Grammar writing for a grammar-reading

audience*

Michael Noonan
University of Wisconsin-Milwaukee

Thousands of languages are currently in danger of extinction without having

been adequately documented by linguists. This fact represents a tragedy for
communities in which endangered languages are spoken, for linguistics as a
discipline and for all of humanity. One major role of the field of linguistics is to
describe languages accurately and thoroughly for the benefit of all concerned.
This paper presents the results of an informal survey of major users of grammat-
ical descriptions and gives lists of dos and don’ts for those contemplating a de-
scriptive study of one of the many endangered languages of the world. Concrete
suggestions are provided that will help grammar writers produce user-friendly,
thorough and useful grammatical descriptions.

1. Introduction

I’ll begin this paper with a recitation of a set of familiar and depressing facts. Of
the approximately 7000 languages spoken today, the following can be said with
regard to demographics, all gleaned from the Ethnologue (Grimes 1996):
– Fully 50% of the human race has as their first language one of just 10 lan-
guages;
– 52% are spoken by fewer than 10,000 people;
– 28% are spoken by fewer than 1,000.
Long term estimates concerning language vitality vary considerably, the most pes-
simistic being that of Michael Krauss (1992) who predicts that by the end of the
present century up to 95% of the world’s languages will be moribund or extinct.
The linguists associated with the Volkswagen Foundation’s DOBES project opt
for the rather more optimistic prediction that 60–70% of the world’s languages
will have died out by that time. Most everyone would probably agree that by the
middle of this century, there will be less than 2000 languages spoken by integral
114 Michael Noonan

communities, that is by communities in which both children and adults speak

the language, and many of those languages will be changed in fundamental ways
through contact with languages of national and global importance.
All of this would be discouraging enough if the world’s languages were ad-
equately documented. But here too the situation appears grave. We’re confronted
with the following, all of which are my [probably overly optimistic] estimates:
– We have full-scale grammars and dictionaries and abundant textual material
for perhaps 500 languages;
– We have grammatical sketches or short grammars, and dictionaries for per-
haps another 2000 languages — many of these are of poor quality;
– We have only rudimentary documentation [word lists, a few sentences, per-
haps a paper on some aspect of the grammar] for another 2500 languages;
– We have little useful grammatical or lexical data for about 2000.
So, not only are most of the world’s languages inadequately described, but given
the rate at which languages are becoming moribund or extinct, we are engaged in
a race against time [and against war, poverty, and the good and bad effects of glo-
balization] to preserve as much of the world’s linguistic heritage as possible.
In running this race, we face, among others, the following three major ob-
stacles: the standard of grammar writing is not uniformly high, there is not a lot
of funding to support grammar writing, and the profession does not sufficiently
value or support the writing of grammars.
This paper will be concerned primarily with the first of these problems, namely
the establishment of higher standards for grammar writing. As regards the second
problem, funding, there have been many encouraging developments recently, and
I will have nothing further to say about this issue. About the third issue, the evalu-
ation of grammar writing within the profession and the professional support pro-
vided to grammar writers, I will make a few comments at the end of this paper.

2. Standards for grammar writing — full-scale grammars

When you teach the craft of writing to students, one of the things you try to impress
upon them is the importance of taking into account the audience to whom they
are addressing their writing. The success of a piece of writing is usually measured
by its reception. One has to know who will be reading the thing written and what
their needs and expectations are in order to assess whether a work is successful.
Reference grammars have two sorts of audiences. The first consists of those
who use a grammar to help them learn to speak or write the language described
 Grammar writing for a grammar-reading audience 115

therein, and the second consists of those who consult a grammar to obtain in-
formation for typological or theoretical studies, i.e. professional linguists. The
first sort of audience will most likely be learners of languages taught in formal
classroom settings, languages which are used by people with political or economic
power and which have significant literary traditions. Linguists, of course, may also
be interested in obtaining information about these languages, but linguists are also
likely to be interested in gathering information about languages which are not
taught in formal classroom settings, which have little or no tradition of literary
production, and which are not spoken by people with political or economic power
— that is the great majority of the world’s languages. In fact, the primary audience
for the grammars of such languages are linguists, and those writing grammars of
languages of this sort should write their grammars with the needs and expecta-
tions of professional linguists in mind.
One should also bear in mind that many of the grammars written in the next
few decades will someday be the only sources of information about the languages
they describe — not just for linguists, but also for the communities that currently
speak them. The transition from vital, to moribund, to extinct can happen sur-
prisingly quickly. For example, when I started working on the Nar-Phu language
of Nepal in 1996, the language seemed quite vital — protected, it seemed, by the
very hardships faced by the people who spoke it and the remoteness of their home
region. By 2001, the young people had switched almost completely to the national
language, Nepali, and when speaking Nar-Phu, many had difficulty recalling Nar-
Phu words, in particular the special and distinctive honorific vocabulary, and were
unable to produce the correct [i.e. historic] tones for many words they could recall.
So, in preserving languages, it is important to record them not just before they
die, but while they are still spoken by integral communities and while they still
preserve that which made them distinctive representations of the human mind
and spirit. The Nar-Phu that the younger generation will remember will not be the
same language as that of their forebears.
With the preceding as background, we should be aware that when we are writ-
ing grammars of those languages which will likely be moribund or extinct by the
end of the century — that is, the great majority of the world’s languages — that we
are writing for the ages. So, we must make sure that what we are doing reaches for
a very high standard.
What follows is a set of prescriptions for grammar writers intended to inform
them about matters of form and content which would help make their grammars
meet the needs and expectations of linguists and help give their grammars lasting
value. In putting this list together, I consulted with a number of linguists — major
users of grammars — who graciously took the time to contribute ideas on what
116 Michael Noonan

a good grammatical description should be or should contain. These linguists are,

in alphabetical order: Sasha Aikhenvald, Balthasar Bickel, Bernard Comrie, Bill
Croft, Bob Dixon, Matthew Dryer, Mike Hammond, Martin Haspelmath, Larry
Hyman, Ian Maddieson, Edith Moravcsik, and Randy LaPolla. Where one of the
these linguists has suggested to me one of the points below, s/he is identified.
The list of twenty-eight dos and don’ts is organized informally into three general
categories: "user friendliness", "descriptive adequacy", and "comprehensiveness".
User friendliness: make your grammar one that is easy to use and obtain infor-
mation from.
1. Where possible, avoid theory-specific terminology and use instead "basic linguis-
tic theory" as a source of terms: Experience has shown that grammatical theories
have a short shelf-life. Special terminology and notational conventions employed
by new theories are appropriate for journal articles but not for grammars, which
ought to have a shelf-life longer than any given grammatical theory. Libraries are
littered with virtually unusable grammars employing the very specialized conven-
tions of extinct theories. Grammar writers should try, therefore, to write gram-
mars for the broadest possible audience of linguists, employing only notational
conventions and terms that trained linguists of any theoretical persuasion could
reasonably be expected to understand, using terms drawn from what has recently
come to be called "basic linguistic theory". Where new or theory-specific terms
are introduced, they should be carefully defined. [Dixon, Hammond, Haspelmath,
Maddieson, Moravcsik]
2. Provide a detailed index and table of contents: The index and table of contents
of a grammar are important tools, especially useful when they are sufficiently de-
tailed and properly organized. Though both serve to help grammar readers find
information, they serve somewhat different functions. The index provides an al-
phabetical listing of topics, whereas the table of contents provides a schematic
overview of the topics covered. The latter can be very important where the terms
used by the grammar writer differ from those used by [or expected by] the gram-
mar reader: the grammar reader may well find the information s/he needs quickly
with a logically organized, detailed table of contents. In addition, in organizing an
index, some conventions employed by Martin Haspelmath in his Lezgian grammar
should be followed, namely noting in boldface type the main entry [or entries] for
a given feature, and noting the absence of a given feature directly in the index with
the dash. So, for example, one could indicate the absence of tone as follows: tone:
— . [Bickel, Comrie, Croft, Haspelmath]
 Grammar writing for a grammar-reading audience 117

3. The text should be divided into numbered and titled sections and subsections, and
there should be ample cross-referencing within the text: Within the text itself, the fol-
lowing conventions make the text easier to use and should be encouraged: the text
should be divided into numbered and titled sections and subsections, and, where
appropriate, there should be crossreferencing between parts of the text where re-
lated issues are discussed; important terms should be highlighted by boldface type
and/or by placing the terms in the margins; and finally, section numbers and/or
titles should be placed in the margins as headers. [Moravcsik]
4. Provide plenty of examples: Properly glossed and translated examples are a nec-
essary accompaniment to descriptive statements in all components of the gram-
mar. Care should also be taken in the choice of examples: made-up sentences are
appropriate for the presentation of information about basic grammatical struc-
tures, but examples should be drawn also from naturally occurring discourse.
Further, the grammar writer should avoid using examples which are drawn from
translations of foreign texts since such sentences may have been influenced by
structures in the source language. [Aikhenvald, Maddieson]
5. Provide interlinear morpheme translations [glosses], as well as translations of the
whole, for all examples: The necessity for this should be obvious, but too often this
courtesy to the grammar reader is ignored or applied inconsistently. There should
be a conveniently accessible list of all abbreviations and symbols used in mor-
pheme glosses and all other parts the grammar and dictionary.
6. A typological sketch, consisting of no more than three to five pages, should be
included at the beginning of the grammar: A short typological sketch is useful to
the grammar reader because it helps the reader put the more detailed informa-
tion contained in the grammar proper into appropriate context. The sketch should
outline prominent features of the phonetics and phonology; should provide in-
formation about basic word order patterns, the presence of basic nominal and
verbal inflectional categories, the presence of concord classes or classifiers, special
lexical features, and so on; and should briefly describe if the language is head- or
dependent-marking and if [and to what degree] the language exhibits properties
of ergativity, inverse marking, split intransitivity, and so on. [Aikhenvald]
7. The absence of a feature should be noted just as reliably as the presence of a feature:
Too few grammars explicitly note the absence of commonly encountered gram-
matical features in the languages they are describing. Grammar readers cannot
always be certain, therefore, whether the lack of discussion of a particular feature
results from its absence in the language or from the fact that the grammar writer
simply did not discuss the matter. Since the grammar writer cannot reasonably be
118 Michael Noonan

expected to note the absence of any of the full set of possible grammatical features,
s/he will have to choose carefully which features to note the absence of. The absence
of any feature which might be expected on areal, genetic, or typological grounds
should be noted [e.g. the absence of concord classes in a Bantu language or ejective
consonants in a language of the Pacific northwest], as should features which are
commonly encountered worldwide [e.g. tone]. [Aikhenvald, Bickel, Comrie]
Descriptive adequacy: Make your grammar one in which topics are described
in adequate detail, in a manner that reveals important generalizations, and with
adequate backing.
8. The grammar writer should prepare him/herself by obtaining a good background
in general typology and the linguistics of the relevant language family: A sound
background in general typology is imperative for grammar writers; it is also im-
portant for the grammar writer to acquire a sound knowledge of the linguistics of
the relevant language family and/or speech area and to consult works dealing the
typology of features prominent in the language under investigation [e.g. works on
ergativity for works on languages exhibiting ergative structures, works on classi-
fier systems for languages having classifier systems, etc.]. [Aikhenvald, LaPolla,
Maddieson]
9. In explaining phonetics, standard IPA characters should be used: Standard IPA
phonetic symbols, not "practical orthography" symbols [even where these are
sanctioned by tradition for a given speech area or language family], should be used
in presenting information about the phonetics of a language. Even when IPA sym-
bols are used, they should be accompanied by a statement describing their value
[which could be done by reference to a standard work such as the IPA Handbook].
When non-IPA characters are introduced for use in the practical orthography for
the language, their relation to the IPA characters used to describe the phonetics
should be made transparent. [Hammond, Maddieson]
10. To the degree possible, detailed instrumental documentation of the phonetics
should accompany descriptive statements: Descriptive statements of the phonetics
of a language should be accompanied by instrumental documentation of at least
the following features: acoustic characterization of vowels in formant space, mea-
surement of the acoustic durations of vowels and consonants in a controlled set of
data [so that some aspects of the duration patterns of the language can be quanti-
tatively described], and for tone or pitch accent systems the shapes of F0 patterns
associated with the contrasts. Any typologically unusual phonetic features should
be described together with appropriate instrumental documentation. [Hammond,
Maddieson]
 Grammar writing for a grammar-reading audience 119

11. Provide a full description of the segmental and suprasegmental contrasts and
an explanation of the basis for arriving at them: A grammar should contain a full
description of the segmental contrasts and an explanation of the basis for arriv-
ing at the analysis. In particular, the analysis must demonstrate contrast between
segments [in the form of minimal pairs] and present arguments in favor of [or
against] the unitary vs sequential analysis of complex elements. In addition, the
analysis must include careful descriptions of suprasegmental contrasts: tone [if
phonemic], stress, and intonation. Intonation and stress may have implications for
morphology and syntactic structure. [Maddieson]
12. Provide a description of distributional patterns of the elements of the phonology:
A grammar should contain a discussion of the basic distributional patterns of the
elements of the phonology in terms of syllables, words, and any other units which
seem to control, or be marked by, different distributional possibilities. Minimally,
canonical syllable structure, possible onsets and codas, and nucleus-margin re-
strictions should be described. [Maddieson]
13. Provide full paradigms showing combinations of all relevant morphemes, not just
lists of affixes: In morphologically complex languages, full paradigms, arranged in
some accessible form [e.g. as tables] should be provided and not just a list of the
affixes. When the language has complex morphophonemic processes, there should
be a sufficient number of paradigms provided to illustrate the various combinato-
rial possibilities. [Hyman]
14. Define grammatical categories used in the grammar: When using grammatical
category labels, the grammar writer should make clear what grammatical tests are
being used to define those categories. This applies especially to labels for gram-
matical roles [e.g. subject, direct object, etc.] and word classes [e.g. adverb, adjec-
tive, etc.]. [Aikhenvald, Croft]
15. The choice of labels for grammatical features is not as important as a thorough
presentation of the facts: Grammar writers should not worry too much about
whether a label for a given feature is the most apt. More important is a thorough
description of the phenomenon. [Croft, Haspelmath, LaPolla]
16. It’s better to admit ignorance about a grammatical feature than to say nothing:
Saying nothing about a particular grammatical feature leaves the matter open to
incorrect interpretations, for example that the feature does not exist in the lan-
guage. When the grammar writer lacks sufficient information about an issue, s/he
should not be afraid to admit this, using perhaps a statement like the following:
"There is insufficient data at this time to determine conditions under which this
form is used." Further, it is important to emphasize that grammar writers should
120 Michael Noonan

present data about grammatical phenomena that their theories either cannot ac-
count for or are not concerned with: technical papers in journals can be narrow in
their scope, but reference grammars must endeavor to describe the entire language
system. [Comrie, Hammond, Haspelmath]
17. Existential statements are not sufficient: In writing grammars, it is not sufficient
simply to note that the language has some feature X; one must also describe under
what conditions that feature is used. For example, noting that Chantyal has a plu-
ral suffix -ma and then limiting the discussion to a few examples is not enough: an
adequate description will also say under what conditions the suffix must be used
and under what conditions it cannot be used. [Haspelmath]
18. Indications of frequency should be provided where appropriate: Indications of
the frequency of grammatical constructs can provide useful information. These
include segmental frequency counts [based on texts or the lexicon] and indica-
tions of whether a given morphological or syntactic feature is rare or marginal [e.g.
lexically limited]. [Bickel, Maddieson]
19. In general, it’s best to describe morphology with a form-to-function orienta-
tion and syntax with a function-to-form orientation: It’s probably impossible [or
at least impractical] to write a grammar which is exclusively form-to-function [as
the Bloomfieldians attempted in principle] or function-to-form [as Comrie-Smith
grammars attempted in principle]. The best compromise seems to be to write the
morphology primarily form-to-function and the syntax primarily function-to-
form. The reason for this arrangement is that form is very easy to recognize in
morphology, but is rather less so in syntax. This does not mean that function-to-
form should be entirely absent from the morphology, nor form-to-function from
syntax. For example, alongside a form-to-function discussion of the meanings of
verbal affixes, it would also be useful to have a function-to-form discussion of
TAM categories, describing how they are expressed in the language. And in syntax,
in addition to function-to-form chapters on nominal modification, interrogation,
expression of negative senses, etc, it would also be very useful to have a form-to-
function chapter on the uses of various word order possibilities. [Haspelmath]
20. A vocabulary consisting of all the lexemes which occur in the grammar is a neces-
sary component of a good grammar: Grammatical descriptions longer than a sketch
should contain a vocabulary consisting of all the lexemes used in the grammar and
accompanying texts. These lexemes should be properly labeled for word class or
concord class [if appropriate] and, for grammatical morphemes, there should be
an indication of the section in the text where the item is discussed. Where variant
 Grammar writing for a grammar-reading audience 121

forms of a lexical item are found in the grammar and texts, the variants should be
noted in the vocabulary. [Dixon, Aikhenvald]
21. A good collection of texts should be included with the grammar: A good collec-
tion of texts with morpheme glosses and translations is of great importance for a
successful grammar. The texts provide illustrations of the constructions described
in the grammar, but unlike the examples in the grammar itself, textual examples
show the constructions in context. Further, grammar writers cannot be expect-
ed to write about, or even know about, all possible grammatical constructions.
A good collection of texts can permit later analysts to gather information about
constructions that the grammar writer, for whatever reasons, did not discuss or
did not discuss in much detail. There should be at least twenty to thirty pages of
texts. [Dixon]
Comprehensiveness: make your grammar one which is comprehensive in scope.
22. The grammar writer should consult survey questionnaires and wellregarded
grammars to make sure that important topics are not missed: The most detailed
survey questionnaire that is generally available is the Comrie-Smith [Lingua De-
scriptive Series] questionnaire, which has formed the basis for a number of gram-
mars. This function-to-form questionnaire [see above] can be profitably consulted
by the grammar writer, as can well-regarded grammars, of which several lists are
available. [Comrie, Dryer, LaPolla]
23. The grammar should contain information about genetic and areal affiliations
of the language: A description of the genetic affiliation of the language should be
provided, along with information about the linguistic area and possible areal influ-
ences, where such information is known. This can be fairly brief unless there are
special issues to be discussed [e.g. controversies concerning genetic affiliation].
[Aikhenvald]
24. The grammar should contain information about the sociolinguistic context:
There should be a short description of demographic and socio-cultural facts relat-
ing to the language being described, including the number and geographical dis-
tribution of speakers, the demographics of language use [e.g. whether the language
is spoken only by adults and not children], the degree and nature of multilingual-
ism in the speech community, and the degree of literacy and access to education.
In addition, the author should reveal how data about the language was obtained
and should provide information about the native speakers who served as language
consultants.
122 Michael Noonan

25. Grammars should contain ample references to previous scholarship on the lan-
guage and its culture: The bibliography should contain ample references to previ-
ous scholarship on the language; if the language has not been the subject of much
previous scholarship, the list of references should be comprehensive. Previous
scholarship on the culture of the speakers of the language should also be noted.
26. Grammars should contain good descriptions of the phonetics and the phonol-
ogy, as well as of the morphology and the syntax: Some grammars include little
more than a list of phonemic or "practical orthography" symbols and only a brief
description of the phonology with just a few examples. Proper descriptions of the
phonetics and the phonology of a language are just as important to the success of
a grammar as proper descriptions of the morphology and syntax. And here too
there should be a plenitude of examples, narrowly transcribed. [Hammond, Hy-
man, Maddieson]
27. A good dictionary is an powerful adjunct to a good grammar: As noted,
vocabularies, consisting minimally of the lexical items employed in the grammar,
should be included with any grammatical description longer than a sketch. The
existence of such a vocabulary, however, does not diminish the value of a good, full-
scale dictionary, which, apart from any other uses it may have, is also a powerful
adjunct to a grammar. A good dictionary comes with lots of example sentences
from which a great deal of useful grammatical information can be gleaned even
about topics that the grammar writer/dictionary creator has no knowledge of or
interest in. For example, very few grammars, even good grammars, contain much
information about clausal complementation. Typically there will be discussions
of complement-types with some example sentences that hint at the distribution
of these forms, but very rarely will the grammar contain much explicit discussion
on the distribution of complement-types vis-à-vis complement-taking predicates,
or contain a sufficiently large number of examples to allow the analyst to infer
the distribution. A good dictionary, on the other hand, can be a goldmine of
information about such topics.
28. Where practical, audio and video recordings should be made of various lan-
guage genres: Where practical, audio and video recordings of a variety of language
genres and cultural activities should be made. A subset of these should be anno-
tated, provided with appropriate commentary, and deposited in an archive where
they may be studied by scholars. [A number of such archives are now in existence.]
Further, photographs and drawings can be very useful in documenting informa-
tion about cultural artifacts, the natural environment, and botanical and zoologi-
cal vocabulary.
 Grammar writing for a grammar-reading audience 123

Points 27 and 28 are desiderata rather than requirements for successful grammars.
However, with regard to point 28, it’s well to keep in mind that over the years the
materials to be included in “good” grammars, i.e. ones that would fully meet the
highest expectations of their day, have progressively expanded with advances in
the field and in technology. At this time, we can’t anticipate the degree to which,
say, intonational information, ordinarily not transcribed in published discourses,
will be considered crucial to syntactic analysis. It’s best to archive audio and visual
records of speakers, if at all feasible.
Needless to say, not everything that a grammar should contain is noted in the
points above since the list focuses mostly on those things that are left out or not
done with some regularity. This accounts for the fact that the prescriptions relating
to phonetics and phonology are much more specific than those for morphology
and syntax, reflecting the generally lower standards of training of field linguists in
phonology and, in particular, phonetics.
The standard of grammar writing detailed above raises a pretty high bar, and
I’m the first to admit that a grammar can be very useful and still not conform to all
of the prescriptions listed. However, the more that are observed, the more useful
the grammar will be.

3. Standards for grammar writing — sketches and short grammars

The list of dos and don’ts already given applies, of course, only to full-scale refer-
ence grammars and not to grammar sketches, which necessarily will fail to con-
form to many of the prescriptions on the list, in particular those placed under
"descriptive adequacy" and "comprehensiveness." Grammar sketches and short
grammars have, of course, a legitimate function. In a perfect world, we would
have full-scale grammars for every language, and grammar sketches would serve
as convenient summaries of the facts. In our very imperfect world, often all we
have for many languages — and maybe all we will ever have — is a sketch or short
grammar. These shorter works, then, should be prepared with the same care as
full-scale grammars. Of the list of dos and don’ts, the following are applicable even
to sketches and short grammars. The numbering of the prescriptions is preserved
from the original list:
User friendliness
1. Where possible, avoid theory-specific terminology and use instead "basic linguis-
tic theory" as a source of terms.
2. Provide a detailed index and table of contents.
124 Michael Noonan

3. The text should be divided into numbered and titled sections and subsections,
and there should be ample cross-referencing within the text.
4. Provide plenty of examples.
5. Provide interlinear morpheme translations [glosses], as well as translations of
the whole, for all examples.
7. The absence of a feature should be noted just as reliably as the presence of a fea-
ture.
Descriptive adequacy
8. The grammar writer should prepare him/herself by obtaining a good background
in general typology and the linguistics of the relevant language family.
9. In explaining phonetics, standard IPA characters should be used.
11. Provide a full description of the segmental and suprasegmental contrasts and an
explanation of the basis for arriving at them.
12. Provide a description of distributional patterns of the elements of the phonol-
ogy.
13. Provide full paradigms showing combinations of all relevant morphemes, not
just lists of affixes. [applies to short grammars, not sketches]
14. Define grammatical categories used in the grammar.
15. The choice of labels for grammatical features is not as important as a thorough
presentation of the facts.
16. It’s better to admit ignorance about a grammatical feature than to say nothing.
18. Indications of frequency should be provided where appropriate.
19. In general, it’s best to describe morphology with a form-to-function orientation
and syntax with a function-to-form orientation.
Comprehensiveness
22. The grammar writer should consult survey questionnaires and well-regarded
grammars to make sure that important topics are not missed.
23. The grammar should contain information about genetic and areal affiliations of
the language.
24. The grammar should contain information about the sociolinguistic context.
25. Grammars should contain ample references to previous scholarship on the lan-
guage and its culture.
Of the twenty-eight prescriptions on the list, all but 6, 10, 17, 20, 21, 26, 27, and 28
apply to sketches and short grammars as well; 13 applies to short grammars, but
not necessarily sketches. Needless to say, the more of these additional features are
incorporated into a grammar or even a sketch, the more useful the it will be.
 Grammar writing for a grammar-reading audience 125

4. Community support

I’ll close with a few remarks on how the community of linguists can support gram-
mar writing.
Among professional linguists, grammar writing is subtly discouraged by the
way in which grammars are evaluated for purposes of hiring, tenure, and promo-
tion. At research universities in particular, published articles on theoretical mat-
ters are valued more highly than published grammars or grammar sketches. Pro-
fessional organizations could help change this situation by, for example, raising
awareness among linguists concerning the urgent need for grammar writing at
this time and by encouraging grammars as topics for dissertations. The latter could
be accomplished by the creation of special dissertation fellowships for grammar
writing and by skewing the selection criteria for existing dissertation fellowships
in favor of grammar writing. Professional organizations such as the LSA or ALT
could also establish awards for the best published grammar of the year, thereby
raising the profile of grammar writers.
Further, the community of linguists could provide support for grammar writ-
ing through the establishment of consultancy services for qualified grammar writ-
ers. This service is of greatest importance in the area of phonetics, where the stan-
dard of training throughout the profession is probably the lowest among those
skills that are required for writing grammars.
Finally, online publication of grammars and dictionaries has a number of ad-
vantages over paper publication: online grammars and dictionaries can easily be
updated and revised, thereby increasing the likelihood that they will conform to
the prescriptions discussed earlier. They can also be made available to a wider
audience [especially if access is free] than is possible with paper publication. And
lastly, online, or at least electronic, publication can facilitate the addition of audio
and visual materials to the written text of the grammar.
There are two problems with online publication. The first is that, in many cas-
es, it is not evaluated as highly as paper publication for purposes of hiring, tenure,
and promotion. If the venues for online publication meet the same standards as
high quality paper publication — that is, meet high standards of editorial review
— then they should be evaluated equally. Again, professional organizations have
a role to play here in encouraging academic departments to accept high quality
online publication.1 The second problem relates to the relative impermanence of
electronic and online publication media. Here too, professional organizations can
play a role in arranging for the archiving of grammars published electronically in
longer-lived formats.
126 Michael Noonan

Notes

* I’d like to thank Edith Moravcsik and an anonymous reviewer for helpful comments. In addi-
tion, I’d like to thank the linguists, named below, who provided me with their insights on gram-
mar writing. Needless to say, all the shortcomings of this paper are mine alone.

1. Online publication has a number of virtues, but it also poses a number of challenges both for
academic departments evaluating candidates for hiring, promotion, performance raises, and so
on and for editorial boards. For example, one of the virtues of online publication is that it allows
for easy revision of texts. For editorial boards, a question arises as to how much revision should
be allowed, how it should be monitored, and how much revision is allowable before the work
must be reviewed again. Different standards might well be applied to journal articles and refer-
ence works such as grammars. In the case of grammars, I would suggest a fairly liberal standard
for revision, drawing the line at changes that affect the framework or mode of analysis. In such
cases, the revised grammar would have to undergo another review process.

References

Grimes, Barbara F. et al. 1996. Ethnologue. Dallas: SIL International.

Krauss, Michael. 1992. The world’s languages in crisis. Language 68(1):4–10.