The Basis of Language: The Human Mind/brain

We have by now, fairly substantial evidence that one of the components of the mind/brain is a language faculty, dedicated to language and its use - where by "language," now, we mean human language, not various metaphoric extensions of the term. Noam Chomsky, Language and Thought, 34.(1)

1.0 Introduction

To Chomsky, language use, language and language acquisition are explained by facts about an individual's psychology and ultimately by facts about the individual's brain. I shall present this position of Chomsky in the following order. 1.1 is concerned with his way of explaining language use, in particular, meaning. He takes linguistic activities as instances of language performance, which we can trace back to knowledge of language (or linguistic competence), and further, to the internal mechanism (I-language) that is cognised by a language user.(2) In 1.2, I shall introduce Chomsky's conception of I-language and account for its intensional character. In his view, having an I-language is reaching a mature state (S_L, or sometimes abbreviated as S_S) of the language faculty of the human mind (which means, an array of capacities specifically dedicated to language acquisition, mastery of linguistic principles and rules, and language use.) Then our discussion will move from S_L to the initial state (S₀), that is, the innately given condition of natural language. A child is supposed to be biologically endowed with a universal grammar (UG), which consists of a limited number of principles and parameters. We shall consider the rationale of this principles-and-parameters (P-P) approach in 1.3 and some hypothetical idea of Chomsky for explaining language variation on that approach in 1.4. Another consequence of the P-P approach is the assumption that language growth (or acquisition) is a process of parameter-setting. This subject matter will be the topic of 1.5.

In these sections, I hope to show that, according to Chomsky, linguistic phenomena have their origin in the human mind/brain. (In his view, the mind, of which the language faculty is one component, is a natural object, in the sense of being an abstract description of the brain. That is why he uses the sign '/' in the term 'mind/brain'.) Chomsky says, "The basic explanation must lie in the properties of the language faculty of the brain."(3) The natural constraints of language, Chomsky maintains, are biological facts, uniquely pertaining to human beings.

A query arises as to how I-language is on the one hand taken by Chomsky as an "abstract entity", but on the other as a set of facts of biological nature. This does not involve a conceptual problem for Chomsky. In 1.6, before closing this chapter, I shall give a brief answer on behalf of Chomsky to clarify his position.

1.1 Explaining the use of language

Our words are amalgams of some sound and meaning, and put in some order to express our feelings, thoughts, information, beliefs and so on, for different purposes in life. We also understand others by interpreting their words, which may involve certain background knowledge, perspectives, and presuppositions. Chomsky is well aware that in actual circumstances, language use is guided and affected by many factors, which he thinks are not amenable to scientific studies. But in his view, some general properties of meaning and sound can be identified and explained by internal principles and rules. His major ideas for adopting this internalist approach to explaining language use will be the concern of this section.

1.1.1 Common language and common thought

Many people, philosophers included (4), tend to think that linguistic activities are instances of the use of a common language or publicly accessible set of well-formed sentences, like English, Chinese and German, etc. It is contended that members of different communities inherit different systems of language, and that such a shared language is passed on to the next generation with a "common treasure of thoughts" . The main reason is that without the same language and thought, communication is impossible.

Chomsky is sceptical about this view. Above all, he queries whether the pre-theoretical notions of common language and of common thought are useful or coherent. There are intermingling factors that determine our identifying a language or dialect in the real world: "colors on maps, oceans, political institutions, and so on ..."(5) At most some rule of using the word 'language' in various contexts might be generalised, but, it seems to Chomsky, this lacks any explanatory value. Moreover, it is not merely that the notion of a publicly accessible language is exceedingly vague. The problem is that it is hopeless to abstract such a 'thing', whose identity is relative to different perspectives and circumstances: shifting interests, expectations and power structure. It is just our loose practice, Chomsky contends, to say that there are Chinese, Dutch, German, Danish, Swedish and other languages in the world. In his view, we do not literally mean that these languages exist as some abstract entities independent of our cognition of them. Chomsky offers some examples as follows. "We speak of Chinese as 'a language,' although the various 'Chinese dialects' are as diverse as the several Romance languages. We speak of Dutch and German as two separate languages, although some dialects of German are very close to dialects that we call 'Dutch' and are not mutually intelligible with others that we call 'German.'"(6) And to say that we have the same thought and language is no more than to say that we think (very much) alike and talk (very much) alike. The claim that we possess something called 'a thought' does not seem to resemble the claim that we possess a diamond, since our thoughts are much more subject to various interpretations under different circumstances.(7) " Communication is a more-or-less matter," says Chomsky. To him, a fair degree of mutual understanding does not depend on any set of public meanings and of accurately determined pronunciations.(8)

1.1.2 Reference

Of a piece with the idea of common language and that of common thought, is the idea of a predetermined framework of reference. The publicly shared language is supposed to represent the world, and a linguistic sign is supposed to refer to an object. Chomsky puts into doubt this conception, which takes reference to be a two-place relation. Take London as an example. Using the name 'London', we can talk about its population, life-style, location, the people living there, the buildings, and the air, etc. Chomsky thinks that there is no object or thing-in-the-world, which embraces such "properties of the intricate modes of reference".(9) Another example: the referent of the word book in the sentence The book which he wrote in his mind weighs two pounds is simultaneously abstract and physical. Chomsky asks: Is there any object like this? In his view, natural language consists of, among other things, nouns, whose function is not to supply a determined set of references in the external world. For as a matter of fact, the object names we use in ordinary life are not the same as those employed in sciences.(10) "In general," Chomsky suggests, "a linguistic expression provides a complex perspective from which to think about, talk about, and refer to things, or what we take to be things" rather than some external references.(11) Take the word see. People who know English can be certain that the speaker of the sentence I see the house sees at least the exterior surface of the house; one would not imagine the speaker standing inside the house and looking at a corner of it. "If I am inside an airplane, I see it only if I look out the window and see the surface of the wing, or if there is a mirror outside that reflects its exterior surface."(12) Take near.If Mary is inside the house, we cannot say Mary is near it. Nor are we correct to say that something inside a container is close to it.

Reference, according to Chomsky, is a relation involving, apart from the sign and referent, the person who uses the sign and the circumstances under which the sign is used.(13)

1.1.3 Meaning and interpretation

The fact that Chomsky highlights the actual contexts where a language is used and where a thought is conveyed does not imply that he abandons the analytic-synthetic distinction. According to him, it is clear that words have an internal structure, and are analytically related. For instance, the word give necessarily involves a subject, object and theme, and the words persuade and chase necessarily have to do with intentions.

He admits the contingent and complicated character of interpretations in real life, where there are no rigid rules to rely on. But it does not follow that he thinks that there is no explanation of language use at all. Meaning and interpretation, he affirms, can be studied.

Besides (i) analytic relations between words, Chomsky suggests, we can also investigate the following aspects of meaning:

(ii) referential dependence (the reference of anaphors like themselves and each other, and pronominals like them, are dependent on some antecedent word(s) in a given sentence),

(iii) thematic roles or θ-roles (like the roles of agent, patient and goal) and thematic relations (like comment, topic and focus), and

(iv) quantification and scope, (for instance, the interpretation of this scope-ambiguous sentence Everybody loves somebody depends on whether somebody lies within the scope of everybody, which means that the sentence is paraphrased as 'Everybody has somebody whom he/she loves. If the scope relation is reversed, then the sentence is to be paraphrased as 'There is someone who is loved by everybody').

1.1.4 Positing an internal mechanism

The aspects of meaning just mentioned are shared by all natural languages. One may generalise certain principles and rules to account for these linguistic phenomena. However, Chomsky argues, they are not (internalised) conventions; they are not explicit linguistic knowledge taught to the child by his/her parents or caretakers. Consider the following sentences (1a, 1b, 2a and 2b)(14), which are similar and yet have very different semantic properties:

(1a) John is too stubborn to expect anyone to talk to Bill

(1b) John is too stubborn to visit anyone who talked to Bill

(2a) John is too stubborn to expect anyone to talk to

(2b) *John is too stubborn to visit anyone who talked to

When Bill is deleted, John, in (2a), becomes the object of talking. (2a) may be paraphrased as: John is too stubborn that one cannot expect anyone to talk to him. Yet one can hardly suppose that in (2b), the person to whom someone talked is also John. Unlike (1b), (2b) is meaningless; one can hardly assign any sense to it.

On the basis of a limited linguistic experience, children are able to understand these sentences, which are probably new to them, without having undergone any specific training. So Chomsky thinks that the only promising way to explain meaning (and interpretation) is to posit internal principles and rules of derivation and representation. These principles and rules are to a great extent innate and universal.

1.1.5 Competence and performance

The production and perception of speech, like those of meaning, are best explained in terms of mental representations and rules, according to Chomsky.(16) A speaker/hearer possesses a lexicon, and in addition, he/she masters a computational procedure of language (C_HL). The lexicon consists of a definite number of lexical entries. Each includes information of three types: (i) phonological, (ii) semantic and (iii) formal, for the interpretations of sound, the interpretations of meaning and the operations of C_HL, respectively. (17) The lexicon and C_HL constitute the language that the speaker/hearer has. Chomsky calls it 'I-language'. (18) When the computational system operates, his/her lexicon is accessed. The products are certain derived, structured, strings of mental representations of sound and meaning. The operations of C_HL are insulated from any influence of the individual's belief or other cognitive systems.

A speaker/hearer's mastery of C_HL and possession of a lexicon, Chomsky calls 'competence'. It is regarded as a kind of knowledge, not practical abilities or dispositions. Linguistic activities are just manifestation of this knowledge, albeit various factors (memory, attention, ideas, and social expectations, etc.) are involved in the use of language in concrete circumstances, which Chomsky calls 'performance'. These are contingent factors, which are relevant in shedding light on particular cases of language use. But in linguistic explanation, it seems to Chomsky, competence, rather than performance, should be given the primary status.(19)

1.2 I-language

'I-language' refers to an inner state of (i) an individual's mind/brain, which has a certain independence from other elements in the world. The idea that a linguistic community speaks the same language is accordingly derivative. It only suggests that a group of people share more or less the same I-languages. An I-language is (ii) an internal mechanism, conceived as a function specified in (iii) intension, which generates an infinite number of, not only well-formed sentences like (3a), but structural descriptions (SDs) like (3b) (whose constituents are hierarchically organised as indicated by labelled bracketing):(20)

(3a) David bought the house

(3b) [_S [_NP [_N David]][_VP[_V bought] [_NP [_DET the] [_N house]]]]. (21)

The notion of I-language is meant to be opposed to that of E-language. The latter notion suggests that first, in describing a given set of sentences, a number of grammars can work equally well. As alternative axiomatisations in mathematics clearly show, different axiomatic systems can generate the same set of theorems. These grammars or axiomatic systems are said to be extensionally equivalent. An extensionally equivalent grammar determines "recursively, the same infinite set of well-formed ... sentences."(22) Second, a particular (extensionally equivalent) grammar has no difference from another that generates the same set of well-formed sentences; what is at stake for the E-language approach is not grammar but a delimited extension of well-formed sentences. And third, it does not make sense to say that one grammar should be better than another extensionally equivalent one, and that one grammar is correct while another is not.

Above all, Chomsky points out, advocates of the concept of E-language take language to be independent of the properties of the human mind or brain.(23) As a consequence, there is no determinate way for analysing a language; all characterisations of language are thus contingent and relative to certain interests and purposes. It seems to Chomsky that this arbitrary character of E-language prevents it from being the object of scientific inquiry. For Chomsky, what a linguist is concerned with are facts of nature, not artefacts.

We shall highlight Chomsky's major reasons for defending his intensional concept of language in 1.2.2, after providing, in 1.2.1, an essential sketch about the internal mechanism of language (C_HL), assumed in Chomsky's current research program.

1.2.1 C_HL

As Chomsky remarks, "The rules of the language are not rules of some infinite set of formal objects or potential actions but are rules that form or constitute the language, like Articles of the Constitution or rules of chess (not a set of moves, but a game, a particular rule system)."(24) So to Chomsky, linguistic rules or principles are not subsequent to a set of well-formed expressions as characterisations of them, but descriptions of the computational system of language, that is, a function mapping a given set of lexical items to a sound-meaning pairing (π, λ). These items, each of which is supposed to be a mixture of formal (for instance, number, tense, gender, case and so on), phonological (for instance, green is pronounced as /grin/, not /gren/) and semantic features (for instance, airplane [artefact] and run [action]), are selected one-by-one, and merged (that is, combined) in a pairwise and successive manner to form new and larger syntactic objects, such as the steps involved in (4).

(4) You will find it.
i. Select it.
ii. Select find.
iii. Merge it and find to form a structural unit (a verb phrase (VP)) find it.
iv. Select will.
v. Merge find it with will to form a higher-ordered structural unit will find it.
vi. Select you.
vii. Merge you and will find it to form you will find it.(25)

The output of this procedure is a couple of structural representations. One is phonological, to be interpreted at the interface level called Phonetic Form (PF) (that is, to be determined by the rules of the phonological component of the mechanism); another is semantic, to be interpreted at the interface level called Logical Form (LF). These interpretations are instructions to the articulatory-perceptual (A-P) and conceptual-intentional (C-I) systems, respectively. So it is assumed that there should be a point at which a chain of computational operations is split off. After this point (called Spell-Out) semantic operations continue 'covertly' (in a sense that they have no phonetic or morphological realisation) to produce the LF representation at LF (like quantifier raising: from Mary visited everybody to [Everybody_i [Mary visited x_i]). And at the same time new features and elements (like intonation contour, and perhaps temporal order), are introduced to the phonological features of the syntactic construct at PF to produce the PF representation on another side.

The idea is that each element of an LF or PF representation should be interpretable, so that it can be accessed by the C-I or A-P systems which are connected to C_HL, respectively. This is a crucial representational constraint, called the Principle of Full Interpretation (FI), which is 'optimal' or 'economical', in a sense that it does not allow any uninterpretable formal features to remain in the final product of a derivation of C_HL. They are supposed to be 'erased' during a series of feature-checking steps. Compare (5a) and (5b), the former being a grammatical sentence while the latter an ungrammatical one:

(5a) He has come.

(5b) *Him has came.

The impropriety of (5b) is due to its violating FI. All the uninterpretable formal features have been 'erased' in the case of (5a), not in the case of (5b).

Let us consider the interpretable and uninterpretable features of the words in (5a) first. The number feature of the word He indicates that what it refers to is only one object, this being an aspect of meaning and thus interpretable at LF. Whereas its nominative-case feature is uninterpretable, since there is no difference in meaning between he and him in She believes he has come and She believes him to have come, respectively. Different from the word He, the function of the number feature of the auxiliary verb has is just to agree with the number feature of its specifier (that is, subject). But the present-tense feature of has is interpretable, insofar as it indicates time. Regarding the word come, its being an n-participle (that is, perfect or passive participle) here is a consequence of the auxiliary verb, and so is uninterpretable (that is, plays no role in semantic interpretation) at LF. There are requirements on the auxiliary verb has that its specifier be in nominative case and a third-person singular noun phrase. These restrictions are regarded as the specifier-feature of has, which motivates its checking with the word He. Their matching renders the nominative-case feature of the word He and the specifier-feature of the word has - both of which are uninterpretable - to be erased. The word has also has a complement-feature, that is, a restriction such that its complement be in the form of n-participle. So the words has and come match; the former's uninterpretable complement-feature and the latter's uninterpretable feature that it is an n-participle form are erased.

The checking breaks down in the case of (5b). The objective-case feature of the pronoun Him is not compatible with the specifier-feature of the auxiliary verb has, which restricts its specifier to be in nominative case. And the feature of the word came that it is in the past-tense morphological form also does not match the complement-feature of auxiliary verb has. All these unchecked and uninterpretable features remain, and consequently (5b) violates FI. (26)

Feature-checking requires some constituent part of a syntactic object to move, in order to determine whether some relevant features match or mismatch. Syntactic movements, it is contended, are in accordance with certain economy principles of derivation. There are operations observing the Minimal Link Condition (MLC), according to which 'shorter'(27) syntactic movements are preferred, and thus the 'shortest move' in a particular derivation is selected. (6a) and (6b) illustrate this condition.

(6a) Who do you think will buy what?

(6b) *What do you think who will buy?

At some stage of a derivation, we have this syntactic object do you think who will buy what. The next step is the raising of a wh-word to the front position. Who, being closer to that position, is thus preferred to what.

Another principle is called Procrastinate. It prefers those derivations, which take place as late as possible - that means, the covert movements that occur after Spell-Out. Contrast (7a), which is well-formed, and (7b), which is not:

(7a) David sometimes paints the house

(7b) *David paints sometimes the house

(7b) involves an improper move of the verb paints to check off its tense feature without waiting until after Spell-Out.

Derivations, it is assumed, are also constrained by Suicidal Greed. The intuition is that N(oun)P(hrase)-movement should not be done for the 'benefit' of the other, such as checking the features of another category, but only if a lexical item has a 'selfish' reason to move (for instance, its Case-feature needs to be deleted), does it move. (8a), (8b) and (8c) are ungrammatical:

(8a) *John seems [t is leaving]

(8b) *John seems to t [that Bill is leaving]

(8c) *John was said to t [that Bill is leaving](28)

The small letter 't' represents the original position of John, or the 'trace' of John, which moves to the front. These examples are similar. Let us consider (8a) only. At one point, [John is leaving] obtains, where John is located in a Case position, and so John's Case-feature has already been checked. There is no need for John to move to a higher Case position. Its being attracted, wrongly, to the higher Case position is due to the offending inflectional features of seems, which need to be checked off. The fact that (8a) does not have a correct form is explained by the presence of such an 'altruistic' operation.

In the foregoing, I provided an outlook of C_HL, which is a structure-building and feature-checking mechanism. In particular, a representational constraint, FI, and some principles or conditions of derivation were illustrated. Their postulation is an outcome of the working hypothesis of the on-going 'minimalist program' in Chomskyan linguistics that C_HL should operate optimally.

1.2.2 An intensional concept of language

The internal mechanism with which Chomsky's linguistics is concerned is supposed to enumerate (generate), not only (1) a particular range of well-formed sentences infinite in number, but (2) the structural descriptions (SDs) of them. He distinguishes between (1) and (2) by calling them weak generation and strong generation, respectively. (29) It is (2), or strong generation, that characterise his intensional conception of language. 'Intensional' is a technical term that he uses. By 'natural language is a function specified in intension', he means that what is generated is the SDs of well-formed sentences, rather than an infinite set of well-formed sentences. (30)

An objection is that there is no empirical means for linguists to attain such an intensional grammar. It is not possible to ask a person or invent any experiment in order to ascertain the grammar that really guides him/her to construct sentences in certain structures, for his/her knowledge of such a grammar is allegedly tacit. Even we ourselves cannot give definite answers to questions concerning the constituent structures of the sentences that we speak every day. Using experiments, at most, we can distinguish different linguistic dispositions, which are merely true of indefinitely many extensionally equivalent systems. To achieve a true description of natural language, it is contended, what we need and can achieve is no more than an extensionally equivalent theory. (31)

But in Chomsky's view, such an idea is incompatible with our motivations to distinguish the following kinds of facts:

(i) When Jones is dropped, he falls

(ii) After Jones has eaten the sleeping pill, he falls into sleep

(iii) Jones understands the sentence Everybody loves somebody as 'There is someone who is loved by everybody' (see 1.1.3)

(iv) Jones acquires a body of knowledge, in virtue of which he can form and interpret the meaning of English sentences, on the basis of a certain linguistic data obtained from his environment.

(Actually Chomsky does not mention (ii) in the place where he discusses the issue.(32) (ii) is included here to further distinguish between physical and neurological facts.) To explain (i) and (ii), we attribute Jones mass and a brain, and appeal to some mechanical and neurophysiological laws, respectively. But in order to explain (iii) (language use) and (iv) (knowledge of language), further structures need to be imputed to the individual's mind (or brain) and further abstraction is certainly required. In Chomsky's linguistics, the steady state (S_L) and initial state (S₀) of the language faculty are attributed to Jones in order to account for (iii) and (iv) respectively. Postulating an extensionally equivalent system of laws may well be sufficient to deal with (i) and (ii), but not (iii) and (iv). For in the case of (iii) and (iv), Jones's being guided by internal rules, not merely Jones's behaviour, is involved. Theories about the mental structures involved in (iii) and (iv), admittedly, cannot be straightforwardly tested or be allowed to rest on behavioural evidence alone. Linguistics, as other scientific practices, should not be biased and restrictedly depend on behavioural criteria and evidence, according to Chomsky. Data from other languages, findings in child language research, discoveries in neurosciences and so on are recognised as 'empirical evidence' by scientists in general. Chomsky maintains that whatever conforms to the standards of rational thinking may contribute to the advancement of theoretical linguistics.(33)

For instance, studies in agrammatics have given substantial support to a postulate in Chomsky's linguistics, namely, 'trace'. Many referential phenomena and phenomena concerning the acceptability of grammatical structure are currently explained in terms of syntactic movements and trace, such as (8a) above. Now take the case of wanna-contraction. The (wh-)sentence (9) is ambiguous. It may induce two different answers as shown in (9a) and (9b).

(9) Who do you want to visit?
(9a) I want to visit Jane.
(9b) I want Jane to visit.

If the person who makes the question intends an answer like (9a), he/she is allowed to use the colloquial wanna, like (9c):

(9c) Who do you wanna visit?

But if an answer like (9b) is intended, the words want and to should not be replaced by wanna. The reason is that such a contraction is blocked by a trace, which is left behind by the movement of the word who to the front, as contrasted by (9d) and (9e):

(9d) you want to visit who → Who₁ do you wanna visit t₁
(9e) you want who to visit → Who₁ do you want t₁ to visit

It has been found that agrammatics have difficulties particularly in understanding the ambiguity of sentences like (9). (34) Note that the problem has nothing to do with the length or the number of arguments of the sentence. The explanation is that traces are deleted in agrammatism. (35) As confirmed by more evidence from agrammatic studies, the theory of 'traces' can be said to be psychologically real. It is not speculative, but empirically supported and predictive. In contrast, E-language theorists can only rely on a range of data, that is, an extension of ambiguous sentences like (9). Based on evidence of this kind, their theories have nothing to do with psychological reality and they are no more than generalisation, with much weaker explanatory and predictive power.

Comparisons between different languages, of course, can bring to light the topological characteristics of each particular language, of which most native speakers of this language are not aware. The data of one natural language may also modify the understanding of another one. Suppose that some newly found facts of a particular language undermine the theory of the initial state of C_HL, that is, UG. Then a revised UG needs to be put forward, to explain particular languages, which are instances of UG, and consequently, the idea on the purported character of a language will be updated. This nativist model of natural language and the relevant arguments involved will be dealt with in the next two sections.

1.3 Principles and parameters

Every human being shares the initial state (S₀) of the language faculty, which consists inter alia in an innate schema of universal principles and parameters (or options) constraining the operation of C_HL. Let us first consider the motivation behind this principles-and-parameters (P-P) approach in this section.

1.3.1 The tension between descriptive and explanatory adequacy

There is always the limitation that the more restrictive a general linguistic theory (UG) is (so that it maps a unique grammar on the basis of certain data about the use of a particular language), the more unpromising such a theory will be (to be able to describe other languages). But if the theory is not constraining enough, there comes the problem that the range of possible grammars will become too broad. If this were true of the initial state of the language faculty, in normal situations where the given data are meagre the child would not be able to learn a language as rapidly as what we usually see. (36)

Given a corpus of linguistic data, there will be a restricted number of grammars, which are capable of assigning structural descriptions to them. These grammars (or theories), which purport to describe the tacit knowledge (competence) of language of the native speakers, are said to be descriptively adequate, if indeed externally justified by their intuitions. (37) An explanatorily adequate theory is not only capable of describing the structure of the sentences of a particular language, but also of revealing the reality about the language faculty and language acquisition.

1.3.2 The P-P approach: resolving the tension

In the current framework of Chomsky's linguistics, C_HL is assumed to work optimally. There being no redundant moves is not due to a restriction imposed by a preferred system of rules.(38) The notion of evaluative procedure has been abandoned. Non-redundancy in the sense of either derivation or representation is now regarded as a designed property of C_HL. This optimal character is not contingent upon any implicit application of an innate language-specific idea of simplicity in the process of language acquisition on a set of possible grammars that are equally compatible with a given set of empirical data. A particular grammar is just an instantiation of the innate UG with parameter options chosen - and parameters are currently believed to be limited to morphological or lexical features. (39) According to this P-P conception, what is learned is restricted to the lexicon; all specific rules are subsumed under some innate principles tacitly known by the speaker/listener. Therefore, the tension between the twin goals of linguistics is better resolved by the P-P approach: accommodating at the same time the widest range of particular grammars and the fact that language is rapidly acquired. (40)

1.4 Language variation

The various ways of parameter-setting accounts for language variation. For instance, a native speaker of English and a native speaker of Chinese are in fact biologically endowed with the same innate schema; the mature state (S_L) of their language faculty is an instantiation of this schema, with the values of the parameters fixed in a different way. Following Otto Jespersen, Chomsky thinks that UG is restricted to syntax, and that morphology accounts for the variety of natural languages. (41) The part of the innately given schema which provides parameters or options is, it is thought, restricted to the lexicon. The differences of particular languages, accordingly, lies in their idiosyncratic morphological features, in particular, their inflectional morphemes or functional elements like Tense (T) and Case. (42) The morphology of a language, and its consequences for word order, are explained in terms of overt or covert feature-checking.(43)

For instance, English has a strong EPP-feature of T (that is, a feature that a clause must have a subject), which induces the pre-Spell-Out (overt) movement of the subject. This results in the subject-first property of the English-type (SVO) languages. The V(erb)-features of Japanese, which is a typical example of verb-final (SOV) languages, are supposed to be weak and according to Procrastinate to be checked covertly. Celtic languages such as Welsh, which is a VSO language, is supposed to have strong V-features, for instances:

(10a) Gwelodd�� y�� dyn�� y�� ci
Saw� �� the�� man�� the�� dog
(10b) Rhoddodd y dyn y ffon i'r ci
Gave the man the stick to the dog(44)

The notion of strong features may also explain the difference of sentence structures regarding the relative positions of the main verb (in particular, of finite sentences) and the adverbial(s) adjoined to the VP. French and English vary in this respect, among others, though they are both subject-initial languages. Compare the following pair of sentences (11a) (that is, (7a)) and (11b):

(11a) David sometimes paints the house
(*David quelquefois peint la maison)
(11b) David peint quelquefois la maison
(*David paints sometimes the house)

It is explained that unlike English, the V-features of T in French is strong and thus there must be overt V-raising from within the VP after the overt subject (nominal)-raising attracted by the strong EPP-features of T before Spell-Out. In addition, similar to English, the D(eterminer)-feature of T is weak in French; consequently its determiner phrases (DPs) do not raise out of the VP overtly. The outcome is that French sentences have the following ordering: subject-main verb-adverbial-object.

1.5 Language growth

The P-P approach has also implication for language acquisition, as shown in 1.5.1. Chomsky's central argument for it is to be discussed in 1.5.2.

1.5.1 From S₀ to S_L

According to Chomsky, I-language is the mature state (S_L) of an individual's language faculty. Experiencing only a meagre amount of data, a child can acquire a rich information system, which enables it to produce and understand even those sentences it never came across before. This linguistic issue is regarded by Chomsky as a version of "Plato's Problem" as presented in the Meno via a thought experiment to show how geometric principles are acquired. While Plato explains the geometric knowledge of the untutored slave boy in terms of a cognitive state of the boy's "pre-existence", Chomsky explains the knowledge of language by appeal to the language learner's innately given physical condition, that is, his biological endowment. Chomsky is not satisfied with the term 'language learning' as it is usually imbued with the behaviourist connotation that learning a language is basically a matter of training or 'moulding' by external factors. To him nature is of far more significance than nurture. Experiences merely trigger the biological development of the innate language faculty. So the word 'growth' is, Chomsky suggests, more appropriate than the word 'learning'. (45)

Let us begin with the 'primary linguistic data' (PLD) relevant to language acquisition, that is, the experience of a particular language to which a child is exposed, especially in its pre-school days. One should not confuse PLD with the data available to the linguists. The PLD include mainly the limited number of sentences or pseudo-sentences that a child hears from its parents or caretakers, some being incomplete, ill-formed, slips of the tongue, false starts, etc. This set of meaningful, structured, sounds (and gestures) in daily life situations only play some triggering and shaping roles in the natural process in which the child attains its language competence. Without an innate mechanism, that is, the 'language faculty', the language learner can hardly extrapolate or induce from the impoverished and degenerate stimuli a language which is so rich and complex in terms of its form and meaning. (46) Moreover, what enables a child to learn how to speak is, in Chomsky's view, by no means a certain 'general(ized) learning mechanism', which includes processes such as trial and error, conditioning, association, induction, making hypotheses, confirmation, making analogy and abstraction, etc., exercising 'multi-purpose learning strategies' on the 'general intelligence' - though such kinds of mechanisms may also be innate. He firmly believes that the mental capacities responsible for our acquisition (and use) of a language must be distinguished from those for learning tasks or activities such as playing chess, physics and history.

Characteristic of the human species, the 'language faculty' consists of a unique, built-in, 'language acquisition device' (LAD), which does not vary across cultures. Almost every individual is genetically endowed with a LAD, a mechanism "which effects a transition from the initial state [S₀] of the language faculty to later states, mapping experience to state attained". (47) UG is concerned with the biologically given and unique condition of language acquisition and thus is the "essence of human language". (48) Language acquisition is understood as fixing these biologically-constrained options (parameter setting) on the basis of invariant, innate, principles and of an exposure to a linguistic environment.

1.5.2 Lack of negative evidence

Suppose a child grows up in an English-speaking community. Then the PLD that he/she receives enable him/her to 'turn on' the head-first parameter and to 'turn off' the pro-drop parameter. English, unlike Japanese, requires the verbs, prepositions, adjectives and nouns (that is, heads) to be put in front of their complements, (49) and unlike Italian, requires a sentence to have a subject obligatorily, except in casual speeches or diaries. (50) , (51) It is believed that the parameters are almost binary ('on' or 'off'), that is, with only two options of variation. This idea, that UG highly restricts the range of hypothetical grammars for testing in language acquisition (and yet is compatible with the properties of all human languages), is sustained by the fact that what is available for a language learner is overwhemingly positive evidence. Chomsky distinguishes three kinds of evidence that the learner might get: (i) positive evidence, (ii) direct negative evidence, and (iii) indirect negative evidence. By 'positive evidence' Chomsky means the correct expressions pertaining to a particular language, which indicates, in particular, word order (like the SVO pattern of English) and irregular verbs, etc. 'Direct negative evidence' refers to explicit remarks on the incorrectness of the structure or words employed in linguistic performance, on the occasions where an adult corrects his/her own or the others', especially children's, linguistic mistakes. As for 'indirect negative evidence', it is supposed to be there when, say, a linguistic principle or rule hypothesised by a child, is not instantiated in even very simple sentences that he/she hears. (52)

If UG is not so restrictive as the P-P approach assumes, then the child will much more likely overgenerate. And the 'shrinking' back from the false grammars that he/she hypothesises to the target language relies tremendously on either (ii) direct negative evidence (that is, corrections by adults) or (iii) indirect negative evidence (that is, the non-occurrence of certain ungrammatical structures). But (ii) is not prominent in ordinary language-learning setting. Whereas (iii) is incompatible with the relatively fast and easy character of first language learning, for this kind of evidence depends on an exposure of an exceedingly great amount of linguistic experience - otherwise, the child could not be sure whether the non-occurrence of a structure is ungrammatical.

The nativist conception of language that Chomsky advocates is strong, for it suggests that UG is highly restrictive, so restrictive that mere positive evidence, which depends only on observation of a few instances, can play a decisive role in attaining some aspects of a grammar. When the language learner is exposed to a limited number of head-first expressions, he/she can automatically set the relevant parameter, since it is innately determined that if a language is head-first, then it is not head-last - and there are no other alternatives. (53)

1.6 Natural language: abstract yet biologically determined

I have shown in the previous sections how Chomsky explains the various linguistic phenomena by reference to the language faculty of the human mind. It is claimed that every person inherits a human genome providing 'instructions', which specify the growth of the brain in such a way that capacities of acquiring, mastery and use of a language are formed. And UG and I-language should reflect a certain physical reality of a language learner or user. Regarding the idea that there should be some biological endowments determining the way linguistic information is learned, stored and accessed for use, there seems to be no problem. But can the possession of I-language be identified with some facts of the mind/brain?

Chomsky is accused of committing a conceptual confusion in taking the I-language to be an "abstract entity" and yet "some element of the mind of the person who knows the language". (54) The charge is that Chomsky conflates what is known with knowing. The former is the object of knowledge; the object itself is not necessarily mentally represented or something necessarily related to the cognitive system of the mind/brain, whereas the latter must be concerned with the mind/brain, being a state or process of it. In addition, the comment continues, when we say Jones knows a language, we do not mean that what he knows are some components of his mind or brain, but something of an abstract nature whose identity does not depend on any cognition about it. (55)

But what is this abstract entity? Is it a mathematical object of a Platonic kind, which is non-spatial and non-temporal, independent of the mind and the physical world? (56) Chomsky has made it clear that if there were such a language, there must be some fact about it; yet we cannot find any linguistic fact that is extra to those concerning the human mind. (57)

Furthermore, human language has its uniqueness. It is distinguished from other symbolic systems, thanks to its nonredundancy feature:

"for example, FI is not observed in standard notations for quantification theory that permit vacuous quantifiers in well-formed expressions, as in [i], which is assigned the same interpretation as [ii]:

[i] (Ax) (2+2=4)(for all x, 2+2=4)
[ii] 2+2=4

But FI is a property of natural language." (58) Another crucial feature of natural language is that it involves "displacement". Analysing a sentence's structure, we need to distinguish two representation levels, namely, LF and PF. A word's position in the thematic structure of a sentence is usually not matching its position in the phonological structure of the same sentence. Wh-questions in English are obvious examples. (59)

In saying that I-language is an abstract entity, Chomsky does not presuppose a Platonic World. It is abstract because it is not described in physical (or physiological) terms. A discipline may characterise or study the human brain at different abstract levels. Chomsky's linguistics of the computational-representational approach is an example; another example, with different degree of abstractness, is neural net theory. (60) Both are theories of the human mind. There should be no conceptual problems, it seems to Chomsky, in calling I-language an 'abstract entity' "beyond those familiar in discourse involving theoretical entities". (61) The mind is composed of 'abstract entities' in this sense, I-language being one of them.

According to Chomsky, the mental under naturalistic inquiry, like the chemical, electrical, optical and so on, is just a part of nature. Linguistic theories play an important part in guiding the neuroscientists' research on the material basis of language. Chomsky writes, "Just as nineteenth-century science provided essential guidelines for the physics of the subsequent period, so the study of mind should serve as a guide for the brain sciences of the future, exhibiting the properties and conditions that must be satisfied by the mechanisms of the brain, whatever they turn out to be." (62) Though so far little has been known concerning the link between theoretical linguistics and neurosciences, that is, concerning the physical realisation of a linguistic description or explanation, he believes, in the long run, these scientific researches can be integrated, in the same way as the unification of chemistry and physics. (63)

1.7 Concluding remarks

Chomsky highlights the biological constraints of natural language. Language acquisition is determined by a principles-and-parameters schema (UG), which is genetically endowed to human beings. This schema is thus also the basis of linguistic competence, which means tacit knowledge of (i) a limited number of economy principles of derivation and representation of C_HL and (ii) a lexicon. Linguistic experience involved in attaining S_L consists in to a large measure mastering (ii), and is in effect a process of parameter-setting. What is known (that is, I-language) in Chomsky's conception, is not an infinite number of well-formed sentences conceived as objects independent of the mind/brain, namely, E-language, but the internalised mechanism which derives optimal representations of LF and PF. FI, which requires that there not be any uninterpretable LF and PF representations, is in fact a legibility condition imposed by the other two systems in the mind/brain, C-I and A-P, with which C_HL interacts to produce meaning and sound.

Back to Chomsky for Philosophers

Last modified: Wed Mar 27 13:30:13 Eastern Standard Time 2002