LINGUIST List 7.1778

Sun Dec 15 1996

Disc: Analogy

Editor for this issue: T. Daniel Seely <>


  1. Daniel L. Everett, Re: 7.1765, Disc: Analogy

Message 1: Re: 7.1765, Disc: Analogy

Date: Sun, 15 Dec 1996 07:59:49 -0500 (EST)
From: Daniel L. Everett <>
Subject: Re: 7.1765, Disc: Analogy

Patrick Farrell write on analogy that:

> There seems to be general agreement that:
> 1. Analogy plays a role in language change.
> 2. There are constraints on analogizing that are important to an
> understanding of knowledge of language.

I am not sure if there is such a general consensus. My own posting is
agnostic on the first item. On the second item I do not know that there
are any constraints on analogy, I doubt if there are. There are general
constraints of grammar, some of which turn out to constrain analogies
that people might imagine.

> An extreme version of this stance would be that there are NO
> other constraints. This extreme stance seems untenable. Consider
> the following case discussed by Chomsky in his discussion (in the
> video "Human Language Series, Part II) "of what he considers to
> be the failure of analogy-based approaches to language
> acquisition :

I do not know why the hypothetical extreme view is raised. It plays no
role in previous discussion of any significance. Moreover, in the
quote given, Chomsky is criticizing approaches to language learning that
rely on analogy as the principle explanatory device. He is not ruling out
the possibility of analogy playing a relatively minor role. Chomsky
has written a considerable amount on these topics in which he carefully
lays out the case, works going back to the 50s. It isn't particularly
helpful to criticize remarks he makes assuming that literature without
addressing that literature.

> The boy paints the red barn : The boy paints the barn red ::
> The boy sees the red barn: The boy sees the barn red
> As my students were quick to point out to me after I showed them
> this video, the analogy is not drawn because an act of seeing
> cannot produce a change of state in an object of
> perception. Because of our beliefs about how the world is and our
> knowledge of the meanings of _see_ and the resultative
> construction, the analogy is UNREASONABLE. As far as I know,
> there is no independently motivated principle of grammar that
> precludes the analogy.

Our knowledge of the real world is mediated in language by grammar to a
large degree. Work on processing (see especially Ted Gibson's work in
numerous articles and a forthcoming MIT Press book) shows fairly clearly
that in the type of case raised here there are grammatical principles
ruling out the relevant structures. Whether these derive from analogy in
some 'ultimate' sense is irrelevant. The fact of the matter is that
analogy will not explain the range of processing cases nearly as well as
a combination of constraints on grammar and short-term memory.

If that is so, then I think one must
> recognize that one kind of constraint on analogy is:
> 3. Avoid conceptual anomaly.

Look, quite frankly I am happy that you and your students had a nice
discussion about analogy. But there is a lot of work to answer before you
propose ideas as serious counterproposals. I do not think this
'avoidance' principle has been established at all.

> More generally, the universal grammar stance, according to which
> formal principles of linguistic structure constrain analogy,
> could be said to be interesting if one could point to
> semantically/conceptually REAONABLE analogies that are not drawn
> AND corresponding explanatory principles of linguistic form that
> are independently motivated, technically viable, and non-vacuous.

This is what the literature on processing has dealt with for years,
especially grammatically-based models.

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue