1. Intro

A well-written account is one that makes reading a pleasant experience. If you make a statement, an example to illustrate your point will make the reading less demanding; and if you draw a conclusion, preparing your reader step by step will make this result more acceptable. Whether you are writing a term paper or a journal article, never expect that things should be obvious for your audience, unless you are referring to the most routine aspects of daily life.

For a self-contained piece of writing, you need to provide explanations for your technical terms, concepts and abbreviations. Don't keep your readers guessing about such points, as they will be more likely to put down your text. Most writing skills, and most suggestions in this guide, derive from a single principle: Mind Your Reader.

In terms of contents, an academic paper is nearly always a report on your efforts: a description of your attempts to solve a research question. You will find more on research questions below – but for now, remember that solving one cannot be the purpose of your paper. After all, it is often impossible to know at the outset whether you can solve it at all. In short: reporting on your efforts is a must, solving your research question is a maybe.

Keep your goals in check, and remember that collecting and describing new data will constitute a contribution to science. Try not to overreach yourself: in your paper, you do not need present magic solutions to notorious linguistic dilemmas.

Reporting on your efforts also means that you will need to raise points that you have considered and left largely unexplored. A careful description of an unsolved problem is not a matter of disgrace, but of scientific virtue, since it helps to identify exactly those areas which require further exploration.

Contributions to science typically start from an original observation, which is really no more than you noticing the occurrence of some event. In order to be useful, observations require documentation; in other words, you need to record them carefully. Since linguistic expressions disappear as soon as they have been uttered, be prepared to note them down, together with details about speaker, hearer(s), place, time and other context. You can read more on the relevance (and art) of observation here.

Just as in other fields, a coherent piece of writing in linguistics starts from the statement of a problem, followed by the discussion of arguments which lead to conclusions. The statement of a problem is your research question, i.e. a matter that you raise for discussion and analysis. The arguments are the observations, claims and analyses which you discuss. The conclusions are the results of this discussion.

In your account, make sure to point out the relevance of the problems that you encounter and to share your enthusiasm. If you want, you can also add a personal note to your conclusions, e.g. an incidental discovery that you ran into, or a something new that you have learned on your journey of discovery.

The quality of your account depends on the clarity of your exposé. Try to place yourself in the position of readers who are interested, but unprepared. Keep asking yourself if your words really convey what you intended to say. Check your texts carefully, but don't do this right after you finish writing them.

There is an abundance of books about writing skills. Even without them, you can learn to write by appreciating the efforts of other writers. Read abundantly and read critically: this may benefit your own writing skills more substantially than you expect. In reading other work, try to assess which aspects appeal to you, and which aspects you find less appealing.

For the sake of clarity, try to minimize your use of abbreviations and symbols. If you have to use them, do explain them whenever you first use them in your text, and do not forget to include an alphabetic overview of these conventions in your text. Such an overview is best positioned before the main body of the text, so a reader will not have to wonder whether these explanations are forthcoming.

Footnotes are sadly overrated in academic writing; a good article does not require them. A footnote should never be used to explain terms and principles. These explanations, being crucial to your argumentation, belong in your main text. In linguistic texts, references to secondary literature are likewise best placed in the main body of the text, as will be explained in section 4.

Footnotes can be used for information which may be of interest to the reader despite being unrelated to the subject under discussion. For example:

(1)

The adverb bù 'not' also occurs with other adverbs, such as tài 'too, extremely'. In the next excerpt, speaker A's question promps a negative answer from speaker B:           A: Pǔbiàn ma? pervasive Q 'Is this common?' B:  M...  bú  tài  pǔpiàn.1 mm not too pervasive 'Mm... not especially common.' _________ 1. Both speakers regard pǔbiàn 'pervasive' and pǔpiàn 'pervasive' as "the same word". In Beijing Mandarin, it is pronounced as pǔbiàn; in Taiwan, pǔpiàn is more common.

If notes are used, footnotes (i.e. notes at the bottom of the page) are usually preferable to endnotes (which are listed following the main text) as they will save the reader a journey across pages, and back. In digital publications, this problem of paging back and forth can be solved by cross-linking between the text of a note and the main text.

2. Language about language

Writing about language is special because it has language play a double role, functioning as the object as well as the medium of description. This is metalanguage, in other words, language about language. To distinguish langage from metalanguage in your composition, a number of conventions are required.

In linguistic writing, italics are commonly used to cite the forms of a language, and quotation marks are used to transcribe the corresponding meanings:

(2)	English fine can be used in the sense of 'all right', but it can also mean 'fragile, delicate'.

(3)	In the French expressions style antique 'ancient style' and style moderne 'modern style', the adjectives antique 'ancient' and moderne 'modern' function as adjuncts following a head noun style 'style'.

As suggested by their name, quotation marks are also used to mark quotations cited from authors (for details, see §  3). In a linguistic essay, you can distinguish this usage of quotation marks from their use in the transcription of meaning, e.g. by placing citations between double quotation marks, and meanings between single quotation marks.

In handwritten notes, italics may be represented by underlining. This usage derives from the traditional use of underlining in manuscripts, where it served to warn the typesetter that a cursive typeface was needed. (And in digital type, underlining has evolved to mark clickable hyperlinks, as in the present document.)

When data are cited in the form of numbered examples, they often appear without italics:

(4)	Now compare the following examples: (27) How many people were there? (28) How many people were there there?

Here, italics can be dispensed with, as the numbering itself will signal to your reader that you are citing language data.  

More details on various ways of transcribing language data are discussed in the Introduction to linguistics transcription.

3. Quoting other works

Whenever you quote works by others, make sure to mark cited text explicitly as a quotation. If you fail to do so, you may run the risk of being accused of plagiarism, i.e. of passing off someone else's work as your own.

In citing other works, double quotations marks may be used: 

(5)	Abel-Rémusat's Élémens de la grammaire chinoise came in two parts, describing two Chinese languages: Classical Chinese and the Mandarin language of his time, which he distinguished as "style antique" (1857: 38) and "style moderne" (p. 107).

Note that the transcription here differs from that in example (3), where style antique and style moderne were italicized, since they served as French linguistic data. By contrast, the quotation marks in "style antique" and "style moderne" in example (5) marks these words as text cited from another author.

In example (5), the point was to illustrate which terms the original French text used, so a literal quote was needed. Still, quoting a text in another language never helps to make your text a smooth read. Instead, you may represent your sources with a paraphrase whenever the exact choice of words is less relevant. For instance, don't write this:

(6)	Abel-Rémusat identified 哉 zāi as "le plus souvent une particule interrogative qui se place à la fin de la phrase" (1857: 99).

Here, a reliable paraphrase will make a better read:

(7)	Abel-Rémusat noted that 哉 zāi was most often used as an interrogative particle placed at the end of the sentence (1857: 99).

As illustrated here, paraphrases (like citations) require exact page references, allowing interested readers to consult the original text.

Some texts are less suitable for citation purposes because of the way they happen to be phrased in their original context. For instance:  

(8)	Lyons (1977: 503) maintains that "Every statement that can be made by uttering a simple sentence expresses a proposition, which, if it is informative (cf. 2.1), provides the answer to either an explicit or an implicit question".

In example (8), the reference to section 2.1 of the original work will be irrelevant to your reader; and the capital letter E looks out of place in its new context. In such cases, consider the following:

• You may never make unannounced changes to the original text of your source.
• Extra text may be added only if you place it between square brackets.
• Irrelevant text may be deleted only if you mark its spot with three dots between square brackets: [...].

The same source reference can therefore be represented as follows:

(9)	Lyons (1977: 503) maintains that "[e]very statement that can be made by uttering a simple sentence expresses a proposition, which, if it is informative [...], provides the answer to either an explicit or an implicit question".

For longer quotes, a separate paragraph may be used, preferable indented, and/or in a a smaller font. This will remove the need of using quotation marks. Such quotations are often separated from the preceding and following context with blank lines:

(10)

Sapir (1949: 122-123) has described this type of inherent comparisons as follows:  Such contrasts as small and large, little and much, few and many, give us a deceptive feeling of absolute values within the field of quantity comparable to such qualitative differences as red and green within the field of color perception. This feeling is an illusion, however, which is largely due to the linguistic fact that the grading which is implicit in these terms is not formally indicated, whereas it is made explicit in such judgments as "There were fewer people there than here" or "He has more milk than I". [...] Many merely means any number, definite or indefinite, which is more than some other number taken as point of departure. This "point of departure" derives from the linguistic or non-linguistic context.

Even though cited text may not be changed, the use of square brackets will allow you to provide the reader with some help to interpret the quotation without having to consult the original:

(11)	Li (1981: 57) states that "to use an RVC [resultative verb compound], the agent must have initiated the primary action referred to by the compound [i.e. the meaning of the first element], while the use of néng 'can' only suggests the possibility of initiating the action".

Double-check your direct citations for accuracy. Avoid wrenching a citation from its context, and always treat your sources with respect. The word sic (Latijn for 'thus [it is written]') should never be used in a condescending manner. Don't write this:

(12)	According to Ebeling (1978: 278), the sentence he enjoys schooting [sic!] wild ducks can be analyzed as follows.

By way of exception, such obvious printing errors may even be corrected tacitly. In academic writing, the use of sic is justified only to indicate something which is so out of the ordinary that your reader might suspect a printing error. For example, this might apply to the French book title mentioned above, where modern French would spell éléments, not élémens:

(13)	Abel-Rémusat first pubished his Élémens [sic] de la grammaire chinoise in 1822.

4. Referencing sources

Quotations from a printed source should include a page reference, as shown in example (11):

Li (1981: 57)

Avoid general references like this:

Li (1981)

Excluding the page number would compel any reader who wishes to check the original text to undertake an extensive search to find the text you are citing.

For online resources, similar tactics apply. Don't expect your reader to explore an entire website, but make sure to specify the weblink or URL of your original text as narrowly as possible.

In many other disciplines, source references may be provided in footnotes. In linguistics, it is often preferable to use in-text references, as illustrated in Section 3.

This is because visually, a linguistic text risks making an overcrowded impression even without footnotes, due to its inevitable numbered examples, phonetic symbols and other diacritics, italics, quotations marks and other forms of transcription, all vying for the attention of the reader. In-text references will render the flow of your text more smooth, and make the reading more palatable.

A common form of in-text referencing is in the form of a short label: Author (year: page), as illustrated. If your reference happens to occur in text which itself appears within parentheses, an extra pair of brackets will not be needed:

(14)	English terms for these expressions include coverb (as preferred by DeFrancis 1976: 83) and preposition (Norman's term, 1988: 158).

Such labels refer to a separate section of references which should follow your main text, allowing your reader to identify the source marked by the label. For instance, bibliographical details for Norman's (1988) work cited in example (14) can be found in the following list of references. The labels are are ordered alphabetically:

	[label]	[bibliographical details]
(15)	Dubinsky & Holcomb (2011)	Stanley Dubinsky and Chris Holcomb, "The language police: Prescriptivism and standardization" = Chapter 11 in Understanding language through humor. Cambridge: Cambridge University Press, pp. 166-183.
	Halpern & Kerman (2002)	Jack Halpern and Jouni Kerman, "The pitfalls and complexities of Chinese to Chinese conversion / 汉字简繁转换的复杂性和陷阱 / 漢字簡繁轉換的複雜性和陷阱". Internet http://www.cjk.org/cjk/c2c/c2c.pdf, 7 March 2018.
	Iljic (1983)	Robert Iljic, "Le marqueur laizhe". In: Cahiers de linguistique Asie orientale XI N° 2 (1983), pp. 65-102.
	Norman (1988)	Jerry Norman, Chinese. Oxford: Oxford University Press, 1988.
	Sapir (1978)	Edward Sapir, Language: An introduction to the study of speech. First published by Rupert Hart-Davis Ltd, 1963. Frogmore: Granada Publishing, 1978.

This is one common way of arranging a reference section. For books, the formula used here is: First name or initials, Surname, Title (in italics). Place: Publisher, year. In example (15), this information is preceded by the label used in the main text, i.e. "Norman (1988)".

Editors and publishers often have their own lay-out specifications, but whenever you are not bound by these, any arrangement will do, provided that you adhere consistently to the formula of your choice.

In referencing book titles, the main titles and a subtitle are usually separated by a colon. Like the main title, the subtitle will have its first letter capitalized, as illustrated above.

Example (15) also illustrates a common way of referencing sources which are not entire books, such as book chapters (Dubinsky & Holcomb 2011) and journal articles (Iljic 1983). These titles are not italicized, but placed within quotation marks.

In such cases, the titles of the journal itself is italicized, and may be preceded by "In: ". References to journals usually do not include a place of publication or publishers's name, but you will need to provide the year of publication, the volume and/or number of the journal issue, and the page numbers of the article you are quoting.

The same example (15) illustrates a reference to an online source (Halpern & Kerman 2002) by specifying the URL. Since online data are continually subject to change, you have to state the date on which you found your materials at the given URL.

Also note that the title given for Halpern & Kerman (2002) is exceptional since it consists, in the original, of three different scripts (and two languages). This is here indicated by separating the three scripts by means of slashes.

If you are writing in English, you may provide translations of titles in other languages as a service to your reader. Such additions, like other additions from you as the author, are placed between square brackets:

(16)

Lǚ (1999)

吕叔湘 Lǚ Shūxiāng, ed. in chief, 现代汉语八百词 • 修订本 Xiàndài Hànyǔ bābǎi cí: Xiūdìng běn [Eight hundred words Modern Chinese: Revised edition]. 北京 Beijing: 商務印書館 Shāngwù Yìnshūguǎn, 1999. Original edition 1980.

5. Citing examples

Providing sufficient examples is a major reponsibility in writing about language. Sometimes a single example will effectively bring home whatever you are trying to make clear in a complicated text. An abundance of examples will stimulate your reader to think along with the linguistic points your are making, thus increasing the chances that your reader will keep on reading your text. In short, as the German saying goes, nur das Beispiel führt zum Licht, vieles Reden tut es nicht 'the example will elucidate whatever words cannot say'.

When you cite spoken forms from Mandarin or any other Chinese language in an English linguistic text, quoting Chinese characters will usually be superfluous.

In a linguistic text, the use of characters can often be confusing in the representation of forms. For instance, if you write 一本书, your reader (let alone the linguistic record) will not be informed whether this is pronounced as *yī běn shū, yì běn shū or yi běn shū; and which of these forms correspond to the two meanings of 'one book' and/or 'a book'.

Linguistic texts often provide glosses for example sentences. Linguistic glosses are simply word-by-word translations, sometimes even morpheme-by-morpheme translations. This was illustrated in example (1), repeated below. The first line quotes Mandarin forms, the second line presents the glosses, and the third line is an English translation of the whole sentence:

(1)

The adverb bù 'not' also occurs with other adverbs, such as tài 'too, extremely'. In the next excerpt, speaker A's question promps a negative answer from speaker B:           A: Pǔbiàn ma? pervasive Q 'Is this common?' B:  M...  bú  tài  pǔpiàn.1 mm not too pervasive 'Mm... not especially common.' _________ 1. Both speakers regard pǔbiàn 'pervasive' and pǔpiàn 'pervasive' as "the same word". In Beijing Mandarin, it is pronounced as pǔbiàn; in Taiwan, pǔpiàn is more common.

The purpose of a gloss is to provide a maximally applicable meaning of a word or morpheme. In other words, glossing will be most effective if you select the same gloss in the target language (in this case, English) for any occurrence of this word or morpheme in the source language (Mandarin). This will allow a reader who does not know the source language to have a good impression of the elements making up a sentence.

To illustrate, examples (17) and (18) would violate this principle:

(17)    Zhōngguo yǒu shān.   (18)    Wó yǒu shū.   China be.there mountain   1SG have book   'There are mountains in China.'   'I have books.'

If yǒu is glossed as 'be.there' in example (17), glossing it as 'have' in (18) will fail to bring across that Mandarin uses the same verb in both cases. Consistent glossing will make this clear:

(19)    Zhōngguo yǒu shān.   (20)    Wó yǒu shū.   China be.there mountain   1SG be.there book   'There are mountains in China.'   'I have books.'

Examples (19) and (20) provide intrinsic information about Mandarin by using the same glosses for yǒu in the second line. In other words, the contrast between 'have' and 'be there' does not belong to the line of glosses, which should represent the source language. It is only relevant in the target language, English, which is why this difference is relegated to the third line.

In the case of grammatical function words, a gloss usually consists of an abbreviation, which you must list separately. In example (1), ma is glossed as "Q", as an abbreviation for question. A list of abbreviations would then specify, for instance, that ma is an interrogative particle which may follow a sentence and turn that sentence into a question. Note that as a gloss, 'Q' cannot be replaced by 'question', simply because ma does not mean 'question'; that gloss might be used for the Mandarin form wènti.

Examples (18) en (20) likewise illustrate the use of abbreviations in linguistic glosses": "1SG" designates the (1 = ) first person (SG = ) singular pronoun. Also note the use of a dot in "be.there', which indicates that the two English words together make up the gloss of Mandarin yǒu.

You are free to design such conventions to your taste. Always make sure, however, to communicate these conventions to your reader, and to apply them consistently. To get an idea of the possibilities, have a look at the Leipzig Glossing Rules issued by the Max Planck Institute for Evolutionary Anthropology.

6. Outro

The good contents of your work always prevail, but good presentation will make it more widely acceptable.

• Contents: Formulate a clear and succinct research question, treat its various aspects coherently, and work toward you conclusion step by step.
• Presentation: Make sure your page lay-out invites the reader: provide an easy read, not chaos.

Read your own text critically, always imagining how someone reading this might interpret your words. Do not create false hopes about your contents and do not embellish the facts. Document what you have done, and what this has shown us. And don't forget that unsolved problems, too, may contribute to science if you formulate them carefully and precisely.

Finally, leave your text alone for a while before finalizing your work. Allow a sneak preview to a colleague or friend, preferably an outsider who is an interested and critical reader.