Minimalism, UG and Second Language Acquisition by Vivian Cook

Minimalism, vocabulary and L2 learning	Vivian Cook SLA Topics Writing Topics
(paper at AILA 1996) NB text rather rough and graphics peculiar!

The talk starts by reminding people of the principles and parameters framework and then introduces some of the Minimalist Program ideas concerning the central importance of vocabulary in language acquisition
1. the lexical parameterisation hypothesis which claims that parameters are part of the lexicon
2. the functional parameterisation hypothesis that they are attached to functional phrases, which have their own entries in the lexicon: the lexicon is thus an extended system with entries of two types
3. inflection-driven. Grammatical inflections are added in the lexicon and ‘checked’ in the syntax.
While the details of the Minimalist Programme are exceptionally hard to grasp and its implications are still far from clear, the emphasis on vocabulary is a core aspect. Consequences are that any previous UG work not taking these two hypotheses on board is now impossible to justify and any current work on Universal Grammar theory should have moved to this group!

L2 learning is thus acquisition of L2 lexical entries with their associated parameter settings. A lexical entry, following Levelt, terms has four sides. The MP in effect extends the syntax side beyond argument structures and redefines the lexicon to include inflectional as well as derivational morphology. An L2 user acquires the word X with lemma meaning components and XX grammatical specifications and s/he acquires the heads and settings for functional phrases. This may not of course bring the L2 study of vocabulary in line with UG in anything other than name since little of it is concerned with any of Levelt’s four aspects of lexical entries. We do not for example learn binding as such but lexical entries for reflexives and pronominals.

Most of the sub-field of UG in second language acquisition has concerned the research question of whether such and such a syntactic technique works for L2 users, now to be updated by the MP proposals. The obsessive issue of access to UG can at least be rephrased: do L2 users have access to the same possibility of creating their own lexicon with its own parameter-settings as L1 learners? There is no reason why it should work as a solid block but entry by entry: perhaps the reason for confusing answers to access is attempts to generalise for the whole lexicon rather than individual entries.

Or we could explore the idea that the two lexicons are in fact one. In other areas there is plenty of evidence that the two lexicons are at least tightly connected if not one large store. Learning an L2 means not creating a new lexicon as a separate thing but adding items to the same store, whether lexical or functional, alongside the existing ones. The reason for the frequent result that L2 learners are percentage points off natives in experiments with a variety of constructions such as subjacency may be that the L1 and L2 functional phrases with the parameter-settings are alongside each other in the lexicon, metaphorically speaking, rather than in two distinct whole grammars; the reason for ease of code-switching may be again that there is no switch from grammar to grammar but from phrase to phrasal entry to phrasal entry in the lexicon.

The main point of this paper is to bring to people’s attention the current importance of the lexicon within the UG theory and to see how it revises our ideas about language acquisition in general. The intention then is to be informative, with a bit of speculation at the end.

The current version of Universal Grammar known as the Minimalist Program is another radical departure, which no-one claims to understand fully. It is about as easy to grasp and as uncontentious as the peace process in Ireland.

The simplest thing is to start from two quotations by Chomsky, seen on the OHP. Originally these come from his papers in progress, now available in his 1995 collection the Minimalist Programme. ’there is only one human language, apart from the lexicon, and language acquisition is in essence a matter of determining lexical idiosyncrasies’ (Chomsky, 1989). This is then the reason why I am contributing this here rather than to other symposia. Language acquisition within the Minimalist Programme is now seen as a matter of learning vocabulary. We all speak the same language; we differ in our vocabularies. Hence everything that is learnt is part of the vocabulary; vocabulary acquisition is all. In itself this may great news for those involved with lexical acquisition, who turn out to the champions of the universe. All we need to do is study how people acquire vocabulary and that’s it.

However there is of course a snag: the word ‘vocabulary’ doesn’t mean quite what it would appear to. Vocabulary acquisition is rather different from anything presupposed up to now. Let us take a second famous quotation: ‘acquisition of a language reduces to selection of substantives from a given store and fixing of values of parameters that apply to functional elements and to properties of the lexicon as a whole’ (Chomsky, 1990). What on earth does this mean so far as vocabulary is concerned?

Let us then look at some of the concepts underlying this quotation to show what it means, at least what I think it means. On the OHP I have spelled out three underlying points that are involved

Since 1981 or so the theory has attempted to deal with variation between languages in terms of a choice of settings for a parameter. Thus the familiar pro-drop parameter decides whether a language may have subjects missing from the actual sentence, like Italian, or must have them, like English. A more interesting example is the parameter associated with the binding principles. Again the familiar analysis suggests that a reflexive pronoun such as himself may only be bound within the clause, roughly speaking; in John said Peter likes himself the reflexive pronoun himself has to go with the noun Peter not with John. In Japanese, however, the equivalent reflexive pronoun jibun may also be bound to John within the whole sentence rather than the limited domain of the clause. There is then a parameter setting difference between English and Japanese. Wexler and Manzini suggested that this is not a general property of the language itself, whether English or Japanese, but of the actual words used, whether the English word himself, or the Japanese jibun. An individual pronoun has a parameter that needs setting to show how it may be bound. The parameter is not part of the syntax but makes up part of the lexical entry for that particular word. This is then the lexical parameterisation hypothesis; the parameters that account for much of language variation go with particular lexical entries. Parameters are in the lexicon, not the syntax.

However, this still leaves the bulk of parameters in the syntax. The next stage then involves the functional phrases that have come to dominate syntactic analysis since 1986—inflection phrases, agreement phrases and the like. Basically the idea is that each grammatical element has a phrase of its own; there’s agreement in sentences so there has to be an Agreement Phrase; there are Determiners so you need a Determiner Phrase. At one time linguists were coining new type of phrase as fast as physicists were finding new particles except that the equipment they needed wasn’t quite as expensive. The functional parameterisation hypothesis suggested parameters went with functional phrases, not lexical items.

But this would mean they were in syntax not in the lexicon. The solution is then to combine the idea of functional parameters with the lexical parameterisation hypothesis and to say that functional phrases are actually stored in the lexicon. As well as the actual content words of a language the lexicon also stores the functional phrases with their parameter settings. Learning the lexicon means learning the heads and parameter settings for each functional phrase. The principles of syntax are not learnt; only the functional phrases and everything that goes with them.

The consequence of this is then that inflections such as present tense ‘-s’ in English are now part of the lexicon. They are attached to words in the lexicon and then they are put in the the syntax where they undergo a process called checking to see if they fit. The mechanisms of this are to me totally obscure but obviously if the inflections are shifted off to the lexicon something like this is necessary to ensure that they are used in the right syntactic structures.

The consequences of the Minimalist Programme for Universal Grammar in second language acquisition are many. Many in this field see second language acquisition research as a way of testing outt the latest syntactic idea; last Tuesday’s theory is no use today. For them the change to Minimalism means at worst that virtually all the published research on Universal Grammar is now beside the point, at best that it will have to be painstakingly reanalysed. For those of us who see Universal Grammar as contributing to second language acquisition research rather than the other way round, the Minimalist Programme provides another set of ideas to try out to see what use we can make of them, without necessarily abandoning all our past efforts.

In general then we can see the Minimalist Programme as having specific effects on second language acquisition research of vocabulary, having redefined vocabulary in this broader way. In some ways this is not so novel. Let us take the standard description of the lexical entry from Levelt, he sees it as having two aspects the lemma with its description of meaning and syntax, the morphophonological with its description of morphology and phonology. Mostly second language research with vocabulary has concerned aspects of the meaning part of the lemma, not its syntactic properties or its morphophonological, rather surprising in neglecting the three main areas of interest to linguists. The lemma aspects already include then some syntactic information, in particular the argument structure of the entry, whether for example a verb such as give takes two objects. This already seems to me a vital point for L2 study of vocabulary. Very little of the area has looked at grammatical information in the lexical entry. Words as seen as having lexical meanings rather than grammatical relationships. Knowing the word XXX allows one to put it into sentences etc. This has already been part of Universal Grammar since the beginning of principles and parameters theory instituted the Projection Principle that tied syntax closely in to the structure of the sentence. It is not however something that my knowledge has featured prominently in the L2 study of vocabulary which has been much closer to the psychological tradition.

Secondly there is the nature of lexical knowledge assumed by the Minimalist Programme. We now have parameters in the lexicon. Acquiring an L2 lexical entry is setting parameters as well as acquiring meanings. In particular it is acquiring parameters for functional phrases. When someone learns that English has an inflection ‘-s’ it means they also set the parameters for agreement in English.

One of the issues has been whether there is access to Universal Grammar in second language acquisition. While the question itself is badly phrased, as I have suggested elsewhere, its meaning has now changed. It could be paraphrased as: do L2 learners build up lexical entries for their grammatical knowledge as L1 learners do? Sub-questions would be whether the access to functional phrases is at first denied to them, as seems the case in L1 acquisition and whether the content of their lexical entries for both functional and lexical categories is the same as those exposed to only one language. I wouldn’t know what existing research if any could be interpreted as answering these questions. But it clearly suggests that we should try out the same techniques on the learning of functional phrases as have been applied to lexical items. It also seems to me to reflect an ‘itemisation’ of grammatical acquisition; learners are acquiring items with properties, not rules or generalisations—those can be left to the unchanging principles of syntax that no learner has to acquire. It’s almost like the old days of grammatical teaching syllabus; here are the items to be learnt and the sequence in which learning takes place. True the items bear no resemblance to those once used

However the main point that I want to develop in the time remaining is the way that the knowledge of the L2 is stored in the mind. One problem has always been is whether UG spins off a clone of itself every time a new language is acquired. That is to say first the mind acquires a first language complete with the principles and parameter settings; then some process produces another distinct grammar with its own processes and principles; ans so on for as many languages as the person knows. It’s coordinate ??? bilingualism on and on for ever with each grammar in a separate box. Partly this seems to me ridiculous; these are all grammars in the same mind; partly it ignores all the known relationships and connections between the languages stored in the mind. But certainly this is the model implied by much UG work. each clone grammar has a new set of parameters, the only question being whether they are resettable.

This cloning mechanism sounds bizarre; no wonder UG doesn’t work in an L2, as many say. It also doesn’t explain what seems to be a central finding from research, namely that L2 learners on the one hand do not acquire the L2 grammar as efficiently as the L1 learner, on the other have a knowledge that is far better than chance; in other words they seem to have a parameter-setting partially. My own view for some time has been that there must be a single overall language system in the mind; one grammar with one set of principles and one set of parameters but multiple possible settings for the same parameter. Thus a person who knows another language may not have quite the same setting in the L2 as a monolingual — why should they?— nor will they have quite the same setting in the L1 as a monolingual —again why should they? but will have a setting that veers from one to the other according to which language they are using. I used to call this the speedometer theory: parameter settings go up and down on a dial in some form of strength etc. However mostly UG people were quite supercilious when I tried this on them; psychologists on the other hand seemed to treat it as obvious and wondered why i wasn’t looking for connectionist explanations instead of UG.

But if we take seriously the Minimalist Programme suggestion to put parameters etc into the lexicon we seem to have a way out as far as I can see. In the L2 there has been long debate about whether there is a lexicon for every language with some kind of neutral cognitive entry or whether there is a single lexicon; my own belief, which fits a reasonable amount of the literature is that there is indeed a single lexicon where the entries for different languages are closely related to each other in one store. EG???

Suppose we now introduce the Minimalist Programme suggestions about vocabulary; we now have one store in which information about phrases is stored as well as about lexical items. There are no longer two separate grammars. There is one overall knowledge of principles of syntax and there is an enormous store including functional phrases with their parameter. The functional phrase for language A is alongside that for language B; the bleeding between the values for the first language and the second is because of their close proximity in storage, just as it is for the interference and overlap between lexical items of two languages. parameter = semantic component

Chomsky, N. (1990). Language and Mind. In D.H. Mellor (Ed.), Ways of Communicating. Cambridge: Cambridge University Press. 56-80