Review of Perspectives on Sentence Processing

by Charles Clifton, Lyn Frazier, Keith Rayner
Book Reviewed
Book Title
Perspectives on Sentence Processing
Book Author
Charles Clifton, Lyn Frazier, Keith Rayner
Book Publisher
L. Erlbaum Associates
Place of Publication
Year
1994
ISBN
Book Review Citation
Review Author
Thomas E. Payne
Year
1996
Publication
International Journal of American Linguistics
Volume
62
Issue
4
Pages
432-433
Publisher
Language
English
License
Select License
URL
Updated
November 22nd, 2012
Abstract

432 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

PERSPECTIVESON SENTENCEPROCESSING.

Edited by Charles Clifton, Jr., Lyn Frazier, and Keith Rayner. Hillsdale, N.J.: Lawrence Erlbaum Associates, 1994. Pp. xiv + 483.

Perspectives on Sentence Processing is a collection of seventeen articles repre- senting work presented at the Sixth Annual CUNY Sentence Processing Conference in 1993. The CUNY conference represents the "mainstream" in the emerging, inter- disciplinary field of Sentence Processing, and many of these articles are already be- ing cited in ongoing research. This book promises to be a major and continuing influence on research into how people understand and produce language.

The term "sentence processing" describes research into the logical and biological processes involved in parsing, interpreting, and constructing sentences in natural language. In other words, while linguistic theory aims to characterize and delimit the forms of natural language grammars, sentence-processing research investigates how such grammars are implemented.

The book is divided into six topically defined sections: (1) sentence processing and the brain, (2) phonological processing, (3) syntactic processing: information flow and decision making, (4) syntactic processing and computational models, (5) referential processing, and (6) sentence processing and language acquisition. A wide range of methodologies is employed, from traditional proof-texted examples with "arm-chair" grammaticality judgments, to instrumental measurements and psycho- linguistic experiments.

While cognitive psychologists have always relied heavily on linguistic data, in this particular thread of research, they have pretty much taken the generativist "package" off the shelf and uncritically used it to formulate hypotheses and claims. From a field linguist's point of view, the authors of the articles in this book could benefit from a stronger diachronic perspective, as well as a healthy skepticism of the basic assumptions of generativist linguistics, such as the sentence as the upper limit on linguistic inquiry, the competence/performance dichotomy, and the theo- retical centrality of grammaticality judgments.

The only article in the book that gives actual data from an American indigenous language is "The Finite Connectivity of Linguistic Structure" by Edward Stabler. This article proposes a "bounded connectivity hypothesis," which essentially states that processing complexity increases exponentially when elements of a configura- tion have more than one relation of any given type outside of that configuration. This proliferation of processing complexity explains why certain sentences may be generated by the grammar, yet be unacceptable as surface strings. The hypothesis is supported by limits on extractions and multiple morphological causatives. The fol- lowing Quechua examples of multiple causatives are provided (from Cochabamba, Bolivian Quechua):
(la) Riku-chi-chi-ni

See-CAUSE-CAUSE-~SG

'I have made someone show it'.
(1b) *Riku-chi-chi-chi-ni

('I have made someone make him show it'.)
REVIEWS

Single morphological causatives are common in Quechua. Double causatives such as in example (la) are rare, and triple causatives are unacceptable, though one grammar does provide one example that is rejected by a Quechua speaker con- sulted by Stabler. These facts cannot be due to a simple syntactic restriction on the number of arguments in a clause because both transitive verbs (such as riku 'see') and intransitive verbs (examples with waAu 'die' are given) show exactly the same syntactic behavior. Furthermore, these facts cannot be due to a semantic restric- tion since it is perfectly possible to conceive of a situation which would involve three causatives (the suggested gloss for lb), and because analytic causatives in- volving distinct causative verbs can, in fact, be used to express such concepts.

Unfortunately, Stabler ignores the competing hypothesis that some forms with one causative morpheme are lexicalized. In other words, it may be the case that some transparently complex forms are subject to the same syntactic rules as are simple lexical verbs. It is telling that the examples used by Stabler are of verbs that are often lexical causatives in the world's languages ('kill' < 'cause to die' and 'show' < 'cause to see'). It is probable, from the point of view of simple lexicalization, that rikuchi and wafiuchi are such useful and frequent complex expressions that they are highly amenable to being treated as simple lexical verbs for purposes of further morphological operations, such as causative. I venture to guess that not ALL verbs in Quechua allow multiple causatives as readily as these two. Therefore, an account based on syntactic connectivity, without reference to the semantics (which is the basis for a form's usefulness and frequency) of the verbs involved, will un- doubtedly overgeneralize.

It is encouraging to see that a few of the articles mention languages other than Japanese, English, and German, the usual sources of data for articles in the field of Sentence Processing. However, most of the non-Indo-European languages are men- tioned only in the article by Stabler. It is hoped that future work in Sentence Pro- cessing will continue to incorporate data from a wide range of languages and will develop a deeper familiarity with descriptive linguistics and theories beyond those of the generativist tradition.

THOMASE. PAYNE, University of Oregon

Set your country here to find out accurate prices
Country:

Seller Condition Item Price Shipping Total Cost
AmazonNewUS$ 18.94 US$ 3.99US$ 22.93
AlibrisVery Good VERY GOOD softcover, 15-20 pages have underlining/occasional marginalia, rest of text CLEAN.US$ 15.00 US$ 3.50US$ 18.50
AlibrisGood 0805815821 USED BOOK in good condition| No supplements| Normal wear to cover, edges, spine, corners, and pages | Writing / highlighting | Inventory stickers | Satisfaction guaranteed!US$ 45.75 US$ 3.50US$ 49.25
AlibrisUS$ 93.00 US$ 3.50US$ 96.50
AlibrisUS$ 93.05 US$ 3.50US$ 96.55
AlibrisUS$ 135.44 US$ 3.50US$ 138.94
AlibrisUS$ 135.44 US$ 3.50US$ 138.94
AlibrisUS$ 135.95 US$ 3.50US$ 139.45
AlibrisUS$ 158.95 US$ 3.50US$ 162.45
  • Recommend Us