fury and consternation spread through the Palace; a camel laden with gold was sent in pursuit; the guilty painter, Mao Yen-shou, was executed and his immense fortune sent as a consolation to the Wang family; but all this could not save the young girl from her fate. The Hsiung Nu ambassador refused to ransom her, and she passed out through the Jade Barrier to the "Yellow Sand Fields" beyond.

The banished daughter of Han was true to the prin [200] ciples in which she had been schooled. Instead of committing suicide, as she longed to do, she submitted to the will of the Five Great Ones—Heaven, Earth, The Emperor, her Father, and her Mother—and performed her duties as a wife to the best of her ability in spite of the homesickness from which she suffered perpetually.

Upon the death of the Khan, she felt that her hour of deliverance had at last come and that she was at liberty to poison herself. This she did, and was buried in the desert, but the mound over her grave remained always green.

Because of her pseudonym, "Brilliant-and-Perfect," she is often referred to as "Ming Fei," the "Bright Concubine." Allusions to her story always suggest homesickness.

I desire to send the "harmonious writings."

Letters from wives to husbands are often spoken of as though they carried sweet sounds.

He who wears the dragon robes delighted in the sweetly-scented wind of her garments.

Appointments for the Emperor's use were all spoken of as "dragon" appointments, and the analysis of the character which means the Emperor's love, is a dragon under a roof. Ladies' clothes were, and are to-day, kept in cupboards in which scented woods were burned, therefore as the long sleeves of their dresses swayed back and forth a sweet perfume came from them.

How was it possible for the "Flying Swallow" to snatch the Emperor's love?

The "Flying Swallow" was a famous concubine. (See Note 30.)

RECITING VERSES BY MOONLIGHT

I suggest that men meditate at length on Hsieh Hsüan Hui.

A reference, under a pseudonym, to the poet Hsieh T'iao, whose work Li T'ai-po so much admired. (See Note 72.) "Hsüan" is applied to the names of gods to indicate that they deserve praise and worship, and "Hui" means bright, splendid, or a ray of the sun.

At dawn, I left the Red Bird Gate.

An allusion to the bird which rules the South. (See Note 55.)

At sunset, I came to roost on the White Heron Island.

According to the Chinese commentary, this island lies "in the heart's centre of the river, three li West of the district of the Golden Mound (Nanking), and many herons collect there."

And the longing in my heart is like that for the Green Jasper Tree.

This tree grows in the Taoist Paradise, supposed to lie in the K'un Lun Mountains. (See map.) Those who eat its blossoms become immortal.

These are the famous chasms of the Yangtze River, between Ichang and Chungking. Their names are: "The Terrifying Barrier," "The Sorceress Gorge," and "The Western Sepulchre." Joined together in one great line of precipitous cliffs, they are among the extraordinary natural objects of the world and are most awe-inspiring.

The Serpent River runs terribly fast. The Serpent River can be suddenly exhausted.

A reference to the fact that, although the water of the river flows with terrible speed while the snow waters are coming down, during the Winter it is very low, and many parts are quite dry. (See Note 46.)

Three dawns shine upon the Yellow Ox. Three sunsets—and we go so slowly.

A cliff beneath which are rapids so difficult and dangerous to pass that the utmost care must be taken in navigating them. Boats ascending this stretch of the river often take several days to pass a given point. (See Introduction for a description of the Yangtze River and travel upon it.)

PARTING FROM YANG, A HILL MAN

You are going to pick the fairy grasses And the shooting purple flower of the ch'ang p'u.

"Hill men" is a term applied to those who desire to become worthy of joining the ranks of the Immortals, and for this reason lead a life of contemplation among the hills. The fairy grasses and the ch'ang p'u (see table of plants in Introduction) both grow in the Taoist Paradises.

Riding down from the green-blue Heaven on a white dragon.

The dragon is one of the steeds of the Immortals.

The mercy of the Sainted Lord is far greater than that of Han Wên Ti. The Princely One had pity, and did not appoint you to the station of the Unending Sands.

The allusion is to an incident which occurred in the [203] Second Century B.C. when a famous scholar named Chia was sent to Ch'ang Sha, literally "Unending Sands" (see map), and died there of the damp vapours.

Old Tai is gone down to the Yellow Springs.

The Yellow Springs lie in the nether world, where spirits go after death.

There is no Li Po on the terrace of Eternal Darkness.

This world is known as the World of Light, and below it lies the World of Shades, where the sun never shines.

The garden pool lies and shines like the magic gall mirror.

The Magic Gall Mirror was a square of glittering, polished metal supposed to possess the miraculous power of betraying the thoughts of all who looked into it, by making the heart and "five viscera" visible. The ferocious First Emperor used it to examine his numerous Palace women, and those who, by a palpitating gall, showed lack of faith were put to death.

The Golden Valley is not much to boast of.

A beautiful garden built by the rich and eccentric Shih Ch'ung (died A.D. 300) for his favourite concubine Lü Chu.

This is the hour when the crows come to roost on the Ku Su Terrace.

(See Note 73.)

The silver-white arrow-tablet above the gold-coloured brass jar of the water-clock marks the dripping of much water .

(See Note 22.)

The shrill notes of the bamboo flute reached to Mien and O.

Mien and O are the ancient names for Hankow and Wuchang.

Perpetually casting my fish-line like Yen Ling.

Yen Ling is one of the names of the philosopher Yen Kuang. (See Note 75.)

THE REST-HOUSE OF DEEP TROUBLE

At Chin Ling, the tavern where travellers part is called the Rest-House of Deep Trouble.

An inn fifteen li South of the district in which Chin Ling (Nanking) stands.

Like K'ang Lo I climb on board the dull travelling boat.

K'ang Lo is a pseudonym for the poet Hsieh Ling-yün, who lived in the Fifth Century A.D.

I hum softly "On the Clear Streams Flies the Night Frost."

A line from one of Hsieh Ling-yün's poems.

It is said that, long ago, on the Ox Island Hill, songs were sung which blended the five colours.

The "five colours" are blue-green, yellow, carnation, white, and black. Anything that is perfectly harmonious is spoken of figuratively as being blended like the five colours.

Rapids flow past the Ox Island Hill on the Yangtze, which is not to be confused with the Ox Hill at the Yangtze Gorges.

Now do I not equal Hsieh, and the youth of the House of Yüan?

Yüan Hung lived in the time of the Chin Dynasty. His poems were both erudite and beautiful, but his extreme poverty forced him to take a position on a freight-boat plying up and down the Yangtze. One night, as the vessel lay below the dangerous Ox Rapids waiting for daylight, the official of the place, a learned man named Hsieh Shang, heard Yüan Hung's exquisite songs and was so delighted that he insisted upon the singer's accompanying him to the Official Residence. Here the days and nights were passed in conversation, and upon Yüan Hung's departure, Hsieh gave him much silver and gold, and eventually used his influence to enable the young man to become an official. Since then all men have heard of Yüan Hung. Li T'ai-po compares his lonely lot to that of the youth who possessed a faithful friend.

The bitter bamboos make a cold sound, swaying in the Autumn moonlight.

The ancient Chinese divided bamboos into two classes: the bitter and the tasteless.

In the attitude, and with the manner, of the woman of old.

A reference to a legend of a woman who was turned to stone. (See Note 45.)

Her resentment is that of the Woman of the Hsiang River.

O Huang and her sister Nü Ying were the wives of Shun, the "Perfect Emperor" (2317-2208 B.C. ). When he died, and was buried near the Hsiang River, they wept so copiously over his grave that their tears burned spots on the bamboos growing there, and thus was the variety [206] known as the "spotted bamboo" created. Eventually the despairing ladies committed suicide by throwing themselves into the river.

Her silence that of the concubine of the King of Ch'u.

Ts'u Fei, concubine of the King of Ch'u was much distressed because her lord was of a very wild disposition, and only took pleasure in hunting and such pursuits. She constantly expostulated with him on his mode of life, but at last, finding that all her entreaties were in vain, she ceased her remonstrances and sank into a silence from which she could not be roused.

Besides there are the "embroidered character letters."

In the Fourth Century A.D. , a lady, whose maiden name was Su, embroidered a long lament of eight hundred and forty characters in the form of a poetical palindrome and sent it to her husband who was exiled in Tartary.

The Heavens have revolved. The "Northern Measure" hangs above the Western wing.

The "Northern Measure" is the Chinese name for the "Dipper," and on the fifteenth day of the Eighth Month, when it can be seen sinking in the West before bed-time, a festival is held. This is essentially a festival for women, who object to being parted from their husbands at that time. Incense is burned to the full moon, and many fruits and seeds, all of a symbolical nature denoting the desire for posterity, are set out for the moon goddess.

In the Gold House, there is no one.

(See Note 23.)

The tones of the Chao psaltery begin and end on the bridge of the silver-crested love-pheasant.

"The sê , or psaltery, is made on the principle of the ch'in , and like that instrument has been made the subject of numerous allegorical comparisons. The number of strings has varied ... but the sê now in use has twenty-five strings. Each string is elevated on a movable bridge. These bridges represent the five colours: the first five are blue, the next red, the five in the middle are yellow, then come five white, and lastly five black." ("Chinese Music," by J. A. Van Aalst.) The most desirable specimens came from Chao, a place in Shensi. (See map.) The allusion to the love-pheasants is, of course, symbolical. By it, the lady says that this instrument is only properly used for love-songs, with the implication that it is therefore impossible for her to play it now.

I wish I could play my Shu table-lute on the mandarin duck strings.

The ch'in , or table-lute, lies on a table like a zither, and is played with the fingers. It is "one of the most ancient instruments, and certainly the most poetical of all.... The dimensions, the number of strings, the form, and whatever is connected with this instrument had their principles in Nature. Thus the ch'in measured 3.66 feet, because the year contains a maximum of 366 days; the number of strings was five, to agree with the five elements; the upper part was made round, to represent the firmament; the bottom was flat, to represent the ground; and the thirteen studs stood for the twelve moons and the intercalary moon. The strings were also subjected to certain laws. The thickest string was composed of two [208] hundred and forty threads and represented the Sovereign." ("Chinese Music," by J. A. Van Aalst.) The "Shu table-lute" is an allusion to Ssŭ Ma Hsiang-ju, a great poet and musician, who was a native of Shu. The mandarin ducks are emblems of conjugal love, and in speaking of them the wife expresses the wish that her husband were present to listen.

I wish my thoughts to follow the Spring wind, even to the Swallow Mountains.

The Yen Jan, or "Swallow Mountains," lie several thousand miles to the West of Ch'ang An, in Central Asia.

The neglected lamp does not burn brightly.

The lamps were little vessels filled with natural oil, upon which floated a vegetable wick. Unless constantly attended to, and this was the duty of the woman, the flame was small and insignificant.

The Mantzŭ are an aboriginal tribe still living in the far Southwest of China. It was here that Li T'ai-po was to have been exiled had not the sentence been commuted. (See Introduction.)

Instead, for me, the "long" rest-houses alternate with the "short" rest-houses.

On the "great roads," which we should speak of as paths, rest-houses for the convenience of travellers are erected every five li (a li is one-third of a mile). These are called "short road rest-houses" and are simply shelters. There are also "long road rest-houses" every ten li , where the care-takers serve travellers with tea and food, and which are equipped with altars and idols for the convenience of the pious.

AT THE YELLOW CRANE TOWER, TAKING LEAVE OF MÊNG HAO JAN

I take leave of my dear old friend at the Yellow Crane Tower.

Mêng Hao Jan ( A.D. 689-740) was a very famous poet, one of whose idiosyncrasies was riding a donkey through the snow in a search for inspiration.

The Yellow Crane Tower is still standing at Wuchang.

(See map and Note 60.)

Li Ling is buried in the sands of Hu.

Li Ling lived during the reign of the Emperor Wu of Han (140-87 B.C. ) at a time when the Hsiung Nu tribes were very troublesome. He penetrated far into the Hsiung Nu country, with a force of only five thousand infantry, and was there surrounded by thirty thousand of the enemy. After his men had exhausted their arrows, he was forced to surrender, and spent the rest of his life as a captive in Central Asia.

Su Wu has returned to the homes of Han.

Su Wu lived during the same period as did Li Ling, and was sent by the Emperor Wu upon a mission of peace to the Hsiung Nu. By the time he reached the Court of the Khan, however, relations between the Chinese and the Barbarians were again strained, and he was taken prisoner. Various attempts were made to induce him to renounce his allegiance to China; he was thrown into prison and subsisted for days on the moisture which he sucked from his clothes, but all efforts to undermine his loyalty failed, and eventually he was sent to tend sheep on the grazing fields of the steppes. Years passed, Wu Ti, the "Military Emperor," died, and his successor [210] Chao Ti made peace with the Central Asian tribes and sent envoys to ask for the return of the faithful Su Wu. The Khan replied that he was dead, but the envoy was able to answer that such could not be the case, as, not long before, the Emperor himself while hunting in his park had shot a wild goose, and had found a letter from Su Wu tied to its leg. The loyal official was therefore sent back to China. He had gone off in the prime of life; when he returned, in 86 B.C. , he was a broken-down, white-haired old man.

Wild geese are flying. If I sent a letter—so—to the edge of Heaven.

An allusion to the story of Su Wu. Letters anxiously awaited are often spoken of as "wild-goose" letters.

SAYING GOOD-BYE TO A FRIEND WHO IS GOING TO THE PLUM-FLOWER LAKE

I bid you good-bye, my friend, as you are going on an excursion to the Plum-Flower Lake.

This lake lies about seven miles Southwest of Nanking. The legend is that, many years ago, a raft loaded with flowering plum-trees sank in it, and ever since, during the plum-blossom season, the lake is covered with plum-trees in bloom.

Nevertheless you must not omit the wild-goose letter.

(See Notes 121 and 122.)

Or else our knowledge of each other will be as the dust of Hu to the dust of Yüeh.

Hu is the Mongols' country to the North and West of the Great Wall, and Yüeh is the province of Chêkiang in the Southeast of China. (See map.)

I had already wandered away from the People's Temple.

The Fêng Hsien is one of the so-called Chao Ti temples. These temples are erected by the people, not by Imperial command, which fact is proclaimed on an inscription written on a horizontal board placed over the main doorway. The Fêng Hsien temple stands in the Lung Mên, or Dragon Gate, a defile cut in the mountains of Honan by the great Yü when he drained the Empire about two thousand B.C. (See Introduction.) He is supposed to have been helped by a dragon who, with one sweep of its tail, cleft the mountain range in two, thus forcing the river I, a confluent of the Lo which is one of the tributaries of the Yellow River, to confine itself within the defile through which it runs in a series of rapids.

CROSSING THE FRONTIER—II

Sadness everywhere. A few sounds from a Mongol flageolet jar the air.

The Hsiung Nu soldiers, against whom the Chinese are fighting, are so near that the sounds of their flageolets can be plainly heard.

Perhaps it is Ho P'iao Yao.

(See Note 4.)

The demons where you are rejoice to see men go by.

The demons are of the man-eating variety, the yao kuai . (See table of supernatural beings in Introduction.)

You should hold speech with the soul of Yüan.

Ch'ü Yüan (see Note 62) drowned himself in the Mi Lo River.

And remembering Ko Hung, you are ashamed.

Ko Hung, author of "Biographies of the Gods," lived in the Fourth Century A.D. Although very poor, he pursued his studies with such zeal that he became an official. Having heard that the cinnabar, from which the Elixir of Immortality is distilled, came from Cochin China, he begged to be appointed to a magistracy in the South in order that he might obtain a supply for experimental purposes on the spot. Arrived in Kwangtung, he spent his time on Mount Lo Fo attempting to compound this elixir, and so, working at his experiments, passed into a tranquil sleep. When his friends went to wake him, they found his clothes empty. Ko Hung had ascended to the Taoist Paradise to live forever among the Immortals.

HEARING THE EARLY ORIOLE

The sun rose while I slept. I had not yet risen.

The poem alludes to the curious Chinese custom of holding Imperial audiences at dawn. This custom was persisted in until the fall of the Manchu Dynasty in 1912. One of the most noticeable peculiarities of Peking in Imperial days was the noise during the night, which never seemed to stop. Officials came to the Palace in their carts, while it was still dark, in order to be ready for the audience at dawn. It is clear from Po Chü-i's poem that he is no longer in office, since, although the sun has risen, he himself is still in bed.

AN IMPERIAL AUDIENCE AT DAWN

At the first light of the still-concealed sun, the Cock-man, in his dark-red cap, strikes the tally-sticks and proclaims aloud the hour.

The Cock-men, whose badge of office was a red cloth, were in charge of the water-clock, and their business was to announce the time of day. Near the water-clock were kept bamboo tallies, one for each division of the twenty-four hours. (See Introduction.) When the arrow of the water-clock registered the moment of the change from one division into another, the Cock-man on duty struck the appropriate tally-stick on a stone set for that purpose beside the door of the Palace. At sunrise, which took place during the hour of the monkey (three to five A.M. ) or during the hour of the cock (five to seven A.M. ), according to the season, he gave a loud, peculiar cry to warn the inmates of the Palace that day had come.

At this exact moment, the Keeper of the Robes sends in the eider-duck skin dress, with its cloud-like curving feather-scales of kingfisher green.

The "Keeper of the Robes" was one of the six offices instituted by the Ch'in Dynasty (255-209 B.C. ), the other five were those of the "Imperial Head-dresses," "Food-stuffs," "Washing Utensils," "Sitting Mats," and "Writing Materials." Robes were, and are, made from the skins of the various eider-ducks found in Northern Asia. The king eider's head is blue; the Pacific eider's, black and green; while the spectacled eider has a white line round the eye, which accounts for its name. The feathers are so close and soft that garments made of them feel exactly like fine fur.

In the Ninth Heaven, the Ch'ang Ho Gate opens.

The Ninth Heaven is the centre from which the points of the compass radiate, and it is there that the first of all the entrances to Heaven, the Ch'ang Ho Gate, stands.

The immediately-arrived sun tips the "Immortal Palm."

The "Immortal Palm" was a very tall bronze pillar [214] which the Emperor Wu of Han erected in the grounds of the Variegated Colours Palace. On the top was a colossal hand, with the fingers curled up so that the falling dew might be caught in the palm, for, of course, the ancient Chinese firmly believed that dew fell. As dew was the drinking-water of the Immortals, to drink it was to advance a step on the road to Immortality. The hand was brightly polished, and was one of the first objects about the Palace to glitter when the sun rose.

SEEKING FOR THE HERMIT OF THE WEST HILL

On the Nothing-Beyond Peak, a hut of red grass.

Huts were built of a certain hill grass, now very rare. It turns red in the Autumn, and is fine and strong like wire.

I look into the room. There is only the low table and the stand for the elbows.

Much of the furniture in the T'ang period was like that used now by the Japanese. It was customary to sit on the floor and write at a low table, and the use of the elbow-stand was general.

I have received much—the whole doctrine of clear purity.

The principles of Taoism are called literally "the clear pure doctrines."

Why should I wait for the Man of Wisdom?

An allusion to the eccentric Wang Hui-chih ( A.D. 388), who made a long journey through the snow to see a friend, but missed him.

The sacredness with which the Chinese regard their family ties is well known, but it is perhaps not realized [215] that the Chinese conception of the duties owed to friendship entails very great responsibilities. If a friend dies, it is a man's duty to see that his family do not suffer in any way. Wei Ying-wu is probably addressing the daughter of some dead friend whom he has brought up in his own family, or she may be a poor relation on his mother's side, but that she is not his own daughter is clear from the fact that her clan name differs from his, which is Wei.

ONCE MORE FIELDS AND GARDENS.

But for thirteen years it was so I lived.

The text reads "three ten," which is the way the Chinese say "thirty," but native commentaries state that it should read "ten three," or thirteen. This is far more in accordance with the facts of T'ao's life. He lived A.D. 365-427, and although he became an official, he soon resigned his post, saying that he "could not crook the hinges of his back for five pecks of rice a day." (See Note 54.)

Mine is a little property of ten mou or so.

A mou is a Chinese land measurement which is equal to about one-sixth of an acre.

A very famous song written during the Liang Dynasty ( A.D. 502-557). Allusions to it always suggest homesickness.

There seems to be no doubt that although King Hsüan of Chou (876-781 B.C. ) is not mentioned by name in the poem, which appears in the "Decade of Tang" division [216] of the "Book of Odes," he is the King referred to. All the old Chinese commentators agree in ascribing the authorship to a certain Jêng Shu, an officer of the Court during the reign of that monarch, who is known to have had a profound admiration for the King. Opinions differ as to the exact date of the great drought, but the standard chronology places it in the sixth year of King Hsüan's reign, 821 B.C. This ode illustrates the Chinese conception of kingship described in the Introduction.

How the Cloudy River glitters.

The Chinese call the Milky Way the "Cloudy" or "Silver River." Stars are peculiarly bright and glittering during a drought.

My stone sceptres and round badges of rank.

The badges of office were made of nephrite. There are references in both the "Book of History" and the "Book of Odes" to the fact that, after certain sacrifices, they were buried in the ground. In this case, the sacrifices had been performed so often that the supply of these tokens was exhausted.

I myself have gone from the border altars to the ancestral temples.

According to Confucius, the sacrifices to Heaven and Earth were performed at the border altars, and those to the ancestors took place at the temples especially provided for the purpose.

Hou Chi could do no more.

Hou Chi is the deity of grain, and from him King Hsüan was supposed to be descended.

Shang Ti does not look favourably upon us.

Shang Ti, literally the "Above Emperor," is the supreme ruler of the universe. Earthly Emperors receive the decree which empowers them to rule from him.

Why should I not be terrified Since all the ancestral sacrifices will be ended?

To the Chinese, this is the greatest calamity that can be conceived, since without these sacrifices the ancestral spirits would suffer greatly, and might visit their wrath upon their descendants.

Drought, the Demon of Drought, has caused these ravages.

The "Book of Spirits and Prodigies" states that in the Southern regions there is a hairy man, two or three cubits in height, with eyes in the top of his head and the upper part of his body bare. His name is Po. He runs with the speed of the wind, and in whatever part of the country he appears a great drought ensues.

I offered the yearly sacrifices for full crops in good time.

It was the custom for the King to pray and make offerings to Shang Ti during the first Spring month (February), in order to propitiate this chief of the Chinese pantheon and ensure good harvests from the grain then being sown. During the first Winter month (November), other prayers and sacrifices were offered to the "Honoured Ones of Heaven" (the sun, moon, and stars) for a blessing on the year to follow.

I neglected not one of the Spirits of the Four Quarters of the Earth.

Sacrifices of thanksgiving to the "Spirits of the Four Earth Quarters" were offered at the end of the harvest season.

SONG OF GRIEF

Pan Chieh-yü, the talented and upright concubine of the Han Emperor, Ch'êng, is one of the ladies most often referred to in literature. She was supplanted by the beautiful, but unscrupulous, "Flying Swallow," who accused her to the Emperor of denouncing him to the [218] kuei and the shên . (See table of supernatural beings in Introduction.) The Emperor, therefore, sent for Pan Chieh-yü who, kneeling before him, answered him as follows: "The Unworthy One of the Emperor has heard that he who cultivates virtue still has not attained happiness or favour. If this be so, for him who does evil what hope is there? Supposing that the demons and spirits are aware of this world's affairs, they could not endure that one who was not faithful to the Emperor should utter the secret thoughts hidden in the darkness of his heart. If they are not conscious of this world's affairs, of what use would the uttering of those secret thoughts be?" Then, rising, she left the Imperial presence, and immediately obtained permission to withdraw from the Palace. Not long after, she sent the Emperor "A Song of Grief," and ever since then the term, "Autumn Fan," has been used to suggest a deserted wife.

LETTER OF THANKS FOR PRECIOUS PEARLS

One of the ladies swept aside by Yang Kuei-fei (see Note 30) was the lovely Chiang Ts'ai-p'in, known as the "Plum-blossom" concubine. As she liked to differ from other people, she painted her eyebrows in the shape of wide cassia-leaves instead of the thin-lined willow-leaf, or "moth-antennæ," the form so much used. Soon after her departure from the Palace, some pearls were received as tribute, and the Emperor, who still had a lingering regard for "Plum-blossom," sent them to her in secret. She refused the pearls, and returned them to the Emperor with this poem.

I gaze far—far—for the Seven Scents Chariot.

The "Seven Scents Chariot" was a kind of carriage used [219] in old days by officials, and only those above the sixth rank might hang curtains upon it. It was open on four sides, but covered with a roof. The hubs of the wheels were carved. Ai Ai implies that the person she is waiting for is very grand indeed.

This song, which was probably written about 600 B.C. , has been elucidated by succeeding generations of Chinese commentators in the following tale.

The lady was a daughter of the Lord of Wei, and the divorced wife of the Lord of Sung. On the death of her husband, her son succeeded to his father's position as feudal chief of Sung. Because of her divorce, the unhappy woman, who was deeply attached to her son, was forbidden to enter Sung, where he lived.

The lamp-flower falls.

An old-fashioned Chinese lamp was simply a vessel in which a vegetable wick floated in oil. If the oil were very pure, the wick burned evenly, leaving no charred end; but if the oil were impure, the wick turned red-hot and formed a glowing tip called the "lamp-flower." Its appearance was looked upon as the happy omen which foretold a lover's speedy return.

But what is the rain of the Sorceress Gorge.

The Sorceress Gorge (see Note 87) is often referred to in a figurative sense, as it is in this poem. The allusion is to the story of a certain prince who dreamed that a fairy, calling herself the Lady of the Sorceress Mountain, came and passed the night with him. On leaving in the morning, she told him that it was she who ruled over the [220] clouds and rain, which would ever after be symbols of their love. Since then, the expression "clouds and rain" has become a euphemism for the relation of the sexes.

The writing of Li Po-hai.

Li Yung ( A.D. 678-747) is often called "Po Hai" in reference to a place where he held office. He was a person who displayed astounding knowledge at a very early age, and rose to be very powerful. When he was nearly seventy, he was overthrown by the machinations of his enemies and put to death. He wrote many inscriptions and was noted for his beautiful, spirited calligraphy.

The writing of Chia, the official.

Chia K'uei ( A.D. 30-101) was known as the "Universal Scholar." He was an eminent teacher, and many of his pupils came from great distances. As the payment he received was in grain, he was said to "till with his tongue," which phrase has now become a current expression for earning one's living as a teacher. Toward the end of his life, he was appointed Imperial historiographer. He was also a noted calligraphist. (See Note 77.)

Are many sweet-olive trees.

The olea fragrans , or sweet-olive, is employed in a metaphorical sense to denote literary honours. Scholars who have successfully passed their examinations are said to have gathered its branches.

Because the Yellow Emperor considers them of importance.

The Yellow Emperor is one of the five mythical sovereigns who ruled circa 2697 B.C. and is supposed to have reigned a hundred years.

THE SOLITARY TRAVELLER

He has attended an Imperial audience at the Twelve Towers.

The "Twelve Towers" was a palace built by Ming Huang (see Note 30) for the use of his ladies. It was an attempted imitation of a building supposed to have been erected by the Yellow Emperor (see Note 164) for the use of the Immortals. By his reference to it, one knows that the traveller has been to Court and is returning disappointed.

SPRING. AUTUMN. WINTER

It makes me think Of the Peach-Blossom Fountain.

An allusion to a well-known allegory, "The Peach-Blossom Fountain," by T'ao Yüan-ming. (See Note 142.) It tells how a fisherman, who was lost, found himself in a beautiful country where the people all wore strange clothes of very old-fashioned cut. On coming home, he told many stories about this enchanting land, but it could never be found again. The gods had permitted the fisherman to return for a short time to the "peach-blossom" days of his youth, although he could never remember the road he had taken, nor even point out the direction in which it lay.

PLAN OF A TYPICAL CHINESE HOUSE OF THE BETTER CLASS

KEY TO PLAN OF A TYPICAL CHINESE HOUSE OF THE BETTER CLASS

TABLE OF CHINESE HISTORICAL PERIODS

The Riverside Press CAMBRIDGE, MASSACHUSETTS U.S.A

The following pages contain advertisements of books by the same author

Books by Amy Lowell

PUBLISHED BY

PICTURES OF THE FLOATING WORLD

CAN GRANDE'S CASTLE

MEN, WOMEN AND GHOSTS

SWORD BLADES AND POPPY SEED

A DOME OF MANY-COLOURED GLASS

( IN COLLABORATION WITH FLORENCE AYSCOUGH )

FIR-FLOWER TABLETS: POEMS TRANSLATED FROM THE CHINESE

TENDENCIES IN MODERN AMERICAN POETRY

SIX FRENCH POETS: STUDIES IN CONTEMPORARY LITERATURE

Can Grande's Castle

By AMY LOWELL

Fourth edition

"The poems In 'Can Grande's Castle' are only four in number, but two of them ... touch magnificence. 'The Bronze Horses' has a larger sweep than Miss Lowell has ever attempted; she achieves here a sense of magnitude and time that is amazing.... Not in all contemporary poetry has the quality of balance and return been so beautifully illustrated."— Louis Untermeyer in The New Era in American Poetry .

"'Can Grande's Castle' challenges, through its vividness and contagious zest in life and color, an unreluctant admiration ... its rare union of vigor and deftness, precision and flexibility, imaginative grasp and clarity of detail."— Professor John Livingston Lowes in Convention and Revolt in Poetry .

"'Sea-Blue and Blood-Red' and 'Guns as Keys: and the Great Gate Swings' ... are such a widening of barriers they bring into literature an element imperceptible in poetry before ... the epic of modernity concentrated into thirty pages.... Not since the Elizabethans has such a mastery of words been reached in English ... one had never surmised such enchantment could have been achieved with words."— W. Bryher in The Art of Amy Lowell. A Critical Appreciation . London.

"The essential element of Miss Lowell's poetry is vividness, vividness and a power to concentrate into a few pages the spirit of an age. She indicates perfectly the slightest sense of atmosphere in a period or a city.... But the spirit of these poems is not the fashioning of pictures, however brilliant, of the past; it is the re-creation of epic moments of history made real as this present through her own individuality and vision."— The London Nation.

"We have come to it—once Poe was the living and commanding poet, whose things were waited for.... Now we watch and wait for Amy Lowell's poems. Success justifies her work.... Each separate poem in 'Can Grande's Castle' is a real and true poem of remarkable power—a work of imagination, a moving and beautiful thing."— Joseph E. Chamberlain in The Boston Transcript .

"'Can Grande's Castle' is, in the opinion of the present reviewer, not only the best book which Miss Lowell has so far written, but a great book per se.... It is a frank and revealing book. It deals with fundamentals.... In 'Sea-Blue and Blood-red' we have the old story of Nelson and 'mad, whole-hearted Lady Hamilton' retold in a style that dazzles and excites like golden standards won from the enemy passing in procession with the sun upon them."— The New York Times Book Review.

Second Printing

"I read 'Legends' last night, and again this morning. I like them the best of all your poems.... I like best Many Swans , which I have read twice and which I feel really speaks inside my unexplained soul. I should not like to try to explain it, because of the deep fear and danger that is in it. But it isn't a myth of the sun, it is something else. All the better that we can't say offhand what. That means it is true. It rings a note in my soul."—D. H. Lawrence.

"The subjects fit the poet like a glove.... The book is highly original, immensely interesting, and in its choice of themes, of the first significance."—Prof. John Livingston Lowes in The New York Evening Post .

"These clever dramatic tales are so brilliantly successful that we can only hope for more of their kind. Here is a canvas broad enough for the strokes of that untiring brush! Both in subject-matter and technique Miss Lowell has surpassed herself in these legends."—John Farrar in The Bookman .

"Miss Lowell builds—or composes—her poems as well as a painter of the first rank.... Her verse becomes increasingly supple.... I cannot say pompously that this latest volume contains Miss Lowell's best work, but it contains her work that I like best.... She is, at any rate, one of the three graces or nine muses upon whom our poetry stands or falls."—Malcolm Cowley in The Dial .

"There is no writer in America to-day, of either prose or poetry, who can manage such brilliant color effects in description.... In 'Legends' she has produced weirdly beautiful work that could never by any possibility be mistaken for the work of anyone else."—William Rose Benét in The Yale Review .

"'Legends' is, I think, Miss Lowell's best book ... the book that achieves the idiom, the convention that makes her work integral."—Padraic Colum in The Freeman .

Sword Blades and Poppy Seed

Fifth edition

OPINIONS OF LEADING REVIEWERS

"Against the multitudinous array of daily verse our times produce this volume utters itself with a range and brilliancy wholly remarkable. I cannot see that Miss Lowell's use of unrhymed vers libre has been surpassed in English. Read 'The Captured Goddess,' 'Music' and 'The Precinct. Rochester,' a piece of mastercraft in this kind. A wealth of subtleties and sympathies, gorgeously wrought, full of macabre effects (as many of the poems are) and brilliantly worked out. The things of splendor she has made she will hardly outdo in their kind."— Josephine Preston Peabody , The Boston Herald .

"For quaint pictorial exactitude and bizarrerie of color these poems remind one of Flemish masters and Dutch tulip gardens; again, they are fine and fantastic, like Venetian glass; and they are all curiously flooded with the moonlight of dreams.... Miss Lowell has a remarkable gift of what one might call the dramatic-decorative. Her decorative imagery is intensely dramatic, and her dramatic pictures are in themselves vivid and fantastic decorations."— Richard Le Gallienne , New York Times Book Review .

"Such poems as 'A Lady,' 'Music,' 'White and Green,' are wellnigh flawless in their beauty—perfect 'images.'"— Harriet Monroe , Poetry .

"Her most notable quality appears in the opening passage of the volume. The sharply etched tones and contours of this picture are characteristic of the author's work.... In 'unrhymed cadence' Miss Lowell's cadences are sometimes extremely delicate, as in 'The Captured Goddess.'"— Arthur Davison Ficke , Chicago Dial .

Tendencies in Modern American Poetry

Fourth Printing, illustrated

"I have no hesitation in insisting that Miss Amy Lowell's 'Tendencies in Modern American Poetry' is one of the most striking volumes of criticism that has appeared in recent years."— Clement K. Shorter in The Sphere , London.

"In her recent volume, 'Tendencies in Modern American Poetry,' Miss Lowell employs this method (the historical) with excellent results.... We feel throughout a spirit of mingled courage, kindness, and independence illuminating the subject, and the result is the note of personality that is so priceless in criticism, yet which, unhoneyed on the one hand or uncrabbed on the other, is so hard to come by ... her latest book leaves with the reader a strong impression of the most simple and unaffected integrity."— Helen Bullis Kizer in The North American Review .

"A new criticism has to be created to meet not only the work of the new artists but also the uncritical hospitality of current taste.... That is why a study such as Miss Amy Lowell's on recent tendencies in American verse is so significant.... Her very tone is revolutionary.... Poetry appears for the first time on our critical horizon ... as a sound and important activity of contemporary American life."— Randolf Bourne in The Dial .

"Its real worth as criticism and its greater worth as testimony are invaluable."— O. W. Firkins in The Nation .

"The feeling she has for poetry is so genuine and catholic and instructed, and her acquaintance with modern activity so energetic, that she is one of the most interesting and illuminating persons with whom to visit the new poets, led by the hand."— New Republic .

Men, Women, and Ghosts

"... In the poem which gave its name to a previous volume, 'Sword Blades and Poppy Seed,' Miss Lowell uttered her Credo with rare sincerity and passion. Not since Elizabeth Barrett's 'Vision of Poets' has there been such a confession of faith in the mission of poetry, such a stern compulsion of dedication laid upon the poet. And in her latest work we find proof that she has lived according to her confession and her dedication with a singleness of purpose seldom encountered in our fluid time.

"'Men, Women, and Ghosts' is a book greatly and strenuously imagined.... Miss Lowell is a great romantic.... She belongs to the few who, in every generation, feel that poetry is a high calling, and who press undeviatingly toward the mark. They are few, and they are frequently lonely, but they lead."— New York Times Book Review .

"... 'The Hammers' is a really thrilling piece of work; the skill with which it is divided into different moods and motifs is something more than a tour de force. The way the different hammers are characterized and given voice, the varying music wrung from them (from the ponderous banging of the hammers at the building of the 'Bellerophon' to their light tapping as they pick off the letters of Napoleon's victories on the arch of the Place du Carrousel), the emphasis with which they reveal a whole period—these are the things one sees rarely."— Louis Untermeyer in the Chicago Evening Post .

"... Beautiful ... poetry as authentic as any we know. It is individual, innocent of echo and imitation, with the uniqueness that comes of personal genius.... Miss Lowell strives to get into words the effects of the painter's palette and the musician's score. And life withal. Does she succeed? I should say she does, and the first poem in this book, 'Patterns,' is a brilliant, æsthetic achievement in a combination of story, imagism, and symbolism. 'Men, Women, and Ghosts' is a volume that contains beautiful poetry for all readers who have the root of the matter in them."— Reedy's Mirror , St. Louis.

"The most original of all the young American writers of to-day."— The New Age , London.

"Brilliant is the term for 'Men, Women, and Ghosts'—praise which holds good when the book is put to the test of a third reading."— Edward Garnett in The Atlantic Monthly .

A Dome of Many-Coloured Glass

"These poems arouse interest, and justify it by the result. Miss Lowell is the sister of President Lowell of Harvard. Her art, however, needs no reflection from such distinguished influence to make apparent its distinction. Such verse as this is delightful, has a sort of personal flavour, a loyalty to the fundamentals of life and nationality.... The child poems are particularly graceful."— Boston Evening Transcript , Boston, Mass.

"Miss Lowell has given expression in exquisite form to many beautiful thoughts, inspired by a variety of subjects and based on some of the loftiest ideals....

"The verses are grouped under the captions 'Lyrical Poems,' 'Sonnets,' and 'Verses for Children.'...

"It is difficult to say which of these are the most successful. Indeed, all reveal Miss Lowell's powers of observation from the view-point of a lover of nature. Moreover, Miss Lowell writes with a gentle philosophy and a deep knowledge of humanity....

"The sonnets are especially appealing and touch the heart strings so tenderly that there comes immediate response in the same spirit....

"That she knows the workings of the juvenile mind is plainly indicated by her verses written for their reading."— Boston Sunday Globe , Boston, Mass.

"A quite delightful little collection of verses."— Toronto Globe , Toronto, Canada.

"The Lyrics are true to the old definition; they would sing well to the accompaniment of the strings. We should like to hear 'Hora Stellatrix' rendered by an artist."— Hartford Courant , Hartford, Conn.

"Verses that show delicate appreciation of the beautiful, and imaginative quality. A sonnet entitled 'Dreams' is peculiarly full of sympathy and feeling."— The Sun , Baltimore, Md.

Six French Poets

STUDIES IN CONTEMPORARY LITERATURE

Third edition, illustrated

A brilliant series of biographical and critical essays dealing with Emile Verhaeren, Albert Samain, Remy de Gourmont, Henri de Régnier, Francis Jammes, and Paul Fort, by one of the foremost living American poets. The translations make up an important part of the book, and together with the French originals constitute a representative anthology of the poetry of the period.

William Lyon Phelps , Professor of English Literature, Yale University, says:

"This is, I think, the most valuable work on contemporary French literature that I have seen for a long time. It is written by one who has a thorough knowledge of the subject and who is herself an American poet of distinction. She has the knowledge, the sympathy, the penetration, and the insight—all necessary to make a notable book of criticism. It is a work that should be widely read in America."

"In her 'Six French Poets' I find a stimulating quality of a high order.... I defy any English critic to rise from this book without the feeling that he has gained considerably. This is the first volume in English to contain a minute and careful study of these French writers."— Clement K. Shorter in The Sphere , London.

"I can conceive of no greater pleasure than that of a lover of poetry who reads in Miss Lowell's book about modern French poetry for the first time; it must be like falling into El Dorado."— F. S. Flint , formerly French critic of Poetry and Drama , London, in The Little Review .

"Amy Lowell's 'French Poets' ... ought to be labelled like Pater's studies 'Appreciations,' so full of charm are its penetrative interpretations ... and it is not too bold to say that her introductions to and interpretations of French poets will live as long as interest in these poets themselves lives. Her book is a living and lasting piece of criticism ... a masterly volume."— New York Sun.

"A very admirable piece of work."— The London Bookman.

"Une très interessante étude."— La France.

"An excellent book."— Emile Cammaerts in The Athenæum , London.

"Miss Lowell has done a real service to literature. One must be limited, indeed, who fails to appreciate the power of these writers as set forth through the comment, the discriminating extracts, and the appended prose translations in her book."— North American Review.

Transcriber's Notes:

Every effort has been made to replicate this text as faithfully as possible, including obsolete and variant spellings and other inconsistencies.

Obvious punctuation errors and minor printer errors repaired.

Moved "Books by Amy Lowell" ad page to end of book.

Deleted duplicate ad pages for "Can Grande's Castle" and "Tendencies in Modern American Poetry".

THE FULL PROJECT GUTENBERG LICENSE

This eBook is for the use of anyone anywhere in the United States and most other parts of the world at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at www.gutenberg.org . If you are not located in the United States, you will have to check the laws of the country where you are located before using this eBook.

• You pay a royalty fee of 20% of the gross profits you derive from the use of Project Gutenberg™ works calculated using the method you already use to calculate your applicable taxes. The fee is owed to the owner of the Project Gutenberg™ trademark, but he has agreed to donate royalties under this paragraph to the Project Gutenberg Literary Archive Foundation. Royalty payments must be paid within 60 days following each date on which you prepare (or are legally required to prepare) your periodic tax returns. Royalty payments should be clearly marked as such and sent to the Project Gutenberg Literary Archive Foundation at the address specified in Section 4, “Information about donations to the Project Gutenberg Literary Archive Foundation.”
• You provide a full refund of any money paid by a user who notifies you in writing (or by e-mail) within 30 days of receipt that s/he does not agree to the terms of the full Project Gutenberg™ License. You must require such a user to return or destroy all copies of the works possessed in a physical medium and discontinue all use of and all access to other copies of Project Gutenberg™ works.
• You provide, in accordance with paragraph 1.F.3, a full refund of any money paid for a work or a replacement copy, if a defect in the electronic work is discovered and reported to you within 90 days of receipt of the work.
• You comply with all other terms of this agreement for free distribution of Project Gutenberg™ works.

IMAGES

The Difference Between Paraphrasing And Summarizing
Difference Between Paraphrasing and Summarizing
paraphrasing versus summarizing
Factors That Differ Paraphrasing From Summarizing
The Difference Between Paraphrasing And Summarizing
Summary vs Paraphrase: Difference and Comparison

VIDEO

What is Paraphrasing? Everything You Need to Know #shortvideo
Writing Center Lessons: Quoting, Paraphrasing, and Summarizing
How to Quote, Paraphrase and Summarize in Research? Amharic
QUOTING,PARAPHRASING AND SUMMARIZING 2
Power up your Debates with Paraphrasing and Summarizing
Unlocking Effective Communication: Paraphrasing and Summarizing

COMMENTS

Paraphrasing vs. Summarizing (Differences, Examples, How To)
Summary: The article discusses paraphrasing vs. summarizing by explaining the two concepts. It specifies when you should use paraphrasing and when you should summarize a piece of text and describes the process of each. It ends with examples of both paraphrasing and summarizing to provide a better understanding to the reader.
Difference Between Summary and Paraphrase (with Comparison Chart)
The first and foremost difference between summary and paraphrase is that a summary is nothing but a shorter version of an excerpt or passage. On the contrary, a paraphrase is the restatement of the original text or excerpt. ... Make sure that the sentences and words used are your own and it should not be a mere substitution or swapping of words ...
What Is the Difference Between Paraphrasing and Summarizing?
Summarizing is correctly mirroring the original writer's message and main ideas. For both paraphrased passages and summaries, you must attribute the information back to the original source. . Knowing the difference between paraphrasing and summarizing is an important writing skill.
Quoting, Paraphrasing, & Summarizing
Quoting, paraphrasing, and summarizing are all different ways of including evidence and the ideas of others into your assignments. Using evidence from credible sources to support your thesis is an important part of academic writing. Citing the source of any quote, paraphrase, or summary is an important step to avoid plagiarism.
Paraphrasing Vs. Summarizing: The Difference And Best Examples
Paraphrasing vs Summarizing — Differences. Here are the 3 primary differences between paraphrasing and summarizing writing techniques: Paraphrasing is rewriting a text in your own words while summarizing is writing the most important parts of a piece in your own words.
Paraphrase and Summary
To summarize: Identify what reading or speech is being summarized. State the author's thesis and main claims of their argument in your own words. Just like paraphrasing, make sure everything but key terms is reworded. Avoid specific details or examples. Avoid your personal opinions about the topic. Include the conclusion of the original material.
2.7: Summarizing and Paraphrasing
A summary is written in your own words. It contains few or no quotes. A summary is always shorter than the original text, often about 1/3 as long as the original. It is the ultimate fat-free writing. An article or paper may be summarized in a few sentences or a couple of paragraphs. A book may be summarized in an article or a short paper.
Paraphrasing
Paraphrasing and summarizing are similar. When we paraphrase, we process information or ideas from another person's text and put it in our own words. The main difference between paraphrase and summary is scope: if summarizing means rewording and condensing, then paraphrasing means rewording without drastically altering length.
The Difference Between Summarizing & Paraphrasing
Paraphrasing is rephrasing something in your own words; the word comes from the Greek para -, meaning "beside" or "closely resembling", 1 combined with "phrase," which we know can mean a string of words or sentences. 2 Paraphrasing isn't practical for entire sources—just for when you want to highlight a portion of a source.
The difference between paraphrasing & summarizing
Differences between paraphrasing & summarizing Definition and Purpose. Paraphrasing involves rewriting someone else's ideas or a specific text in your own words while maintaining the original meaning and often keeping a similar length to the source material. The primary purpose of paraphrasing is to use another person's ideas in your work without resorting to a direct quotation, thereby ...
Paraphrasing & Summarizing
Paraphrasing allows you to use your own words to restate an author's ideas. Summarizing allows you to create a succinct, concise statement of an author's main points without copying and pasting a lot of text from the original source. What's the difference: Paraphrasing v. Summarizing. Explore the rest of the page to see how the same ...
The Writing Center
Paraphrase, summary, and analysis are important for accomplishing different jobs in the essay: Quotation allows the writer to fully use the original author's words using quotation marks in order to make a point or to provide support for an idea. Example: The professor said, "The sun is shining.". Paraphrase enables the writer to ...
Paraphrasing vs. Summarizing: What's The Difference?
Another difference between paraphrasing and summarizing is the length of the piece of writing that results from each process. When people paraphrase a written document, the paraphrased document usually ends up being a similar length to the original piece. Similarly, if someone paraphrases the events of a meeting or presentation, the length of ...
What is the difference between summarizing and paraphrasing
Both involve taking ideas, words, or phrases from a source and crafting them into new sentences within your writing. Summarizing is when you briefly state the main concept/s of a topic in your own words. Paraphrasing is when you briefly recount a specific passage from a piece in your own words. Whether summarizing or paraphrasing, credit is ...
Summarizing vs. Paraphrasing: What's the Real Difference?
Summarizing involves condensing a large amount of information into a concise version while maintaining the main points. On the other hand, paraphrasing involves rephrasing the text in your own words while retaining the original meaning. Summaries are shorter than the original text and omit details, while paraphrases are usually the same length ...
Quoting, Paraphrasing, and Summarizing
Paraphrasing involves putting a passage from source material into your own words. A paraphrase must also be attributed to the original source. Paraphrased material is usually shorter than the original passage, taking a somewhat broader segment of the source and condensing it slightly. Summarizing involves putting the main idea (s) into your own ...
Paraphrasing and Summarizing Flashcards
Paraphrasing may lead to text that is just as long as the original but is written using one's own words and language. Gist Summary. Focusing on the central idea, but identifying the who, what, where, when, why and how of a text. Gist summary is best for summarizing: Expository texts like newspaper articles or any text about an event.
What's the Difference? Summarizing, Paraphrasing, & Quoting
Oppositely, the main difference between quoting, paraphrasing, and summarizing is that quoting is done word for word from the original work. Both paraphrasing and summarizing only touch on the key points and are written with some variation from the initial author's work, usually in the style and tone of the new author.
Quoting vs. Paraphrasing vs. Summarizing
What is summarizing? Next, we come to summarizing. Summarizing is on a much larger scale than quoting or paraphrasing. While similar to paraphrasing in that you use your own words, a summary's primary focus is on translating the main idea of an entire document or long section. Summaries are useful because they allow you to mention entire chapters or articles—or longer works—in only a few ...
On the Difference Between Paraphrase & Summary
A paraphrase recreates a facsimile of the passage from the narrator's point of view. A summary describes the passage from the reader's point of view. In the book, we use the example of this short passage from To Kill a Mockingbird in which Scout is talking about her Aunt Alexandra who speaks a coded language about the importance of heredity ...
What is the difference between quotation, paraphrase, and summary
Paraphrasing is usually expected in research and argumentative essays. These type of papers benefit from paraphrasing because it shows that you understand the source and are therefore a reliable voice on that source. Paraphrasing can make the evidence more straightforward. Another reason to paraphrase is to adjust your tone for your audience.
Direct Quotes vs. Paraphrasing vs. Summarizing Explained
Learn the main differences between direct quotes, paraphrasing, and summarizing. Plus, see examples for each to ensure you follow best practices. Talk to Us. ... Changing the first letter of the quote to an upper or lowercase so that the quotation matches the context sentence's syntax. Can also modify the punctuation at the end of the quote.
7.1: Using and Organizing Information Details
It's a good way to describe what you've learned and also help the reader understand the significance of the information. Paraphrasing is a common writing technique, but it's also where many students unintentionally plagiarize. Students unintentionally plagiarize because they: Take a sentence from a source and rearrange the words.
Paraphrase
Paraphrase. A paraphrase or rephrase ( / ˈpærəˌfreɪz /) is the rendering of the same text in different words without losing the meaning of the text itself. [1] More often than not, a paraphrased text can convey its meaning better than the original words. In other words, it is a copy of the text in meaning, but which is different from the ...
What are Tone Words? List of 300+ Useful Words
The word for tone in literature is the mood or atmosphere that the author intends for the reader to experience while reading the text, while mood is the experience that the reader has while reading the story.. How the author feels about the setting or the character and how he wants the reader to feel determines the tone. On the contrary, the reader's emotional response establishes the mood.
Murder trial of O. J. Simpson
The People of the State of California v. Orenthal James Simpson was a criminal trial in Los Angeles County Superior Court, in which former National Football League (NFL) player and actor O. J. Simpson was tried and acquitted for the murders of his ex-wife Nicole Brown Simpson and her friend Ronald Goldman.The two were stabbed to death outside Brown's condominium in Los Angeles on June 12, 1994.
Beyond Trend and Periodicity:
This sentence is crucial as it provides temporal context that significantly impacts the prediction of both daily and annual periodicity. While other sentences receive moderate attention, the sixth sentence, which describes atmospheric pressure as detailed in Table 1, consistently receives no attention across all channels.
Evaluation of GPT-4 ability to identify and generate patient
Best practice guidelines recommend that clinicians communicate the presence of AIFs to the patient at the time of the ED visit and document this information into the patient's discharge instructions. 4 Although rates of verbal communication are not easily measured, specific documentation of AIFs into a patient's discharge summary are ...
An alternative theoretical approach to develop a new conception about
The theoretical approach presented in this paper describes a novel experimental-theoretical methodology to conceptualise pain in people with dementia. Existing procedures for assessment of pain rely on subjective self-report using pain questionnaires and rating scales that have proven to be highly problematic where a person has dementia. Consequently, pain in people with dementia can be ...
FIR-FLOWER TABLETS
The Chinese text, for rhyme-scheme and rhythm; the dictionary meanings of the words; the analyses of characters; and, for the fourth, a careful paraphrase by Mrs. Ayscough, to which she added copious notes to acquaint me with all the allusions, historical, mythological, geographical, and technical, that she deemed it necessary for me to know.

Paraphrasing vs. Summarizing (Differences, Examples, How To)

What is paraphrasing?

When should you paraphrase?

How to paraphrase effectively

Example of paraphrasing

What is summarizing?

When should you summarize?

How to summarize effectively

Example of summarization

Core lessons

Popular lessons

Stay awhile. Your weekly dose of grammar and English fun.

Difference Between Summary and Paraphrase

Content: Summary Vs Paraphrase

Definition of Paraphrase

Key Differences Between Summary and Paraphrase

Steps for Summarizing

Steps for Paraphrasing

You Might Also Like:

Leave a Reply Cancel reply

Paraphrase and Summary

When to Use Paraphrasing vs. Summarizing

Module 11: The Research Process—Using and Citing Sources

What is Paraphrasing?

What Does Good Paraphrasing Look Like?

The Difference Between Summarizing & Paraphrasing

Summarizing vs. Paraphrasing: The Biggest Differences

Write with Confidence using Editor

What is Summarizing?

How to Summarize

What is Paraphrasing?

How to Paraphrase

Get started with Microsoft 365

Topics in this article

What is independent publishing?

What are literary tropes?

What are genre tropes?

What is literary fiction?

Everything you need to achieve more in less time

Explore Other Categories

What is the difference between paraphrasing & summarizing

Detail and Length

Usage in Academic Writing

How to paraphrase in a few steps

Reading & Understanding the content

Taking Notes

Using Synonyms

Changing Sentence Structure

Rephrasing Concepts

Comparing with the original content

Cite the Source

Common paraphrasing mistakes

How to summarize in a few steps

Keeping it short and straightforward

Maintaining objectivity

Using a summarizing tool

Don’ts of paragraph summarization

What is paraphrasing?

What are the steps to paraphrase?

What are some tips to paraphrase?

What are the steps to summarize content?

The Sheridan Libraries

Paraphrasing & Summarizing

Definitions

What’s the difference: Paraphrasing v. Summarizing

Original Text

Compare the Original and Paraphrase

Original Text

Compare the Original and Summary

Cheat Sheet

Writing Center FAQs

Q: What is the difference between summarizing and paraphrasing?

Answer Last Updated: May 22, 2017 Views: 411

Related Topics

Summarizing vs. Paraphrasing: What’s the Real Difference?

Key Takeaways

Definition of Summarizing

Purpose of Summaries

Main Points in Summarizing

Steps in Summarizing