ICON programming for humanists By Alan D. Corre (review)

Abstract

648 LANGUAGE. VOLUME 67, NUMBER 3 (1991) This diversity is evident in the papers of the four plenary speakers. Roger W. Andersen ( 1 24 ) examines the use of verbal inflections, in context, to consider 'what makes it possible for native speakers of a language to convey complex expressive messages quickly and easily to accomplish their purposes in verbal interaction' (2). Rod Ellis's topic (25-60) is the use ofmetalinguistic judgments in second-language research ; he concludes that the reliability of grammaticality-judgment tasks in such research is 'circumspect'. Henning Wode (85-116) questions 'the prevailing assumption' that there is a 'discontinuity' between first-language acquisition and adult second-language acquisition, and proposes a 'continuity' view of language acquisition abilities. Perhaps the most useful paper for not only understanding second-language acquisition but also applying that knowledge to improve second -language teaching is T. Givón's (61-84). Givón's argument is that the natural order of both first- and second-language acquisition involves , first, acquisition of the lexicon (in a 'nongrammatical pidgin mode') and then acquisition of grammar (in a 'grammatical communication mode'). At the outset of language learning, neither word recognition nor grammar use is a habituated task, and both require attention . Only after vocabulary recognition becomes automatic can the learner be free to focus his/her attention on grammar. Givón proposes experiments to test his hypotheses about the 'cognitive basis for acquisition order', 'the efficiency of structured language learning', and 'vocabulary size and grammar-lexicon competition ', and he distinguishes his approach from 'communicative', 'immersion', and 'lexicalpidgin ' approaches. Those with special interests in second-language acquisition are likely to find among these 36 papers at least two or three that address their interests. All in all, these volumes provide, as Burmeister & Rounds claim, 'an up-to-date representation of the relevant issues and current state of affairs in L2 acquisition research' (iii). [Timothy J. Riney. International Christian University.] ICON programming for humanists. By Alan D. Corre. Englewood Cliffs, N.J.: Prentice-Hall, 1990. Pp. xiv, 157. Some linguists will remember SNOBOL, a language with unusual facilities for character manipulation and pattern matching in which considerable linguistic programming was done, especially in the 1960s and early 1970s. Aside from these special features, SNOBOL is a very primitive language, especially in its low-level control structures. ICON is a modern descendant of SNOBOL, syntactically not unlike C and Pascal, designed and implemented by a research group at the University of Arizona led by Ralph Griswold, one of the creators of SNOBOL . ICON has been implemented for a wide variety of machines and operating systems and is distributed electronically at no charge and by mail at cost. ICON'S easy string handling, pattern matching , and mapping facilities make it useful for searching for information in a corpus, translating one representation into another, and constructing word lists, reverse dictionaries, and concordances. These, combined with its automatic storage allocation and rich data structures (lists, queues, stacks, records, and tables), which make it easy to represent graphs (of which the trees and feature structures used in linguistics are special cases), make it very suitable for parsing and morphological analysis. An especially useful feature is the fact that arbitrary values, not just integers, may be used to index into tables. Thus, a word-counting program could increment the count of the word the by this statement: words ["the"] + : = 1 . Corré's book provides an introduction to ICON suitable for those with little or no background in programming. It presupposes very little , explaining such things as how to use the EDLIN and EMACS text editors and how to execute programs once they are written, as well as basic programming techniques. It is very clearly written, with a variety of simple examples . The reader who completes this book should be able to write simple ICON programs and will have a basic knowledge of ICON' s pattern matching facilities. Two factors render the book incomplete as a textbook of ICON for linguists. First, the applications illustrated are almost all concerned with the computation oftext statistics (e.g. computing the standard deviation of the lengths of sentences in a text, or correlating the frequencies of words...