Language genes

[fluffyhamster]

There's nothing really that surprising about work being divvied up around the brain - I mean, a neuron or two somewhere has got to fire, hasn't it? And just because a very specific part enables us to recognize faces doesn't necessarily imply that there is a "language module" (but I'll "concede" though that certain parts of the brain are indeed involved with "language" - re. psycholinguistics, studies of aphasics etc). The thing with some nativists, however, is that they actually seem to believe that words themselves are innate, even though the words would seem to be very culture-specific (I recall Dabrowska - do a search here on Dave's for her - mentioning in her book someone going on about the innateness of 'cricket'. I'll check the specific reference if anyone wants me to).

Now, that may be merely the extreme loony fringe of the nativist camp speaking, but what comes from the more central members sometimes isn't much better and can often just seem (to me at least) like so much hand-waving. Yes, we'd all like to know exactly how the brain works, and to what extent a disposition or ability in language is (always has been or has become) genetically inheritable, but I doubt if it'll be Chomsky who'll be providing us with any substantial or convincing answers (that's not to say that the questions and proposals he has always been dreaming up haven't been interesting or inspirational (if flawed or impossibly ambitious)).

I became aware of the controversy surrounding Sampson a year or two ago, around the time I'd bought and was reading Educating Eve. I think it's unfortunate that he felt he had to comment on "racial" research, because it has distracted rather from his more worthwhile pursuits. But what's actually so inconsistent about 1) claiming that there are individual differences (if not racial averages) in intelligence and 2) saying that language is predominantly a cultural (collective) "intellectual/cognitive" achievement? Are the two points that related?

[Stephen Jones]

It is completely inccnsistent of Sampson to accept that intelligence is mainly genetically determined but that language skills are entirely controlled by nature and genetic determinism has nothing to do with it.

This is particularly foolish when there are more differences in the development of intelligence between individuals than there are in language skills.

The development of language has all the hallmarks of genetically determined behaviour. It follows a pattern of development that is consistent across the species, and requires a stimulus to be switched on but only a sufficient stimulus; increasing the stimulus does not proportionally (or possibly not even marginally) increase the ability. As Chomsly has said, his theory has near-universal acceptance from geneticists, so it is naive of psycho-linguists to accuse him of soft science.

There is also a hierarchy in language traits - it is not for nothing that Baker has written a book called 'The Atoms of Language', in which he claims we are approaching the possibility of having a periodic table of the elements of language, as we have one of the elements of the physical world.

To claim that language is a totally cultural phenomenum is to claim it is totally different from other aspects of human and animal development and behaviour. Our concepts of how genes interact with the environment is becoming clearer and clearer, and the theories of pure connectionsists that had a brief rebirth in the mid 90s are no longer taken seriously.

[fluffyhamster]

I read Baker's book a short while before Sampson's, and have to say that Sampson's was by far the more convincing and (therefore) interesting read of the two for me.

Larry Trask's review of the Baker:
http://human-nature.com/nibbs/02/trask.html

[Stephen Jones]

Trask works at the same department as Sampson, which may or may not have something to do with his attitude.

It seems to me he is setting up strawmen in his argument; I will try and check out more carefully.

[abufletcher]

It seems to me he is setting up strawmen in his argument

I'd have to agree with this. Still, Educating Eve is an interesting read just for balance in a field that badly needs balance.

[fluffyhamster]

Trask works at the same department as Sampson, which may or may not have something to do with his attitude.

Actually Trask died a few years ago. I forgot he'd probably have had some contact with Sampson!

There can be little doubt however that he was a very capable and respected linguist (probably moreso than Sampson), and more than conversant with Chomskyan grammar (see Trask's dictionary of grammatical terms), Historical Linguistics, the Basque language etc etc, so again, his views merit serious attention. I'm sure that whatever "attitude" he displays came as a result of working through all the competing approaches and realizing which captured the facts or worked best (for him, at least).

It seems to me he is setting up strawmen in his argument; I will try and check out more carefully.

It's been a while since I read the review - I should probably read it again.

[fluffyhamster]

It might be helpful in this debate (well, in the part of it raging on Dave's, at least) to make clear what exactly it is we mean when we use the term 'language' (as in 'Language is innate' - 'No it isn't' etc) - it's easy to start taking the term for granted and therefore possibly not address central issues.

As I said before, particular (developed) language and/or culture-specific lexis and morphosyntax is almost certainly not "innate" (though it may be arbitrary), but obviously language generally has to come from somewhere, be based on something (originally).

What could this basis (for 'language') be (i.e. have been)? Does the sort of taxonomy of mental operations proposed by Cognitive Linguistics (see e.g. Croft & Cruses's Cognitive Linguistics (CUP 2004) provide a comprehensive (at least a constructive, initial) framework for modelling the abilities of the (pre-?)human mind? Would such a range of cognitive processes perhaps give rise to a form of "mentalese"? Are e.g. Wierzbicka's 'semantic primes' (which form a 'natural semantic metalanguage') a sort of mentalese? Is such a 'language' the basis, basis enough, for the development of more complex language(s)?

The above is necessarily sketchy...no explicit mention of evolution, philosophy etc. I didn't even manage to slip in words, names or phrases like 'symbolic', 'transmission of culture', 'the ratchet effect', 'Deacon', 'Tomasello', 'Fodor' etc etc etc.

Right, I'm off to buy a second copy of Pinker's The Language Instinct' (unless Kinokuniya's sold out of it, hmm, is it likely?:o )...

[fluffyhamster]

Some reviews of the first edition of Sampson's book:
http://www.ling.ed.ac.uk/~jim/sampsonreview.html
http://www.linguistlist.org/issues/9/9-734.html

[fluffyhamster]

A review of Tomasello's The Cultural Origins of Human Cognition:
http://cogweb.ucla.edu/Abstracts/Hauser ... lo_00.html

[Anuradha Chepur]

For those in the Chomskyan camp, the eyes are always green on the other side of the fence ! And hey, Chomskyans never say words are innate, as fluffyhamster says somewhere in this thread.

There are strict rules which allow only certain sequences of words and dissallow the rest. Who wrote the rules? Nobody? If nobody did it, then nature did it. We are born with the rules and the template is innate. This is the long and short of Chomskyan linguistics.

[Stephen Jones]

There are strict rules which allow only certain sequences of words and dissallow the rest. Who wrote the rules? Nobody? If nobody did it, then nature did it. We are born with the rules and the template is innate. This is the long and short of Chomskyan linguistics.

No, it's not. It's a grotesque over-simplification.

[Anuradha Chepur]

No, it's not. It's a grotesque over-simplification.

O no! Why would an ardent Chomskyan distort him? Yes, it was a quick attempt at brevity for the sake of brevity, and also to avoid being accused of thrusting Chomsky on non-Chomskyans. But if I may know why exactly you disagree with me, I could try to clarify.

[fluffyhamster]

And hey, Chomskyans never say words are innate, as fluffyhamster says somewhere in this thread.

I'll have to dig up the relevant bit in my Dabrowska then (and perhaps in whichever other books, where similar points are made). But while we're on the subject of words, isn't a lot of grammar actually quite "wordy" (Wierbecka's NSM etc)? That is, maybe we have to "coin" and learn a lot more than Chomskyans would care to admit, and the seeming similarities between languages (UG etc) might just be to do with how we perceive and refer to the world around us. Sapir-whorf it ain't, and Chomsky it certainly ain't!

[Anuradha Chepur]

But while we're on the subject of words, isn't a lot of grammar actually quite "wordy" (Wierbecka's NSM etc)? That is, maybe we have to "coin" and learn a lot more than Chomskyans would care to admit,

The lexicon, where words are stored, is always given its due in generative grammar.

The following is a relevant section from a research paper by Eirikur Rognvaldsson:

"Since the beginning of generative grammar (Chomsky 1957), its main objects of study havebeen the basic elements of the lexicon, their structure and relations, on one hand, andsyntactic constructions, i.e., sentences, on the other. In Chomsky (1981:5), it is assumed that`the subcomponents of the rule system' are the following:

(1) (i) lexicon(ii) syntax(a) categorial component(b) transformational component(iii) PF-component(iv) LF-componentSystems (i) and (iia) constitute the base. Base rules generate D-structures (deepstructures) through insertion of lexical items into structures generated by (iia), inaccordance with their feature structure. These are mapped to S-structure by the ruleMove-α, leaving traces coindexed with their antecedents; this rule constitutes thetransformational component (iib), and may also appear in the PF- and LF-components.Thus the syntax generates S-structures which are assigned PF- and LF-representationsby components (iii) and (iv) of (1), respectively. (Chomsky 1981:5.)

For the last 20 years at least (since Chomsky 1970), most generative linguists have assumed that all irregularities and idiosyncracies are stored in the lexicon, whereas thesyntax is completely regular. The meaning of sentences is assumed to be compositional, i.e.derivable from the meaning of the individual words plus some interpretation assigned to thestructure. It has of course been realized that certain sentences - idioms and the like - do notin fact have compositional meaning; the individual words do not retain their `original'meaning. Some examples are given in (2):(2) a. keep tabs on someoneb. pay heed to somethingc. kick the bucketThus, we must assume that such phrases and/or sentences are learnt as a whole and stored inthe lexicon, in much the same manner as individual words. Such `exceptions' are relativelyfew, and it is usually assumed that they do not present a serious challenge to the maintheory.It is clear, of course, that some kind of cooccurrence restrictions will have to be built inthe lexicon. Chomsky (1965:101) illustrates this by means of the following sentence:


(3) He decided on the boat.

Chomsky points out that this sentence can either mean `he chose the boat' or `he made his decision while on the boat'. In the latter case, the phrase on the boat is a place adverbial, modifying the entire sentence, whereas in the former case, the connection between the verband the preposition is much closer. In these close constructions, the choice of Particle is often narrowed or even uniquelyconstrained by the choice of Verb (for example, "argue with X about Y"). We must therefore indicate in the lexical entry for such words as decide, argue, that they takecertain particles and not others, as, in fact, is commonly done in dictionaries. Thisinformation can be presented in various ways. (Chomsky 1965:191.)After the words have been inserted into the deep structure phrase-marker, we need somesort of a filter or checking mechanism to scan the structure and ensure that all coocurrencerestrictions or subcategorization features are respected.Another type of cooccurrence restrictions is exhibited by so-called selectionalrestrictions (Chomsky 1965), which are semantically based. We can argue that the fact thateat takes an animate subject, drink takes some kind of liquid as an object, etc., follows fromthe meaning of these verbs, and hence speakers do not have to learn this separately.Therefore, such selectional restrictions come at no extra cost; and if they are violated, thesentences are judged ungrammatical or just plain nonsense.Thus, the usual assumption has been that the `classical' generative model can accountfor all of these cooccurrence restrictions (idioms, subcategorization, and selectionalrestrictions).

The Minimalist Program 3.1 Lexical insertion in the Minimalist Program

In this section, I will argue that Chomsky's (1992) proposals in the Minimalist Program may make it much easier to account for such features. At present, there are very many aspects of this framework that seem to be open to debate. Among them is the nature of the lexicon and the processes that apply within the lexicon; For our purposes, the most important innovation in the Minimalist Program is that aseparate level of D-structure as an interface between the lexicon and the computational system, is no longer assumed. In earlier approaches, `all items that function at LF are drawnfrom the lexicon before computation proceeds, and presented in the X-bar format'(Chomsky 1992:27). Thus, instead of inserting words from the lexicon into slots in a ready-made tree structure, the tree-structure is built up gradually, as the words are inserted. (1992:30-31) describes this in the following manner:The computational system selects an item X from the lexicon and projects it to an X-barstructure of one of the forms (18), where X = X0= [XX]:(18) (i) X(ii)(iii)[X''[X'X]][...]"
--------------------------------------------------------------------------------

[Andrew Patterson]

Anuradha Chepur wrote:

For the last 20 years at least (since Chomsky 1970), most generative linguists have assumed that all irregularities and idiosyncracies are stored in the lexicon, whereas thesyntax is completely regular. The meaning of sentences is assumed to be compositional, i.e.derivable from the meaning of the individual words plus some interpretation assigned to thestructure. It has of course been realized that certain sentences - idioms and the like - do notin fact have compositional meaning; the individual words do not retain their `original'meaning. Some examples are given in (2):(2) a. keep tabs on someoneb. pay heed to somethingc. kick the bucketThus, we must assume that such phrases and/or sentences are learnt as a whole and stored inthe lexicon, in much the same manner as individual words. Such `exceptions' are relativelyfew, and it is usually assumed that they do not present a serious challenge to the maintheory.

This is tantamount to saying it's 100% true except for all the exceptions.

The number of such exceptions will presumably vary from one language to another. In English they are relatively high.

It sounds like an example of the "special pleading fallacy" to me.