The Format of Teaching Grammars
Ashok R . Kelkar





Grammar-1 as the object of study is to be distinguished from grammar-2 as the outcome of the study of grammar-1.  Grammar-2, in turn, may be a theoretical grammar or a teaching grammar—the latter being intended for the use of a teacher and/or a student of that language.  Teaching grammars may be classified in various ways—in particular, they may be arranged in order of increasing pedagogic deviation from the underlying theoretical grammar, whatever it may be, in certain directions.  The best teaching grammar for a given purpose achieves an optimal deviation.


The four directions are : (1) grading and grouping of the points—at the gross and fine levels; (2) bias for usage rather than system and for transduction rules rather than specification rules; (3) bias for itemization rather than stating constraints; (4) bias for specification by form rather than by meaning.


In so characterizing a teaching grammar, one is taking it to be an information retrieval system that has achieved a pedagogically optimal linearization (principle 1) and a pedagogically optimal balance between readystored information and computable information (principles 2-4).


I think that a potential source of confusion for the present discussion may be forestalled with the help of an ancient Indian conceptual tool.  Words like grammar or history suffer or benefit (as the case may be) from a polysemy—they stand both for something worth attending to (laksya in Sanskrit) and, metonymically, the account one can formulate after so attending to it (laksana).  (See Katyayana’s scholium 14 to Panini and Patanjali’s comment thereon in his Vyakarana-mahabhasya 1:1:1.)  It is immaterial whether a given word like grammar, history, or vyakarana originally referred to the object or to the account.  But it is material whether one means grammar-1, the object of a grammarian’s attention, or grammar-2, the end-product of the grammarian’s labour when one speaks, say, of the influence of English grammar on Marathi grammar or pleads for “grammars for the consumer.”  (I have feeling that Halliday (1964) meant grammar-2 while his transformational-generativist critics meant grammar-1 as a discoverable object.)  It is one thing to say that grammar-2 need not be emphasized while one is teaching the language and quite another to conclude therefrom (without justification) that one can learn a language without mastering its grammar-1.  Obviously a good proportion of language learning (and therefore language teaching) has to do with grammar-1.  granting that grammar-1 and grammar-2 are both human artifacts, one must point out that grammar-2 is the more “artificial” of the two products.


Now what sort of grammar-2 should be in the hands of the teacher and, possibly, of the learner?  Clearly they cannot be quite the same for both.  Nor can either be the same as a grammar-2 drafted by a grammarian for fellow grammarians.  One has to recognize two kinds (at least) of grammars-2—teaching grammars and theoretical grammars.  A theoretical grammar-2 is presumably one that reveals the more transparently to us what the grammar-1 of a language is like.  A teaching grammar-2, on the other hand, may be intended for the learner, for the learner, for the learner who is his own teacher, for the teacher proper, or even for the designer of text material, curricula, tests, and the like.  Whatever the specific use or combination of uses intended are, a teaching grammar is liable to be more or less different from a theoretical grammar, the extent of the difference depending on what piece of information is being presented.  (For some pieces or sequences of pieces the difference may be null.)  I shall assume that the format of a teaching grammar can be conveniently thought of as a point-by-point modification of a theoretical grammar-2 in a definite direction.  The thing to know is, first, what this direction is and, secondly, where along the line one should stop deviating, given the specific purpose of a teaching grammar.  The present study is addressed to the first of these two questions.


It will be noticed that I have left wide open the possibility that there may be different kinds of teaching grammars.  Some may, for example, be more pedagogically deviant than others—that is, some may deviate to a further degree or in further ways from theoretical grammars than others.  A teaching grammar has to select the optimal pedagogic deviation from the underlying theoretical grammar.  Thus a grammar for the student is liable to be more pedagogic and deviate more from a theoretical grammar than one for the teacher himself.  Again, a grammar that is geared to the beginning student’s contact texts (say, stories or dialogues) or to exercises and tests has to be more pedagogically deviant than a grammar used by an advanced student for occasional reference in solving a problem of comprehension (What can the author mean by this?  Let me get the hang of it.)  or expression (How does one manage to say such and such things?  How do I construct a sentence with a contrary-to-fact condition?).  So, different degrees of pedagogic deviation may be desirable for different purposes.


There are of course other ways than the degree of pedagogic deviation in which teaching grammars may differ from each other.  They may be shorter or longer—the more exclusively a grammar is confined to essentials, the shorter it is.  This suggests that grammar-1 itself is susceptible to a distinction between its essential elements and peripheral elements.  If such is the case (and I think such is the case) a theoretical grammar-2 should correspondingly indicate such a grading for its formulations.  The complex nature of this distinction between essential and peripheral features of grammar-1 is not yet exactly understood.  (For a recent discussion see Vachek, ed., 1966.)  One can do no more here than suggest to the reader that the first member in each of the following pairs is more essential than the second, other things being equal :


fully coded / partially coded

discrete distinctions / distinctions along a continuum

unmarked / marked

regular / irregular

nuclear / adjunct

general / specific

less delicate / move delicate (in Halliday’s terminology)

amenable to intuitive assimilation / amenable rather to problem-solving

basic rules of the language game / ground rules for adjustment

to a contingency (an extreme example being those governing the abbreviated syntax of  telegraphic messages)

rules of the game / set strategies for successful play (like “third hand high” in bridge)

grammar / idiom and usage (the traditional distinction)

rules presupposed by many other rules / rules presupposing many other rules


One can go on to other classifications.  There may be general-purpose grammars and grammars with a specialized bias, such as bias in favour of oral-aural skills or of manual-visual skills; or in favour of receptive skills or of receptive-reproductive skills or of receptive-productive skills; or in favour of this or that diatype—colloquial, literary, technical, and so forth.


The bilingual state that the teaching programme aims at producing in the learner may be one of parallel bilingualism or of compound bilingualism.  Naturally, teaching grammars will consequently differ in the extent to which they take note of the resemblances and differences between the target language and the learner’s language.


Last but not least, teaching grammars may differ in the degree to which they present things directly.  In the extreme case all of grammar-1 will be presented indirectly and there will be no grammar-2 at the classroom level, but only a grammar-2 “underlying” the classroom material for the benefit of the teacher and the designer of that material.


The present study in confined only to one aspect of the format of teaching grammars—namely, the features that render a teaching grammar more and more pedagogically deviant.  I suggest that there are four such features or types of deviation from theoretical grammars.  The present study thus does not take up the question to what extent it is desirable, for a given purpose, to deviate pedagogically in a given feature or type of deviation.




The first and most obvious feature of pedagogic deviation is grading.  The arrangement of the formulations in a theoretical grammar will be governed entirely by considerations of the theory that the grammar-2 subscribes to—the theory about the nature of grammars-1.  Now let us suppose that the theory requires an arrangement, say, of the following sort :


            (1)        S; N; V


where S, N, V respectively stand for Sentence Formation, Nominal System, and Verbal System.  We have already suggested that a theoretical grammar will also need to take note of the distinction between essential and peripheral elements.  For simplicity’s sake let us postulate three degrees of descending essentiality :


(2)               1; 2; 3


Combining (1) and (2), we get (3) as an arrangement proper to a theoretical grammar :


(3)               S1 N1 V1

S2 N2 V2

S3 N3 V3


Whether the grammar is written column-wise (S1 S2……) or row-wise (S1 N1…….) is only a matter of convenience and not germane to its theoretical status.  A matrix like (3) is adequate as a description its format.


The contents of a teaching grammar on the other hand cannot avoid being linearized.  This linearization involves two processes of adaptation: the teaching points must be grouped together and the groups must be arranged in a sequence.  These twin processes of arrangement—namely, grouping and grading—have to be considered at two levels: at the gross level, say, arrangement into phases 1, 2, 3 and also at the fine level, say, the division of each phase into units 1, 2, 3.


The gross linearization in a full-scale teaching grammar that is the least pedagogically deviant will take the following shape:


(4)               Phase 1 : S1 N1 V1

Phase 2 : S1 S2 N1 N2 V1 V2

Phase 3 : S1 S2 S3 N1 N2 N3 V1 V2 V3


This phasing follows the so-called concentric method, each phase being relatively self-contained and successively richer in detail.  If one is looking for a shorter version of the grammars as opposed to the full-scale grammar, one need pick up just any one of the phases.


The gross grading in a full-scale teaching grammar that is the most pedagogically deviant will take the following shape :


(5)               Phase 1 : S1 N1 V1

Phase 2 : S2 N2 V2

Phase 3 : S3 N3 V3


The arrangement her is sequential and not concentric.  Shorter versions will simply delete the last phase, the last two phases, and so on.  The principle of linealization is carried to its utmost, while in the concentric method it is mitigated by recapitulation.


What are the principles of gross linearization in respect of grouping and grading?  Gross grouping is governed by the criterion of viability: at the end of each phase the learner must be in control of a slice of the whole language—a subset of the teaching points that is viable, i.e. sufficient to give him a working base.  Naturally this means that the subset must include a little of everything, and thus must be spread over S, N, and V; it must spread across the levels of phonology, grammar-lexicon, and semantics; and it must spread across the ranks from the maximal to the minimal size: sentence, phrase, word, formative.


Gross grading is governed, as can readily be seen from the displays (4) and (5), by the criterion of essentiality.  The relatively more essential precedes the relatively more peripheral.  The traditional reliance on frequency counts of vocabulary justifies itself only as a first approximation—a rough estimate that needs to be corrected in various ways.  In respect of grammar, frequency counts of the use of grammatical markers (e.g., of, ’s, the, -ing, -ly in English) are available for a few languages have had frequency counts of the attestation of syntactic rules.  Grammatical frequency counts are certainly an urgent pedagogic need.


I am aware of three such counts in existence—Fries 1940 for American English undertaken from the point of view of teaching English as Own Language; Keniston 1937 for Spanish; and Clark and Poston 1943 for French (closely modelled on the foregoing)—both undertaken from the point of view of Other Language teaching.  Keniston (1937: p. 12) shows for his count that a vocabulary of 1300-1400 words corresponds to syntax points with a range and frequency of 20-33 and above.  Clark and Poston (1943) based their count on 10,000 running words each from 60 prose texts published from 1881 to 1938.  Some sample entries are condensed below—


2.152  Construction f the type prevenir quelqu’un que… (V + NP + queS) (range 19, frequency 26)

2. 645  Substantive as indirect object of separation (54-168) after verbs like demander (34-61), échapper (18-20), etc.

7.46  Correspondingly for pronouns (66-569) after verbs like demander (54-249), falloir (42-92); manquer (24-36), cacher (15-23), etc.


As one can expect, for 85 grammatical points of obviously high frequency (and essentiality?)  no figures are given (such as substantive as subject of a finite verb or as direct object).  The count reveals that many rules traditionally deemed essential have low frequency, while many rules traditionally deemed peripheral are attested in the corpus with high frequency.  Somehow grammatical frequency-counting has never quite attracted the attention it deserves.


The fine arrangement within a phase will be based on the criteria of relatability in grouping and learnability in grading.  Teaching points that will be brought together within the same unit should preferably be relatable in some way—association by contiguity, resemblance, or contrast, for example.  Thus, tense and aspect contrasts and adverbials of time tend to cooccur and thus be contiguous, synonymous, or near-synonymous construction resemble, polarities like animate/inanimate or few/little contrast.  Such are the cases that lend themselves to grouping.  Sequence within a unit or distribution over earlier and later units will depend on relative learnability and, consequently, relative teachability.  One can assess this by the joint application of the following three sub-criteria:


(6)               Of two learning points within the same phase or subphase of grammar, that one will take precedence which is

(a)  the more interesting of the two,

(b)  the more accessible of the two,

(c)    the more presupposed of the two for purposes of exposition.


The least pedagogically deviant grammar will arrange the three criteria, in case of conflict, in the following order (a), (b), (c).  (Theoretical grammars will of course not be concerned with interest or accessibility at all; only (c) out of the three will have relevance.)  Pedagogically deviant grammars, thus, will sometimes face a problem: how to handle a point put first because of its greater interest or accessibility even when a point presupposed by it has to be postponed.  (Theoretical grammars or teaching grammars that never violate criterion (6,c) will of course not face this problem.)  How does a pedagogic grammar solve a problem of this sort?


Let us take a couple of concrete cases from French and English respectively.


(7)               In French teaching point (7a) precedes teaching point (7b) psychologically (i.e. in accordance with 6a, 6b) but point (7b) precedes point (7a) logically (i.e. in accordance with 6c):

(a)  il y a + NP (+ AdvP of  Place) (cf. English there is / are….)

(b)  conjugation of y avoir (yielding the specific form il y a).

(8)               Similarly, in English:

(a)  John came here (the use of John without the)

(b)  the John you saw yesterday; you mean the John?  Where (8a) is more

       interesting and accessible than (8b) but where (8a) presupposes (8b)        

       in which the is retained rather than deleted as in (8a).


The solution basically takes this form: point (7a) or (8a) is presented tour court as a fait accompli and not allowed to be “computed” from point (7b) or (8b) respectively, as one would do in the theoretically appropriate format.  Thus in (8a) the pedagogic grammar will present the non-use of the rather than its deletion.  The teacher will simply have to learn to be more tolerant or the learners’ errors arising out of their failure to anticipate points not yet covered!









Apparatus for Computing Information










                            Computed Information

            Figure 1. Information Retrieval System



Indeed the basic point about the pedagogic transformation of a theoretical grammar into a teaching grammar is, apart from the linearization of information-storage “addresses” implied in grading, that there are shifts in the relative balance between information stored in readiness at these addresses and information left to be computed by the user.  In making this statement I am of course considering grammar-2 here as an information-retrieval system with two main functions: yielding pieces of ready information and providing an apparatus for computing information not directly stored.  (See Figure 1.)  (The multiplication table, for example, gives ready-made the product of 9 and 7 and the multiplication rules tell us how to use this piece of information in getting the product of, say, 19 and 17.  A theoretical grammar strikes a balance between ready storage (say, the verb may have or lack past, perfect, and progressive in English) and computability (say, leaving the reader to compute the 8 combinations from this formula).  This balance is optimal from the point of view of making the grammar-2 as transparently revelatory of grammar-1 as possible.  A teaching grammar has to find its own optimal balance for its own pedagogical purpose (say, choosing to spell out the 8 combinations in ready storage.)  One information-retrieval system may be different from another yielding basically the same overall body of information in either of two ways.  One may offer some piece of information ready-made while the other leaves it to be computed.  (Thus we can construct a multiplication table in which “2 times 11,” “2 times 12,” “3 times 11,” etc. may be left out along with “2 times 13.” etc. to be computed.)  Secondly, while both may offer some piece of information ready-made, one will store it at one location or address in the storage, the other will store it not at the corresponding address but at some other address.  (Thus we can construct a multiplication table in which “2 times 11,” etc. are deleted but in which two new columns beginning with “11times….” and “12 times…” are added.)  The remaining three principles relate to just such transformations, which either shift the theoretically optimal balance between ready storage and computation or shift information from one storage location (address) to another.




It is no secret that there is not general agreement among linguistic scientists as to the form of grammar-1 and therefore as to the precise format for a theoretical grammar-2.  it is idle to pretend that this does not affect teaching-grammars-2, since a teaching grammar is an adaptation of some theoretical grammar or other which may be said to underlie it.  In the present study we have, however, chosen to focus on the mode of adaptation rather than on the choice of the underlying theoretical grammar, the latter not being a pedagogic question in any case.  The justification for such a procedure is twofold.  First, the departures from the theoretical grammar are thus saved from being ad hoc and shown to be amenable to theorizing.  Indeed the theory of adaptation presented here may in part (as in the case of grading) turn out to be more or less a codification of the existing practice of successful teachers.  The second justification stems from the fact that the choice of the mode of adaptation seems relatively independent of the choice of underlying theoretical grammar.  I say “relatively independent”; for obviously at least a broad agreement as to what capability a theoretical grammar must have in order to be considered respectable will be necessary if we are to be able to formulate the transformations at all.  I submit that, whatever the brand, a theoretical grammar has to make the following distinctions (see Figure 2):


(9)               (i)  between the system and the use of the system—what is usually called        

“competence” is the psychological middle term that makes the use of a system on the part of the individual listener or speaker possible and is not to be confused with the abstract object as such that is to be called “system”;

(ii)  between meaning and form—the linguistic sign and its use have an   

       aspect of form that makes signaling possible, and an aspect of            

       meaning that renders the signaling worthwhile by making signifying   


(iii) between constraints and items—if grammar-2 is a storage of   

       constraints affecting the construction of usable texts out of certain     

       elementary items, the dictionary is a storage of the elementary items    

       to which the constraints apply; the two together (the grammar and the 

       dictionary) describe the language system, of which grammar-1 is a 




Text Box: fication Text Box: les Text Box:         Speci Text Box:              ru

Specific types


















as seen in tokens in context


(a)                                                                                                                                        (b)


Figure 2.  Elementary facts about language


A given grammatical theory proposes its own principles for the optimal balance between system and use, between form and meaning, and between item and constraint in the placement of facts.  The actually written grammars-2, however, are apt to fall short of the ideal balance—invariably so in the initial versions—in that the analyst has to work his way from use to system, from form to meaning, and from itemizations to formulations of constraints.


Indeed, since linguistic analysis is essentially the analysis of the system, what is to be looked for and looked upon as a fact open to linguistic analysis may itself be at stake in disputes under (9) (i)—placement under system and under use.  So far as language pedagogy is concerned, the choice between system and use means a decision as to what is to be taught explicitly and systematically and what is to be left to the learners’  own devices, to ad hoc footnotes, to exposure to and imitation of examples.


Suppose, for example, that one were to decide that the distinction between John is easy to please and John is eager to please is not a part of the English grammatical system but rather a matter of usage.  In that case one would reject (10) (i) and opt for (10) (ii) (b), or in the extreme case (10) (ii) (c).


(10)           (i)  To postulate different structures for the two sentences.

(ii)  To postulate the same structure for the two sentences.

(a)  but to formulate interpretation rules that differ for adjectives of

                                           the easy type and ones of the eager type.

(b) but to relegate the matter to the glossing provided in the   

     dictionary entries for the two items easy and eager.

(c) and to neglect the distinction as just one of those curiosities   

     (one can’t possibly teach everything!).


Now this particular decision, however, is less likely to originate in the theoretical grammar than in its adaptation to a teaching grammar.  For example, one may decide that in a given phase this distinction needs to be taken note of but cannot be accounted for systematically for reasons of grading.  Conceivably one may decide that the distinction has no place at all in the system underlying a pedagogic grammar.  Both such decisions exemplify the second principle of making a grammar more pedagogically deviant.


The greater the tendency to decide in favour of usage rather than system, the more pedagogically deviant the teaching grammar is—more and more pieces of information will be stored in it at the usage addresses rather than be left to be computed from information stored at the system addresses.  Greater reliance will be placed on transduction rules that tell us which forms are to be interpreted in what manner, or which meanings are to be expressed in what manner, than on specification rules that tell us which forms or meanings are available in the first place for interpretation or expression as the case may be.  (See Figure 2 once again.)


It is certainly not as if a teaching grammar were jettisoning a piece of information incorporated in a theoretical grammar.  Rather it is repackaging it, changing its storage location.  What was packaged as specification of two distinct structures is now being offered as two possible transductions (in this case from forms to their interpretations).




The third principle has already been hinted at in (10) (ii) (b)—namely that the greater the tendency to decide in favour of item rather than constraint, the more pedagogically deviant the teaching grammar will be.


Thus, whatever the theoretically optimal decisions may be, it is more pedagogic to present him and them as ready-made forms than to extract –m separately and propose constraints on its use.  The same ground is being covered, but what might be offered in a theoretical grammar (for example, Trager & Smith 1951) as constrained use of –m and hi- and the- is being offered as two ready-made items in a teaching grammar.  (Of course even a theoretical grammar may find justifiable grounds for yielding on this point; not all generalizations are significant!) An early formulation of this insight to be found in Bloomfield (1945:p.8) is worth quoting:


Traditional grammars…often resort to repetition instead of assembling forms which present a common feature.  This habit is pedagogically useful.  When one does not carry it to the length of causing difficulty by sheet bulk….Systematic description, on the other hand, tries to assemble all forms that have any common feature and to unite them under a single statement….The partial resemblances between forms which we describe in morphology are often so whimsically irregular that a less rigorous statement has practical advantages.  It may take more discourse to describe a few eccentric forms than it would merely to cite them.  The author of a rigorous description finds difficulty in making it watertight; even Panini left some holes.  The reader finds difficulty in interpreting, applying, and combining the separate statements; this, too, would be true of Panini’s grammar….A less rigorous statement may be useful even  for scientific purposes.


It  should be recognized that to make a grammar more and more pedagogic in this or in any other respect may not always serve the teacher’s purpose.  Thus, French verbs, whether traditionally considered irregular or not, have stem-alternants for the tonic position (typically 1st, 2nd plural).


(11)           (a)        with no distinction between tonic and non-tonic positions:

dohn-  ‘give’

cour – ‘run’

romp- ‘break’

mett- ‘put’

(b)        with a distinction between tonic and non-tonic stem and 3rd plural calling for the tonic form:

            mén-/men- ‘lead’

            vien-/ven- ‘come’

            reçoiv-/recev- ‘receive’

            doiv-/dev- ‘own’

            boiv-/buv- ‘drink’

                        (c)        likewise but 3rd plural calling for the non-tonic form:

                                    fini-/finiss- ‘finish’

                                    assied-/assey- ‘seat (reflexive)’

                                    vau-/val/ ‘be worth’

                                    maudi-/maudiss- ‘curse’

résoud-/résolv- ‘resolve’

moud-/moul- ‘grind’


In this particular case, then, the traditional pedagogue overshot in the direction of itemization.


We have to arrive at the optimal degree to which a teaching grammar can be pedagogically deviant given the specific goal and situation.  This applies with even greater force to the fourth and last principle.




Finally, the greater the tendency to decide in favour of form rather than meaning as the basis for the specification rules, the more pedagogically deviant the teaching grammar will be.  To use a popular shorthand: the shallower, the more pedagogically deviant; the deeper, the less pedagogically deviant.  Thus, if a generative-semantic type is to be one’s theoretical grammar, it will have to be considerably transformed in the direction of shallowness before a practical teaching grammar emerges.  Its semantic insights will not be sacrificed, but recast as interpretation rules.


The substitution table of French non-tonic pronouns that precede a verb (other than the imperative affirmative) is just such a device that relies on the overt formal sequences without attention to the deeper sources.


(12)      je         me       le         lui        y          en

            tu         te         la         leur

            elle       nous

            nous     vous    





Constraint: Items from columns 2, 3, 4 cannot all three occur together.

(Note : The forms in the second column may be datives like those in the fourth column or be objects like those in the third column.  Further, datives of the second and fourth columns refer to animates and that of the fifth column refers to inanimates.  The table disregards these meaning distinctions and displays only the formal surface.  A theoretical grammar would perhaps have “normalized” the order and offered reordering rules or output constraints for computing the surface order given here.  A pedagogical grammar, on the other hand, will need to specify, in addition to the surface order formula above, rules for computing the case relations therefrom.)


We have so far pointed out that there are four ways in which a teaching grammar can deviate (but need not always do so) from a theoretical grammar and be more pedagogic in accordance with our definition:


(12)           (i)  grouping and grading:

(a)  gross grouping and grading by viability and essentiality respectively.

(b)  fine grouping and grading by relatability and teachability respectively:

                        (ii)  (a)  bias in favour of usage rather than system.

                               (b) bias in favour of transduction rule rather than specification rule;

(iii)  bias in favour of item rather than constraint;

(iv)  bias in favour of specification by form rather than by meaning

(shallower rather than deeper level for selecting the point of entry, the so-called initial symbol of analysis).


As we have said earlier, the best teaching grammar for a given purpose is the one which strikes the optimal balance between system and usage, specification and transduction, constraint and item, and specification by meaning (with expression supplement).


What remains to be done is to establish an evaluation procedure for optimizing the degree of pedagogic deviation for a given pedagogic destination and for the particular phase and unit of the grammar-2.  We are not saying: the more pedagogically deviant the better.  But we are making the claim that the pedagogic optimum will be on a level with or lower than, but never higher than, the theoretically demanded optimum.  Of course the available theoretical grammar may fall short of its own standards for lack of sufficient date, analytic acumen, and so forth.  (“Higher” means polarizing more towards system, specification, constraint, meaning; “lower” means polarizing in the opposite directions—i.e. those specified under (13) (ii) (a), (13) (ii) (b) (13) (iii), and (13) (iv) above.  Principle (13) (i) is not at issue here.)


If this claim is correct, we can predict that teachers will tend to choose that theoretical grammar as a point of departure which calls for the fewest adaptations to teaching purposes.  Naturally it is felt to be more economical to use theoretical and teaching grammars that differ as little as possible from each other.  For a similar reason, teaching grammars for teachers and teaching grammars for students tend to be modeled after each other.  In either case, we are predicting a tendency and not endorsing it.  Theoretical grammarians please note!





Bloomfield, Leonard.  1945.  On describing infection.  Monatshefte für Deutschen

Unterrichte 37:4--5.  8-13.

Clark, Richard E.; Poston, Lawrence, Jr. 1943.  French syntax list: A statistical study of

grammatical usage in contemporary French prose on the basis of range and frequency.  New York: Henry Holt for the Committee on Modern Languages, American Council on Education.                                   

Fries, Charles Carpenter.  1940.  American English grammar: Grammatical structure of

present-day American English with special reference to social differences or class dialects.  National Council of Teachers of English, monograph 10.  New York: Appleton-Century-Crofts.

Hale, Austin.  1974.  Grammars for the lay reader.  Read at the Seminar on Linguistics,

Tribhuvan University, Katmandu, Nepal, Nov. 1974.  (Since published: Seminar papers in linguistics, Katmandu: Institute of Nepal & Asian Studies, Tribhuvan University, 1976.  pp. 53-82; discussion pp.83-84.)

Halliday, M. A. K. 1964.  Syntax and the consumer.  Georgetown University Monograph

series on languages and linguistics 17.

Keniston, Hayward.  1937.  Spanish syntax list: A Statistical study of usage in

contemporary Spanish prose on the basis of range and frequency.  New York: Henry Holt for the Committee on Modern Languages, American Council on Education.

Targer, George L.; Smith, Henry Lee, Jr. 1951.  An outline of English structure.  Studies

in Linguistics, occasional paper 3.  Norman, Okla.: Battenberg Press.  Reprinted under various imprints.

Vachek, Josef (ed.).  1966.  Les Problémes du center et de la périphérie du systéme de la

langue.  Travaux linguistiques de Prague 2.  Praha: Academia, Editions de I’Académie Tchécoslovaque des Sciences.





I must gratefully acknowledge that Austin Hale’s “Grammars for the lay reader” (1974) provided me with the initial stimulus for thinking through this entire problem.  The first version of the present paper was presented at the Seminar on Materials Production, Central Institute of English and Foreign languages, Hyderabad, India, 9-13 February 1975.  The present version is substantially the same as the one presented at the Fourth World Congress of the Association Internationale de la Linguistique Appliquée at Strassburg in August 1975.  it has benefited from comments by Carol Kiparsky.  This was published in Essays in honor of Charles F.Hockett Ed. Frenceriet B. Agard et al. Leiden : Brull, 1083.