LINGUIST List 7.721

Tue May 21 1996

Sum: Iearning vs. innateness of binding principle

Editor for this issue: T. Daniel Seely <dseelyemunix.emich.edu>

Directory

1. "Carsten Breul", Sum: Iearning vs. innateness of binding principle

On 23 April 1996 (Linguist 7-607) I posted the following
query (I repeat the text in its original form, but, please,
read 'Beneficiary' or 'Benefactive' instead of 'Benefactor'
and think of an example sentence like 'Jane hits
her/herself' instead of my 'Jane combs her/herself', which,
as S.E. Kemmer points out, is bad English):

> in Vivian Cook and Mark Newson's _Chomsky's Universal
> Grammar: An Introduction_ (2nd edn., 1996), we find the
> following argument in favour of the innateness of the
> binding principles:
>
> "Step A is to claim that the native speaker knows that in
>
> 4. Helen said that Jane voted for herself.
>
> Jane binds _herself_ [...].
>
> Step B is to see whether children could have worked this
> out from the speech they are likely to have encountered.
> Suppose children wrongly understand that _herself_ is bound
> by Helen [...]. Nothing would tell the children that they
> are wrong; no context could let them unerringly distinguish
> the binding of anaphors and of pronominals." (p.84f.)
>
> On the one hand, I am not totally convinced by this
> argument. I can well imagine that children learn by
> induction and generalization from a number of simple
> sentences in proper contexts - like, e.g.
>
> 'Jane combs her' versus 'Jane combs herself' -
>
> that _her_ cannot refer to the Actor of the action of which
> _her_ is the Patient, while _herself_ must refer to this
> Actor. If we apply this generalization to the example
> sentence given by Cook and Newson (see above) - allowing
> also Benefactors alongside Patients -, we still get the
> correct result: _herself_ must refer to the Actor of the
> action of which it is the Benefactor. The fact that the
> anaphor is part of an embedded clause and that the matrix
> clause also has a feminine subject NP does not interfere
> if the 'rule' is derived by induction and stated this way.
>
> On the other hand, I would be totally convinced by Cook
> and Newson's argument, if children never made the mistake
> of using pronouns instead of anaphors or anaphors instead
> of pronouns during language acquisition. This would
> indicate, I think, that they do not proceed by induction
> and generalization.
>
> Can anyone tell me what the relevant facts of language
> acquisition are as regards pronouns and anaphors?
>
> Comments on and discussion of my arguments in general and
> especially clarification if I did not get the point of
> Cook and Newson's argument, are also welcome.

A big Thank You goes to the following persons for their
replies:

Richard Ingham <llsingamreading.ac.uk>
Suzanne E Kemmer <kemmerruf.rice.edu>
Kazumi Matsuoka <MATSUOKAUConnVM.UConn.Edu>
Steven Schaufele <fcoswsprairienet.org>
Elisa Vazquez Iglesias <iaeviusc.es>

Steven Schaufele writes:

"For what it's worth, my 5 1/2-year-old daughter
frequently uses `anaphors' (i.e., reflexive pronouns) in
circumstances in which ordinary pronominals would be more
appropriate. For instance, i have heard her reporting
`Mom/Sue/Evan/Tory's mother brushed herself', when it was
clear from context (or my independent knowledge) that the
agent and the beneficiary of the action were not the same
person.

I have not made much of this, because i can imagine two
possible interpretations of this phenomenon that do not
involve explicit falsification of Chomskyan Binding Theory
claims.

(1) Margaret is not using the anaphorical `herself' but
the emphatic `herself'.

(2) Margaret `understands' the Binding Theory distinction
between the theoretical entities `pronominal' and
`anaphor', but does not know that `herself' is an anaphor.
 I must admit i'm not sure how i or anyone else could
teach her this, other than by overt correction: `No, she
didn't brush *herself*, she brushed *her*/*you*/etc.'"

Suzanne E Kemmer writes:

"Children do learn language by generalizing over
exemplars. The claims that "Most of language is innate" (as
pronounced for example in the PBS video on Language aired
in the US in the last year or so) depend on redefining
"language" as some tiny slice of knowledge called 'core
grammar'. The principles of UG have nothing at all to say
about the vast array of expressions and constructions that
every speaker of a language must learn (and specific
expressions are just specific instances of constructions,
which occur at all levels of generality). The principles of
UG also have little to say about the real universals of
language, i.e. general cross-linguistic patterns found for
particular construction types (e.g. the constraints on
kinds of possessive constructions --relations of meaning
and form-- found in the languages of the world, as
identified for example by Bernd Heine and other language
typologists).

That's why cognitive linguists, who see no great gulf
between the most general grammatical constructions (which
UG theorists concentrate on), lower-level, more
idiosyncratic grammatical patterns, and specific instances
of those patterns, prefer to have one theory covering all
degrees of generalization in linguistic patterning; and
one which can also apply to even more general
cross-linguistic patterns (which often are motivated by
cognitive factors OUTSIDE language--as Greenberg began to
show back in the 60s). The role of meaning in cognitive
linguistics is integral--both for acquisition and for
adult knowledge of grammatical patterns. Most of the
problems that UG theory is set up to solve are only
problems if we think that meaning plays no role in language
acquisition and use.

Michael Barlow has written a paper called "Corpora for
Theory and Practise" which is about how corpora reveal
patterns that data from pure intuition can never
reach--including frequency information that affects the
generalizations speakers draw (again, both as children and
as adults). It has a case study on reflexive marking in
English, which shows what the REAL patterns of reflexive
marking are that people are exposed to--which is quite
different from the kinds of examples UG linguists think up.
He presents a schema theory to describe the actual usage
data found with reflexives. (Recall that UG isn't about
usage anyway--language is somehow completely divorced from
what happens when speakers use and understand language).
Michael's paper will appear in the journal Corpus
Linguistics--but I'm sure he'd be glad to send you a copy:
barlowruf.rice.edu .

By the way, your example with 'comb' is not a good one, for
one thing because English doesn't use a reflexive marker
with that verb --we say 'She combs her hair'. A better
example would be 'hurt' which is probably the first
reflexive children hear in English. (English doesn't much
use the reflexive marker for 'middle verbs' of personal
grooming like 'wash'--these aren't good examples of
reflexives because they designate inherently reflexive
actions, which many languages, including English, keep
formally separate from true reflexives, which are those
actions that one normally performs on another, and
exceptionally on the self)."

Richard Ingham writes:

"Your enquiry on LINGUIST raised some good points, I
think. Try:

Solan L. (1987): Parameter setting and the development of
pronouns and reflexives. In: T. Roeper & E. Williams
(eds.) _Parameter setting_. Dordrecht:Reidel.

This provides evidence that children do in fact
misconstrue pronominals as reflexives, e.g. giving:

1) the horse kicked him

a reflexive interpretation.

However, I'm not sure that this refutes the position
presented by Cook & Newson. For a while, child grammars
might allow reflexives and (ungrammatical) locally bound
pronominals in parallel distribution, at least in lexically
governed contexts.

The issue you raise is why English-speaking children
should overgeneralise, and allow long-distance binding,
when there is no evidence for it in input. I am pretty
sympathetic to your point of view, except that you would
need to show why children might allow local binding of
prominals when (presumably) there is no evidence in input
for a reflexive interpretation of things like 1) either."

Kazumi Matsuoka refers to seminal articles on the issue:
Matsuoka, Kazumi in: Proceedings of 19th Boston University
 Conference of Child Language Development (Somerville, MA:
 Cascadilla Press). [An experimental study, an updated
 version of which is being prepared.]
Chien & Wexler in: Language Acquisition 1 (1991).
Reinhard & Reuland in: Linguistic Inquiry 24 (1993).

Elisa Vazques Iglesias points out that she is currently
finishing her dissertation, which deals with anaphoric
binding in the history of English and which contains a
chapter on the innateness thesis of binding principles.

Carsten Breul
e-mail: carsten.breulrz.ruhr-uni-bochum.de