LINGUIST List 9.844

Sat Jun 6 1998

Disc: Hypotaxis

Editor for this issue: Martin Jacobsen <>


  • Marilyn Silva, Disc: Hypotaxis

    Message 1: Disc: Hypotaxis

    Date: Tue, 2 Jun 1998 12:25:28 -0700 (PDT)
    From: Marilyn Silva <>
    Subject: Disc: Hypotaxis

    In the ongoing discussion on hypotaxis, Benji Wald introduces some interesting material on the emergence of hypotaxis in child language by referencing the 1976 debate between Piaget and Chomsky:

    > The issue from which the discussion of 'hypotaxis' arises here > emerges from Sampson's attempted criticism of one of Chomsky's > 'innateness' arguments, in this case the innateness of certain > processes in complex syntax. I remember that in his 1976 debate with > Piaget, Chomsky used the sentence: > > is the man who is tall sad? > > as a crucial argument against a Piagetian or 'generalised > intelligence' account of learning complex syntax. He insisted that > on the basis of such examples as: > the man is tall -- is the man tall? > the "simplest" hypothesis that a child could come up with to acquire > question formation of > the man who is tall is sad > > would be: move the *first* verb to the beginning of the sentence, > hence, > ****is the man who tall is sad? > > He crucially claimed that no child ever goes through such a stage in > learning English, and therefore a child is 'innately' programmed to > somehow recognise the proper analysis of the inversion underlying > question formation (that its domain involves the 'highest' = largest > NP, not the first NP).

    Indeed, in my studies of subordination in child language development, I have never found such errors made by children. But that alone does not mean that children are making the kinds of analysis Chomsky contends. As Wald points out a bit later in his posting, children have not been observed to produce relative clauses within the subject NP. That is, relative clauses, when they occur, typically occur at the ends of sentences, with the result that early subordination is actually more of a paratactic than hypotactic strategy. Thus, my daughter produced her first relative clause at about 19 mos.: "That a man that wear a hat." I should note that as was typical for her and for others I have observed, studied, or read about, she did not produce another such construction for many more months. You will also notice that the copula is missing from her utterance, a fact that could have been predicted from Roger Brown's (1973) study of the acquisition of grammatical morphemes. Copulas are omitted until rather late in acquisition, suggesting again that Chomsky's example is rather worthless as evidence one way or the other because English-speaking children don't produce them, at least at first.

    Wald continues his points with the following:

    >(L2 learners also do this.) I omit the interplay of inflection with >inversion as less relevant, though, of course, it is only because >they have acquired inflection that we can see the pleonasm when there >is a 'real' verb (cf. 'did he went there?'). So, I wonder, do we get >intermediateexamples like: > > *is* the man who is tall *is* he (is) sad? > *did* the man who is tall *did* he cry (+ed)?

    Though young children typically do not produce the copula and therefore do not prepose it, I did find that my daughter was able to "prepose" the *do* auxiliary, but lest my statement lead you astray into thinking that the relationship between her behavior and movement was straightforward, let me state that she actually had two verbs *do*--the auxiliary that was always sentence initial and the main verb that was not. How do I know? She had two pronunciations: The sentence-initial aux when produced in present tense, third person singular, was always given its ordinary pronunciation [dz], which happens to be one of three irregular 3rd sing present tense verbs in the language. However, if *do* was used medially, which generally means as a lexical verb (the evidence is that *don't*, which can of course occur medially, is contracted and not recognized as two morphemes, but as a variant of *not*), the 3rd sing present was pronounced [duwz]. These data suggest a strategy in which interrogative Y/N sentence types are largely formulaic, and not derived from declaratives via movement. That is, because she encountered *does* as sentence-intial in interrogatives, my daughter learned the standard pronunciation for the word in initial position, analgous perhaps to learning that wh- question words occur initially. However, the lexical verb *do* was treated as an ordinary verb, undergoing regular suffixation (though in the adult language it is irregular).

    So, in general, I'm suspicious of arguments about children preposing the verb of the matrix clause and not the verb of the embedded clause. The strategies that children use, it seems, bear little resemblance to the hypotheses of formal linguistic theory.

    > So, I'm asking those who study these things closely: How do these > facets of complex syntax develop? And does that have a bearing on > C's arguments? Did he pull a fast one on the non-linguists in his > debate with Piaget (by implying that these things emerge immediately > in their mature state rather than being built up from the > acquisition of smaller pieces of grammar, something that might be > more compatible with Piaget's graduality in "constructivism", > leaving aside his specific proposals for stages in "generalised > intelligence")? Or, does Chomsky's argument hold up no matter what > the facts are? (As long as the facts are that they NEVER produce > "is the man who tall is sad?" or "is the man who tall cried?") I > oppose these considerations to both Chomsky's a priori innatism and > Sampson's a priori empiricism.

    I wouldn't go so far as to say "pull a fast one," but perhaps a "facile" one. Not having studied language development, Chomsky was in no position to make statements of the sort he did. However, given the thrust of the theory at the time, Chomsky's rationalist approach allowed no other explanations. The data he used were clear enough: children don't make the sorts of errors that he said they didn't make. But the question is: Is that enough to validate a position?

    I suspect you know my answer to that question.

    Marilyn Silva California State University, Hayward