WAS CHOMSKY'S GREAT REVOLUTION
IN LINGUISTICS A NEW PARADIGM?

THREE APPROACHES TO ANALYSING LANGUAGES

©  2000-2003 by Orchid Land Publications

[updated 20030421]

      Whatever Noam Chomsky did, it certainly was a revolution and great advance in linguistics, after the list-mentality behaviorism that had prevailed till he came on the scene.  The greatest genius of linguistic analysis of all time, Chomsky of course took over some of the basic ideas of his predecessors.  But even where he failed to complete his revolution, i.e. when he took over an older idea like the virtual reality of a CHANGELESS (and therefore timeless) and invariant idiolect as the object of linguistic analysis, he got there by a very different route:  His predecessors positivism and the list mentality were replaced by positivism's opposite, rationalism, and a great systematizing mind.  These revolutionary approaches were accompanied by an increasing focus on universal grammar--and the brain.  

     In opposition to the regional and social dialectologists (mostly positivists who accepted the idiolectal premise and treated a language as a list of idiolects), Chomsky saw that a grammar is in a mind, not out there in geographical or social space where language is.  Generating a language represented a huge shift from just listing the items and entities of a language in the manner of structuralist grammars.  Chomsky's idiolectal outlook was balanced by his idea of a universal human grammar.  Both resulted from his early rationalist program and at least comported with the notion of a language as a unified system. 
    I have never seen convincing reasons for leaving variants out of individual grammars any more than ignoring language difference--something Chomsky does not do when promoting his universalism.  A timeless, changeless language was accepted because at first, no one could see any other way to study multifaceted language as a unity (Labov's positivist-statistical (non-theoretical) approach had settings for given isolects, but no overall language; he also adopted phonemics and similar usages from positivist linguistics.)   The Chomskian framework has apparently been too static or Platonic to see that development creates implicational patterns among variants--a cline or set of circles within circles of variants--that allows speakers to speak many isolects and understand far more . . . if they fit into one's language system:  They move from one variant to another that implies it or is implied by it, up and down the implicational cline; some of these arrangements get re-arranged as the result of unmarking a rule environment or rule ordering as well as by easily comprehended generalizations of rule forms, including environments.   (False analogies are over-generalizations of a familiar, often morphological, pattern to a situation where an irregularity is still preserved in the grammar or style in question.)   But Chomsky's later stress on evolution-created mechanisms has been anything but static.
     Chomsky and his students' showing that (before the language window closes in puberty for most people) children give evidence of a complex grammar that can predict an endless novelty of expressions they have never heard on the basis of a very partial exposure to their mother tongue.  Chomsky's minimalism is not purely methodological--after all language requires a great deal of redundancy to function under the conditions where it is used--but has a theoretical status.  Given the difficult conditions under which language is used, a non-varying minimalism would leave language shorn of redundancy would not be desirable; but that can be shucked off to "performance"--which, however, removes it from the system it forms part of.   Minimalism in the number of components and parameters is obviously a sign of progress.  Chomsky's minimalism gets rids of proliferated syntax rules.  Chomsky introduced a philosophical sophistication previously lacking in linguistics, plus a whole new approach to systematic analysis--an anti-positivist approach not afraid to employ abstract, or even virtual, entities far beyond the idea of an idiolect.  
       Despite the retained idiolectalism, it would be hard to deny that Chomsky's revolution was a new paradigm.   Those who remember the behaviorist days, the list mentality, the failure to distinguish language-acquisition from language-learning, etc., realize that Chomsky created a new epoch that continues still.  His present work exhibits a breadth of first-hand scholarly knowledge in many fields that others cannot boast.  His current emphasis on comparison (with non-human vertebrates; see Hauser, Chomsky, and Fitch, "The faculty of language:  What is it, who has it, and how did it evolve?" in Science 20021122, pp. 1659-1679) might open up two possibilities:  First, considering comparison among the variants of a language system, and a clearer differentiation of linguistics or cognitive science from grammatical analysis is made easier.  (I mention this, since teachers of grammar get shrecked by the Chomskian idea of a grammar's being a theory of language and most of all by their failure to realize that the major goals of linguistics and grammar are different, and that therefore most of its methods differ from those of practical grammar understanding.)  
     Second, if development were expanded from an emphasis on the origins of language to some attention to the rôle of the way in which a language develops--i.e. with ever-novel variations--and an emphasis on how many analogical errors simply propagate a language system by removing irregularities created by subsequently lost productive processes, that would (for me) represent even more progress.  But of course one person cannot deal with every question and has got to focus it on the things that are crucial for him.  Even Chomsky's minilectalist (idiolectal) assumption --that the object of a grammarians interest is a virtual grammar, one that is virtually invariant--has the positive value that, as just noted, a language like English is a unitary system.  But minimalism may not entail minilectalism.  The next step is, it seems to me, not to relegate the many (overlapping!) variants of a language to the outer Gomorrah of other analytical bin, especially if we wish to relate our work to the capabilities of any normal speaker's brain--to shift styles and to understand many stylistic, ethnic, regional, gender, and social varieties of one's langauge.  (This contention does not appear to be shared even by sociolinguists, despite the positivism of most of them, or by traditional grammarians, among whom the list mentality often reigns.)   

     Another necessity, in my view, would be to see the Old-French ancestry of the grammar system of Middle and current English, separating this from the etymological source of most of its functor words and 28% of its general vocabulary.  (Note that the calques of functors follow the Romance grammatical PATTERN, despite their etymologies:  cf. older the which with Old French laquele/liquels; for that with pourque; and because with à cause de.)
       Chomsky observes (wk. cit., p. 1578, cols. i-ii) that to the extent that it can be established "to what extent the computational system is optimal, meeting natural conditions of efficient computation such as minimal search and no backtracking"--things he has laid emphasis on earlier--"we will be able to go beyond the (extremely difficult, and still distant) accomplishment of find the principles of the faculty of language, to an understanding of why the faculty follows these particular principles and not others."
       Unfortunately, linguists' grammars (other than late, lamented J. McCawley's and one or another that I may not have seen) have tended to retain wholly inappropriate categories for English analysis--an issue addressed in these pages and to be dealt with in a forthcoming publication (English grammars versus the system of the English language).  One thing that all scholars trained in linguistics in the orientation developing out of Chomsky's work share with scholars in most (but not embraced by the positivist linguists of yore) is the necessity of explanation, something that the aforementioned publication harps on.

     Developmentalists assume (and demonstrate) the unitary nature of, e.g., English, despite its variants--but do not wave hands at the variants--which overlap for ages-groups, styles, class-differences, and regional differences.  (The basis of this was shown by Labov, but his statistical approach analysed the quantity [numb-ers], not the quality, of a language; and each variant almost seems to be an entity unto itself; at least, one generates but one at a time.)   If Chomsky decides to treat variants as "superficial," okay.  But a study of the late rules of phonetology shows that the progressive application of these changes pari passu with increasing tempos and decreasing formality represent the most natural, unconscious part of language.  Unfortunately from this reviewers point of view, Chomsky's rationalism (in contrast with mentalism) demands a static rather than developmental view of language.   When one realizes that, aside from the changes to unmarked ordering (that get initiated in informal speech), the implicational ordering of the rules reflects developmental ordering.  (It seems important to observe that the unitariness of language can be preserved with a full analysis of its variants in shown in Essays on time-based linguistic analysis, OUP, p. 21 and Ch. 4.)  It is also to be regretted that attested reversals in marked contexts (see Chh. 5-6 in the volume already cited) are not noticed, since even without their being explained (they can be explained up to a point) they represent spectacular evidence for the dazzling brilliance of the human language faculty ("organ" in Chomsky's usage).      

     Though Chomsky has shown a great deal of interest in the mental origins of language (singular), how a given language originates receives less attention.  (The origins of English are laid out in Ch. 10 of the volume cited above.)  
      Though Chomsky himself (p.c.) preferred Th. Kuhn's later article to that scholar's initial booklet on paradigms, Chomsky did initiate a new paradigm.  The underlying forms and sound units are virtual realities.  That was a radical departure from the overblown empiricism of pre-Chomskian linguistics.  Being virtual does not mean that such underlying entities are not described with natural parameters, for they indeed are so described.  If phoneticians can have real sound waves--though in England, phonetic transcription is traditionally confused with the "phonemic" transcriptions of the thirties and forties--semanticists can hardly have tangible meanings; yet (mental) meanings constitute the whole purpose of language.   Chomsky long preferred a more circumscribed, system-related approach to semantics, though he always mentions it as an essential element of language.  His rejection of his early rule-based analyses is a step toward a higher and increasingly unified view of syntax, a progression from multiplicity to the rationalist goal of the simple.  It is a worthy goal but one that is easier to achieve in syntax than in phonetology.   
      It seems regrettable that all of those who subscribe to the mental view of language and the unitariness of a given language and indeed language in general) cannot come together.  The brilliant Chomsky or one of his students may some day (he has, one hopes, many "days" left) take an interest in those aspects of language which he has ignored.  It would require his giving up hardly any, perhaps none, of his basic principles, though they would have to be integrated with real-world time (or at least temporality).  The development of language has not been simply mental evolution; language evolves daily before our ears.  Since it is true that the system does not evolve so much as details and the external values that society places on varying usages, and since it is true that what those variants communicate is clearly of less interest (to linguists) than the language meanings communicated by the system, studying variants would be of little interest if it did not reveal important aspects of the language "organ," viz. how the brain organizes and deals with the vast array of variations that it is so competent to deal so effortlessly with.  It is the writer's view that proper (i.e. non-quantitative, but so-to-speak qualitative) studies of variants do just that.

[comments/corrections by those more knowledgable on any point are welcome]

THREE PARADIGMS OR PHILOSOPHICAL APPROACHES 
TO ANALYSING LANGUAGES

      Let's call the approaches C[homskians], L[abovians], and D[evelopmentalists].  One can view language by its external appearances, as do the behaviorists and areal or social dialectologists.  One can view language externally, as do the behaviorist L, and "leave the mind alone," as Labov famously said, and set out from the geographical and social distribution of linguistic phenomena.  Or one can view language internally or imminently--in terms of its grammar alone, as do C (who stress the mental side of grammar) and D (who do not ignore the mental side, but also accept the physical side of language as playing a rôle in explanation) and be concerned about psychological reality.   Where C is concerned to learn about mental reality through language and ignores external time and space, D is of course concerned with time as a creator of the implicational patterns discussed below.  None of the three deny that external factors affect a grammar--if only on the level of borrowing from other grammars--though L thinks that analysis begins with external factors.  D accepts that languages begin by creolizing, especially where they are sufficiently different not to be viewed as dialects of one language:  The language of a socially dominant group furnishes the grammatical structure, while the language of the suppressed furnishes the sounds (including prosodics) and words that calque the function words and other words of the dominant group.  (Cf. the way Middle English calqued Old French laquelle as "the which," uninflected  qui? as uninflected "who?," and the half-dozen or more prepositional phrases of French corresponding to Germanic for[e] as "on behalf of, in place of, " etc.; and one Germanic word order is found in northern French.  Of course, sounds may coïncide in both languages, as did Old French and Anglo-Saxon "th.")
     Chomsky come to treat grammars as a set of principles whose parameters get set for each language and language variant.  All variants canin principle be generated serially from a rule-less grammar.  The idiolectalism of the past has in prinicple if not in fact been given up.  Labov does the same for the grammatical fragments he has so ingeniously analysed.  D would establish the implicational patterns of a language that include all variants in a unified grammar.   (Only a half-dozen rules of English are really complex and require three or four special conditions.)  With the help of Bailey's view of markedness--which includes W, Mayerthalers reversals in marked contexts--allows one to predict new variants, as what is marked gets unmarked . . . something that is surprisingly stable even when borrowed elements come into the language.  
     Given that the third approach accepts Chomsky's rejection of extragrammatical findings other than those connected with an analysis of the physical human nervous system but also accepts--what Chomsky rejects--the possiblity of studying a grammar with variants (across time) in favor of a virtual grammar with at most trivial variants, the difference between Chomsky's thought world and this is clear.  See models and examples of this analysis below.

Markedness is an aspect of language universals that does not seem to be important for Chomsky in his stress of universal grammar, since his emphasis is on syntactic structures.

      Discussions of this sort don't always focus on what is desirable and well as possible in explanatory linguistics.  To understand what follows, see the graphics at the end of this writing.   Since the progressively less formal styles have progressively more rules, it is generally true that less formal phenomena imply more formal phenomena.  The rule of change is the converse as speech expands (even by deletion) in the direction of increasing formality.  Since markedness is defined by change (it can be shown that time defines markedness values), it is redundantly true that change goes from more-marked to less-marked.  
     That this approach requires rules can be shown with a simple example.  Suppose some styles of my speech drops underlying or virtual //t//  in the second syllable of tentative, while other styles realize the //t// as some kind of apical stop.  (The //t// in the third syllable varies between [d] in normal speech and [t] in formal, especially contrastive, styles.  Note that //nt// become [nd] between unstressed vowels in the few occurrences of this phenomenon in English; see Bailey, English phonetic transcription [SIL/UTA, 1985], Ch. 7.)   The absence of an apical stop in the pronunciation of the second syllable of tentative implies the presence of an unaspirated [t].  It is possible to arrange all of the eighty or so pan-English rules in this fashion--relating the application to the rule in question to its non-application.  An enormous number of changes result simply from reordering pairs or triads of rules from marked to unmarked order (see Bailey, "Variation resulting from different rule orderings in English phon[et]ology" in New ways of analyzing variation in English [GUP 1973], pp. 211-252).  Applications to syntax are meagre, since English lects vary a great deal less on the syntactic level.  Yet there are examples of variation on that level also.  (The reader is referred to ETBLA, from which the following graphics and table are taken.  Ch. 1 is on general principles, while Chh. 5 and 6 deal with reversal of marked usages in marked contexts; Chh. 7-8 deal with change and history, while Ch. 10 shows on grammatical principles that English is not a Germanic language but must have begun as a creole of Old French [itself influenced by Frankish and then Norman--both Germanic languages].   SEE ALSO HERE.)  Click on the thumbnail of each graphic to bring up the full-sized picture.

Sketches of the basic idea of language birth 

and variation analysis

Solving two of the most intractable problems of dialectology
with variation theory

     The foregoing graphics are taken from the author's Essays on time-based linguistic analysis, of which Ch. 4 shows how the dead ends and fugdgings necessary in extra-grammar (spatial, social, etc.) dialectology get simply resolved with grammar-internal variation analysis.  The idea being advocated is to ignore spatial and social distributions of variant uses and to arrange them within the grammar implicationally.  It turns out that if a person of a given social class, age, and sex says S in a given style and place, s/he will also say R, Q, P, O, N, M, etc. in the same circumstances and--more importantly, everyone who says S in any circumstance will do likewise, given the locutions that allow for the R, Q, P, etc. to apply.  (It should be noted that the implicationalities are not so simply that they display a single path; there may be multiple paths.  Further, it is understood that re-ordering the applicational rules for given pronunciations lead to the (or an) unmarked ordering; see my "Variation resulting from different rule-orderings in English phon[et]ology" in New ways of analyzing variation in English (Georgetown University Press, 1973, pp. 252).   There are various reversals treated in Chh. 5-6 of Essays that have to be taken into consideration in the overall pattern; but re-ordering a rule can easily be treated as a quasirule in the implicational array.  The whole thing rests on a single grammar--which, as Chomsky might well say, is evidenced by the way Americans of all classes, etc., understand the Scots, Irish, Welsh, northern and southern British, Aussies, Kiwis, and even New Yorkers and other Yankees, Southerner Statesers, and Hawai'ians.  
     Ch. 10 of Essays shows the impossibility of considering English to be Germanic--a lineal descendent of Anglo-Saxon, whose grammar is vastly unlike that of English.  This treatment is based on what has been learned in recent decades about how languages get born.

    Whether the brain prefers the neater three-valued features (as some neurological experiments seem to indicate) or binary features (as a straight-forward look at synapses indicates) depends in some degree on whether there is an either/or or even more independent third value between 1 and 0.