Mental Grammar

You might also like

Download as pdf
Download as pdf
You are on page 1of 13
S) Grammar and psyechotinguistion $17 Now do spechent sentences? How do we produce nt a mn! ae processes? useful for the notat) ‘denied that speakers learmed of used such ceaons. scill ocher linguists, perhaps a majority, simply sRoided the dispure altogether and kept on doing whatever sc was that they were doing (making descriptions of various aspects of language) and leaving it to others to worry sbow the ultimate psychological nature of their formulations In the 1960s Chomsky came down heavily on the side of psychological linguistics. His thesis, that linguistics is 9 branch of cognitive psychology, has become so widely held chat few Ynguists today oppose such a view. Even many of Chomzky's staunchest linguist critics are in accord on the goal of linguistics as involving the description of knowledge that people have about language 5.2 Chomsky’s competence and performance distinction” in addition to using the term “gramumar’ to indicate 3 cercain kind of language knowledge. Chomsky uses the tern competence with dhs same sense. Competence isthe knowledge that people have of the grammar of their language and, as such, ic is the goal of linguistics to describe this competence. A linguist will be successful to the extent that his or her tocnulations arc true descriptions of the knowledge that is in people's minds. if the specification of competence ts the primary Now, goal of linguistics, what chen should be the goal of psycholinguistics? In Chomsky’s w. psycholinguistics has two major goals: (1) to specify how people usc competence so that they are able to produce and understand sentences: and (2) to specify how people acyuire competence (grammatical knowledge). It 1s the first of these goals, the use of a grammar for the production and understanding of _ that will be of concern to us here. The second is ther chapters of this book. , the activities involved in producing and it fe processes. Com- Farmar) is just One part, albeit an essential @ crucial performance functions A theory sentence production, i.¢ of performance should explain how ypeakers ke idess and make them ino sentences which are rendered into speech sounds A theory of performance also should explain sentence comprehension, Le. how speakers, on receiving speech sounds, recover ideas from these sounds, In both of these performance proces: according to Chomsky, one grammar, iL. competence, is used _ The relationship of competence to performance for Chomsky, therefore, is me of part f whole, with competence being a part or component of the whole, which is, performance. Competence is the knowledge that persons have of their geammar while performante involves know- ledge for using Seoppereeet so that the processes of sentence production and understanding can be realized 5.3. Chomsky’s grammatical conceptions Before continuing with a discussion of mental grammar and performance, which will be presented in the nexc chapter, it ws first necessary to consider various conceptions of mental ee In this regard. consideration of the ideas of homsky, who is the world's foremost grammatical theorist, is of the highest priority. Im 1957 Chomsky came out with his book, Syntactic Suruerures - 4 remarkable book that revolutionized linguistics and influenced a number of the socialsciences, especially psychology and anthropology. With his notion of a system of nudes, phenomena which hitherto could not be explained could now be dealt with. fn particular, Chomsky demon- strated how such a system could be used to explain how speakers can, in principle, produce and tunderstand an infinite number of grammatical sentences. (The first chapter of my 1982 psycholinguistics book provides 1 description of Chomsky’s brilliant and ideas.) Since the advent of ky, linguistics (and psycho- linguistics) has never been the same. Over 35 years later, we find. Jeading the field. With the help of sble to solidify his revolution with new On the average, Chomsky shakes up decade. Following Syntactic Stuc- pe im the 195Gs, there was che Anpecst Gremiend Phooey) grammar i the 1960s Then there 28 the Rasemded Standacd Theory grammar io the 197 fobowed by che Goverament/ Binding grammar ie ol Ties, Mow. ia the 1A once again we hes rumbling wish Chomsky proposiag 10 abandon Desiwonure. th deep tewet of structure the name of which ha bece famdiar even co many ouside of linguistics Sell, ie far-reaching changes in the details of vammars, Chomsky has continued te memtasn or ‘mation, which is that the syntax of © ge i primary. with meaning (and sound) bees (Abandonment of D-strveture, as Chomesy now suggests. in no way implies that syntax would not continue to be primary.) This is to say that the meaning © sentence is specified as 2 function of its syntactic form, snd mot vace versa. This relanonship, which Chomsky claims to be innate and umiyersal, is shown in Figure 5.1. This idea of how a grammar is tw be organized, however, has the effect as shall be argued later u: this and che following chapter, of rendering Chomsky’s essential grammatical theorizing 25 ically invalid. By his own criterion, the linguists grammar must be psychologically valid. ss haa bs Meanie eee Sound Figure 5.1 Relsoonstip of syntax, meaning and sour st ¥ s 5.3.1 The Standard Theory it was in 1965 that Chomsky proposed, in his Aspects of the Theory of Symax, the theory of grammar that became the Standard Theory (ST). While this version has by later formulations resulting in his (GB) theory, it is useful w get sone Standard Theory before desting wah essentially consists of various sc of rules syntactic, semantic and phonological Each ser of rates is systematically integrated and serves 1 provide, for every seovenee, a linguistic description oF €e tation at four different levels. There is + sound level (Phonetic Interprers Goa) where the phonetic sound pattern of a sentence ix represented; there is a meaning level (Semantic lnterprew tion) where the meaning and logical relanons in 2 senvence Are represented; and, there are two syntactic levels (Deep Serweture and Surfiee Structure) where various syntactic aspects of a sentence are represented. Deep Structure rae the underlying syntactic form of the sentence while Surface Seructure represents its more overt form - Axxchema of the Standard Theory geammar is shown in Figure 5.2, This grammar, while incorporating ideas from his earlier Syntactic Structures’ grammar, is much more complex and comprehensive. In the schema we see that the grammar consists of various componcnts ~ the syntactic, semantic and phonological ~ and a lexicon or vocabulary repository which consists of morphemes and idionis The syntactic component consists of wo types of syntactic cules Phrase Structure cules (also known as Base rules) and Transformational rules. Each of these sets of syntactic rules is responsible for an output: the Phrase Structure rules provide Deep Structure while the Transformational rules, operating on Deep Structure, provide Surface Structure. The jeal t consists of Phonological rulcs which operate on the Surface Structure to provide the Phonetic Interpretation of a sentence. The semantic com- consists of Semantic cules which operate on the same Surface Steucture to provide the Semantic Interpretation ofa sentence. Let us now look at the contents of the various components of the grammar and see how they function to generate representanons. Phrase Structure rules, Lexicon and Deep Structure h ure (PS) rules provide the basic con- a sentence. Essentiatly, they provide an into its underlying, phrases which arc words and word parts, These weeds mes) and idioms are stored in the But whai is the relevance of all this to the semannc expressions? First of all, « competent users of a language are mterpret indefinitely many of that language. Since itis tenpossible for anyone to have learned the sense of every composite expression in the way that he, presumably, learns the sense of lexemes, there must be some function which determines the sense of composite expressions on the basis of the sense of fexemes_ Second, it is reasonable to assume that the sense of a composite expression is a function. not only of the sense of its component lexemes, but also of its grammati- cal strecture. We have made this assumption throughout. and it can be tested empirically in a sufficient number of instances for us to accept it as valid. What we want then. in the ideal, is a precisely formulated procedure for the ae reemetarticalty com posite expressions in a language. coupled with a pro- Stare for determining te sami fe, if any. of each or stage of syntactic composition. This is what semantics seeks to provide. Semantics, as such, is not committed to any theory of syntax. Nor docs it say anything in about the closeness of the correspondence be- @ syntactic and semantic structure in natural lan- SO en a eens eh opines on exch of then 1 will consider only two of the best known i of determining that composi- it is) which assigns sense to expressious of natural a ret Sentences ave boon made vo basic the data truth-condionalty Anyone who doubts this is invited to compare the (Weal meat of ‘any’ and ‘some’ in older and more recent al grammars of English. aot to mention schoi arly articles on the topic He will see the diflerence unmediately What follows is a deliberately simplified account of some of the principal concepts of formal semantics that are of proven relevance to the analysis of the propos tional content of sentences. Limitations of space oblige me to be selective and, at times, more allusive or Gog matic than I should like to be. The allusions arc for the benefit of those who already have some background in linguistics. The dogmatism will be apparent to those who do not share my views. | lake no account of anything other than what is uncontroversially a part of the poposi- tional content of sentences in English. Compositionality The principle of compositionality has becn mentioned already in connection with the sense of words and phrases. Commonly described as Frege's principle, it is more frequently discussed with reference to senience- meaning. This is why | have left a fuller treatment of it for this chapter. It is central to formal semantics in all its nts. As it is usually formulated. it runs as follows (with ‘composite’ substituted for ‘complex’ or *compound’): the meaning of a composite expression is a function of the meanings of its component expressions. Three of the terms used here deserve attention: ‘mcean- ing’, “expression” and ‘function’. 1 will comment upon mand then say why the principle of important, have seen, can be given various restrict it to descriptive meaning, of . we can still draw a distinction ———— Ma between sense aod denotation (see Chapter 3). Frege's awn distinction between sense bnd reference (drwwn German by means of the terms ‘Sinn’ and Bedeutung’) is roughly comparable, and is sceapted in broad outline, if not in detail, by mos formal semant fi take the principle of compositionality to apply to sense. ression’ is usually left undefined when it used by linguists. But it is normally taken to include sentences and any of their syntactically identifiable con. stituents. I have given reasons earlier for distinguishing, expressions from forms, as fai words and phrases are concerned. For simplicity of exposition. | will now include sentences among the expressions of a language. More important, 1 will assume that there is an identifiable subpart of every sentence that is the bearer of its propos tional content, and that this also is an expression to which the principle of compositionality applies. The term ‘function’ is being employed in its math ematical sense: i.c., to refer to a rule. formula or opera tion which assigns a single value to each member of the set of entities in its domain. (It thus establishes ether a many-one or one-to-ont correspondence between the members of the domain, D, and the set of values, V: maps D cither into or on to V.) For example, there is an arithmetical function. normally written y = x’, which for any numerical value of x yields a single and determinate numerical value for x°. Similarly. in the propositional calculus there is a function which for each value of the itional variables in every well-formed expression maps that expression into the twb-member domain (True. False}. As we saw earlier. this ig what is meant by saying that composi tions are truth-functional. 1 have 4 this out in more: detail and deliberately . m , several of the techni- tet Ge we at B ECE powered evel, and 1 lt sommes co whet G UNCON OST: & water of eon of ok Cate Riomel combcet. The [eo approaches wr COhoenskyan transiormavonal gramma oe she ome ell gremenar. om Ube othe: f The Katze-Fodor theery What I wall refer ve as the Katz-Fodor theory of seatiemee- > as 2 theory of formal not deal with here. ladeod. 1 wali not twee aticmge to Et a fall accowmt of the Kaw-Fodor theory im ary of & wersions. | will concentrate upon two things pemeerecr Se 14g Seatenees Hoe simplicity, 1 will discuss the Kate-Fodor theory j, wheat may now be thought of as its classical version, nor ,, as oniginal formulation, but as it was presented in th, period immediately following upon the publication > Chomsky's Aspeces. The main consequence. as far as th, general purpose of this book is concerned, is that I sha be taking for granted a particular nowon of deep structur, which has now been abandoned by almost all linguist: including Chomsky. The arguments for and against th. clagical notion of Structure are interesting an important. But | menses go into them here. Nor will burden the text with unnecessary technical detail. Th advantage of operating with the classical notion of dee; structure, in a book of this kind, is that it is more familia to non-specialists than any of the alternatives. And wha Thave to say about projection rules and selection restric tions is not materially affected by the adoption of on view of deep structure rather than another, or indeed b: the abandonment of the notion of deep structur altogether. According to the standard theory of transformationa grammar, every sentence has two distinct levels of syntac tie structure, linked by rules of a particular kind calle transformations. These two levels are deep structure anc surface structure. They differ formally in that they ar generated by rules of a different kind. For our purpose: the crucial point is that deep structure is more intimate!) connected with sentence-meaning than surface structure is. Surface structure. on the other hand, is more inti mately connected with the way the sentence is pro nounced, (As we shall see in Chapter 9, the process o ing sentences, whether spoken or written, is mor ly non-technical formulatior surface structure and pronuncia Omitting all but the bare essentials relation between syntax, semantic . as in Figure 3 (cf ct Man-made Medals 1 Wale reference ie) thin Hingreie. we sine see Heat Set eramenar (4 the broadest sense Of thee toner) comnperners i transformational grammar The deep sarocture of a sentence is the output of the base component and the component component; the surface structure of a sentence is the output of the transformational component and the input to the phonological listed, in what we may think of as dictionary entries_in the lexicon. They tell us, in effect, which pairs of lexcmes can combine with one another meaningfully in various gram- the adjective ‘buxom’ can modify nouns like ‘girl’, ‘woman’, ‘lass’, ete., but not “boy’, ‘man”, “lad”, etc., that ‘sleep’ can take as its subject nouns like “boy’. ¢tc., but not such nouns as “idea’ or ‘quadrupiicit on. If the selection restrictions are violated. the projec- tion rules will fail to operate. Consequently. they will fail s of the same phi and sentences as able, For example, the word "house = none of the senses (“housewife ,") it ho keeps house, in another (housewife,”) it denotes & pockel sewing kit Many phrases ia which ‘housewife’ is: modified by an adjective (good housewife’. “beautiful housewife’, etc.) will be eomperne ty ambiguous. But ‘buxom housewife", pre- sumably, a pi since “housewife,", suite “housewife,”, cannot combine with ‘buxom’. In general, then, the selection restrictions will tend to cut down the qumber of interpretations assigned to lexically composite expressions, In fact. the failure to assign interpreta- tion at all to a sentence, referred to in the previous roe be seen as the limiting case of this process. rules select from the meanings of an expression those, and only those. which are compatible with the (sentence-internal) context in which it occurs. The Katz-Fodor theory of sentence-meaning is formu- jated within the framework of component analysis (see Chapter 4). For example. instead of listing, in the lexical entry for “buxom”. all the other lexemes with which it can or cannot combine, the theory will identify them by means of one or more of their sense-components. It might say (in an Be cing formal notation) that “buxom cannot be ined, in semantically well-formed expres- sions, with any noun that does not have as part of its meaning the we seen, ts HUMAN and FEMALE. AS tial analysis runs into quite serious if it is pushed beyond the proto-typical, of expressions. It is for this reason that examples used by linguists to Katz-Fodor selection restric- But I am not concerned, at

You might also like