25.4836, Review: Lang Acquisition; Syntax: Culicover (2013)

Mon Dec 1 21:26:32 UTC 2014

LINGUIST List: Vol-25-4836. Mon Dec 01 2014. ISSN: 1069 - 4875.

Subject: 25.4836, Review: Lang Acquisition; Syntax: Culicover (2013)

Moderators: Damir Cavar, Indiana U <damir at linguistlist.org>
            Malgorzata E. Cavar, Indiana U <gosia at linguistlist.org>

Reviews: reviews at linguistlist.org
Anthony Aristar <aristar at linguistlist.org>
Helen Aristar-Dry <hdry at linguistlist.org>
Sara Couture, Indiana U <sara at linguistlist.org>

Homepage: http://linguistlist.org

Do you want to donate to LINGUIST without spending an extra penny? Bookmark
the Amazon link for your country below; then use it whenever you buy from
Amazon!

USA: http://www.amazon.com/?_encoding=UTF8&tag=linguistlist-20
Britain: http://www.amazon.co.uk/?_encoding=UTF8&tag=linguistlist-21
Germany: http://www.amazon.de/?_encoding=UTF8&tag=linguistlistd-21
Japan: http://www.amazon.co.jp/?_encoding=UTF8&tag=linguistlist-22
Canada: http://www.amazon.ca/?_encoding=UTF8&tag=linguistlistc-20
France: http://www.amazon.fr/?_encoding=UTF8&tag=linguistlistf-21

For more information on the LINGUIST Amazon store please visit our
FAQ at http://linguistlist.org/amazon-faq.cfm.

Editor for this issue: Sara  Couture <sara at linguistlist.org>
================================================================

Date: Mon, 01 Dec 2014 16:25:52
From: Geoffrey Sampson [sampson at cantab.net]
Subject: Explaining Syntax

E-mail this message to a friend:
http://linguistlist.org/issues/emailmessage/verification.cfm?iss=25-4836.html&submissionid=33688145&topicid=9&msgnumber=1

Discuss this message: 
http://linguistlist.org/pubs/reviews/get-review.cfm?subid=33688145

Book announced at http://linguistlist.org/issues/24/24-3985.html

AUTHOR: Peter W. Culicover
TITLE: Explaining Syntax
SUBTITLE: Representations, Structures, and Computation
PUBLISHER: Oxford University Press
YEAR: 2013

REVIEWER: Geoffrey Sampson, University of South Africa

Review's Editor: Helen Aristar-Dry

SUMMARY

This book is a collection of thirteen of the author's journal articles,
originally published at dates from 1972 to 2006; some are single-author
papers, but most are co-authored.  Except for the first, the chapters are
equipped for republication with a paragraph or two of ''remarks'' relating the
individual chapters to the book as a whole, and (particularly with older
papers where Culicover would now treat some phenomenon differently) there are
occasional new footnotes.  Otherwise, the texts are unaltered from the
original publications, for instance a chapter refers to itself as ''this
paper'' rather than ''this chapter''.

Chapter 1, ''The simpler syntax hypothesis'' (co-authored with Ray Jackendoff
and first published in 2006) is presented here as a ''Prologue'' to the book
as a whole.  The other twelve chapters are grouped under the headings
''Representations'', ''Structures'', and ''Computation''.  The first of these
headings accounts for half of the book, and Culicover glosses it as ''various
aspects of what the proper domain of syntax is''.  ''Structures'' is about
''how to properly characterize the syntax of a language''.  ''Computation'' is
a short section (63 out of 380 pages in the book as a whole), glossed by
Culicover as ''reasons why some syntactic possibilities might be more likely
to be encountered than others''.  (Footnote 8 on p. 95 makes clear that for
Culicover ''computational work'' does not mean research using computers, but
rather work which brings considerations about how speakers and hearers
mentally process utterances to bear on linguistic theories.)

Since he presents it as a ''prologue'', it is reasonable to take Chapter 1 as
encapsulating the main gist of what Culicover is aiming to achieve with the
book.  The point of Chapter 1 is to argue that linguistics ought not to treat
complex sentences as having structures which include hidden elements that
function to make the logic of the sentences explicit.  Rather, linguists
should postulate syntactic structures that are as close as possible to the
word-sequences actually spoken or written (''simpler syntax'').  A concrete
example would be ''Ozzie tried not to drink'', where we understand that Ozzie
was the ''drinker'' as well as the ''trier'', although at the surface only the
verb ''tried'' has ''Ozzie'' as subject.  Culicover (actually Culicover and
Jackendoff, but it would be tedious to mention co-authors here whenever a
chapter that has one is quoted) says that ''mainstream generative grammar''
responds to those facts by postulating a deep structure or logical form in
which the object of ''tried'' is a clause [PRO not to drink], where PRO is
''an unpronounced pronoun whose antecedent is Ozzie''.  This is said to be
required by an axiom, identified as ''classical Fregean compositionality'',
said to be shared by many approaches to language, including much of formal
logic.  Culicover, on the other hand, prefers to say that the only syntactic
structure for the sentence is one in which no PRO element or other subject for
''drink'' appears; our understanding that the sentence is about Ozzie not
drinking arises from our semantic interpretation of that structure -- it is
not encoded within the structure itself.

In this example, the covert structure rejected by Culicover is a single
element, but in other cases it can be highly ramified.  For the example ''Joe
has put those raw potatoes in the pot'', Culicover's preferred structure has
fourteen nonterminal nodes, whereas the ''mainstream'' or Minimalist Programme
analysis has 33 nonterminals and several unpronounced terminal nodes; and
whereas Culicover's nonterminal labels are all rather traditional grammatical
concepts, e.g. S, NP, Aux, VP, Det, the mainstream labels are heavily
theory-laden, e.g. TP, T', DP[nom], Spec, v.  In Chapter 2 Culicover quotes
Merchant (2001) as a prominent representative of the approach which he
opposes.

''Simpler syntax'' is one theme running through most chapters of this book,
but there is also a second theme linking several of the later chapters which
is quite separate.  This is the idea that language universals can sometimes be
explained in terms of a pressure to use structures which facilitate speakers'
and/or hearers' mental processing tasks.  That idea is not original with
Culicover -- it is perhaps most prominently associated with John Hawkins, who
is cited several times here -- but Culicover examines it in novel ways, and it
would underplay the book to suggest that it is exclusively about the topic
introduced in the Prologue.

EVALUATION

To this reviewer, the idea that surface grammar is ''all the grammar there
is'' is very welcome.  The abstruse, highly-ramified tree structures
postulated by mainstream generative grammarians strain credulity and seem
underdetermined by any observable realities.

Nevertheless, I have problems with Culicover's concept of semantic
interpretation doing what ''simpler syntax'' does not.  Repeatedly (e.g. pp.
16, 26, 40, 72) Culicover assumes that the meaning of a sentence is something
which is to be encoded as a ''semantic representation'' -- that is, another
data structure of some kind, different from the syntactic structure of the
sentence.  That has been treated as axiomatic by many linguists from Katz and
Fodor (1963) onwards, but it is very questionable.  Philosophers of logic, who
have been thinking about the formalization of meaning longer than linguists,
usually conceive of the meaning of a sentence not as a particular data
structure but as a set of relationships with other sentences.  The reason why
''Ozzie tried not to drink'' is about Ozzie drinking is that it implies, for
instance, ''If Ozzie drank, his attempt failed'', where ''Ozzie'' is overtly
the subject of ''drank''.  Formalizing these relationships would be about
defining formal rules of inference specifying when one sentence follows from
another or others, not about defining ''representations'' of the meanings of
individual sentences.

(The reference to ''Fregean compositionality'' might seem to lend the
''semantic representation'' concept some philosophical respectability.  But I
am not aware that Gottlob Frege discussed such a concept, and Jeffry Pelletier
(2001) says that Frege never stated and perhaps did not believe in the
compositionality principle to which Culicover appeals.)

Culicover says (pp. 2-3) ''The fact that Ozzie is understood as the 'drinker'
results from a principle of semantic interpretation that assigns 'Ozzie' this
extra role.  Thus, semantics can have more elaborate structure than the syntax
that expresses it.''  But if syntax is made simpler merely by moving
structural complexities out of syntactic structure into another structure
called a ''semantic representation'', it is not clear what has been gained.

One can also ask how simple ''simpler syntax'' really is.  Chapter 8 argues
that English complement clauses have the complementizer structure:

            [CP Spec C [PolP Spec Pol [IP ... ] ] ]

(where the dots stand for the substance of the clause).  ''IP'' is Inflection
Phrase, roughly speaking a simple clause, and the kind of theory-laden concept
which the Prologue chapter seemed to reject.  This looks like the sort of
complicated covert structure which ''simpler syntax'' holds to be unnecessary.

If there is a contradiction here, that could be because, over the 30+ years
represented by these papers, Culicover like other generative linguists has
heavily revised his conception of how grammar works.  His new footnotes
sometimes say things like (p. 169, added to a transformational analysis first
published in 1982) ''The use of transformational rules to generate tag
questions is a device rooted in the earliest period of generative grammar ...
a more contemporary approach would take a constructional perspective ...''. 
The reader wonders how much of the chapter would remain valid if recast into
the grammatical approach Culicover now favours.  Presumably replacing a
transformational with a constructional perspective could not simply leave the
substantial content of the chapter unaltered, otherwise the two perspectives
would appear to be mere notational variants of one another.  But it would be
taking a lot on trust to assume that, though significant revision of the
content would be needed, if it were carried out the overall message of the
chapter would be sure to survive unscathed.

Culicover might have made his book more valuable if, rather than reprinting
the papers unaltered, he had reworked the contents so as to fit them together
into a coherent book-length exposition, all conforming to a shared framework
of grammatical assumptions which he was currently willing to defend.  The new
''remarks'' and footnotes are a gesture in this direction but fall far short
of what is ideally needed.  Indeed, Culicover does not even correct obvious
mistakes in the originals.  On p. 35, for instance, a series of example
sentences contains an alternation between the phrases ''rob a bar'' and ''rob
a bank''.  The reader's immediate reaction is to take this difference as
significant (small wording differences like this, e.g. ''got the present'' v.
''got the president'' on p. 132, are often crucial to Culicover's arguments),
but on further examination it seems that the bar/bank case is just a typing
error which has been preserved from the original publication in 1972 through
into the new millennium.

The low point of the book is Chapter 12, where Culicover claims to explain
Greenbergian language universals, such as the correlation of verb-first word
order with prepositions rather than postpositions, via a computer simulation. 
The simulation is displayed graphically as a large square comprising many
small squares representing individual speakers; initially the large square is
speckled randomly in eight colours, representing idiolects characterized by
each logically possible combination of three binary properties.  After 150
steps the display has evolved to contain just two domains each of a single
colour:  many logical possibilities, comparable to ''verb-first but
postpositions'', have been eliminated.  But the experiment is lamentably
misdescribed.  Culicover says that the population contains 2500 (50 x 50)
individuals, whereas it is easy to see that it contains 30 x 30 = 900
individuals.  Much more important, we are told nothing whatever about what
happens from step to step in the simulation.  We see that a random arrangement
of eight colours evolves into clumps of two colours, but we have no idea what
drives this in the simulation or whether that is a plausible model of
something that affects languages in reality.  This chapter is effectively
empty.

(In any case, I should have thought that a simple alternative explanation for
the verb-first/prepositions finding, not considered by Culicover, might be
that prepositions have typically developed out of verbs through
grammaticalization.  I do not know whether there is evidence for such an
origin in the Indo-European family, but in some languages it is obvious.)

Chapter 12 is not typical, though.  In general, I found that Culicover's book
provided a valuable window into the thinking of a linguist who is hard-nosed
enough to steer an intellectual course of his own rather than just accepting
the current theoretical consensus, but who nevertheless sees the generative
approach as a suitable overall framework for exploring grammar.  The book is
worth reading.  A book which reworked the original papers rather than just
reprinting them could have been more worthwhile still.

REFERENCES

Katz, J.J. and J.A. Fodor.  1963.  The structure of a semantic theory. 
Language 39.170-210.  Reprinted in J.A. Fodor and J.J. Katz, eds., 1964, The
Structure of Language: readings in the philosophy of language.  Englewood
Cliffs, N.J.: Prentice-Hall, pp. 479--518.

Merchant, J.  2001.  The Syntax of Silence.  Oxford: Oxford University Press.

Pelletier, F.J.  2001.  Did Frege believe Frege's Principle?  Journal of
Logic, Language, and Information 10.87--114.

ABOUT THE REVIEWER

----------------------------------------------------------
LINGUIST List: Vol-25-4836	
----------------------------------------------------------