Variations and Selections:

"An Anthology of Reviews from the Journal of the Experimental Analysis of Behavior

Language is not identical to verbal behavior, and a special contribution of behavior analysis lies in differentiating between them. Bertrand Russell's 1926 review of The Meaning of Meaningby Ogden & Richards provides some of the rationale for focusing on functional verbal behavior instead of on language. In his introduction, Wood, who proposed its reprinting, identifies ironies in the relation between Russell's work and behavioral analyses of language. Encapsulated in Russell's essay is a list of "truisms about words" that fit hand-in-glove with the behavior analytic approach: e.g., "The natural function of words is to have effects upon hearers." Russell was not a behavior analyst, for he also viewed words as means of conveying information, thereby suggesting more traditional views, but his approach to "meaning" and his willingness to challenge conventional assumptions set the stage for later developments.

A landmark later development was Skinner's Verbal Behavior. MacCorquodale begins by describing what Skinner's book is and is not, characterizing it as an interpretive extension from known principles rather than an exposition with supporting data. Skinner omitted relevant data for clarity of exposition (Skinner, 1957, Preface); a research program predicated upon the book took roughly a quarter of a century to take off, but now it supports a journal, The Analysis of Verbal Behavior. MacCorquodale outlines relevant behavioral principles, along with the major taxonomies or functional categories of verbal behavior and the grounds for distinguishing among them (cf. Catania, 1992, chapter 11). He also identifies some potential misunderstandings. Key among these is the role of multiple causation in accounting for complex phenomena in terms of relatively simple processes, because many critics mistakenly assume that a behavioral account posits one unique, proximal environmental cause for each instance of behavior.

Contemporary political struggles have shown that immediate and forceful replies are needed to counter false negative advertising. With 20-20 hindsight, it is clear that behavior analysts seriously erred in not replying in that manner to Chomsky's (1959) vitriolic review of Verbal Behavior. As a result, the review is more widely known than its target, and its assertions are accepted by many who know virtually nothing about behavioral accounts of verbal behavior. MacCorquodale's second essay shows that an effective reply was possible, unmasking distortions and inaccuracies in Chomsky's brilliantly effective rhetorical moves. Despite its seductive embrace of conventional concepts and assumptions, the bald inaccuracy of Chomsky’s review could be its ultimate vulnerability (cf. Andresen, 1990).

Lenneberg's nativist Biological Foundations of Language,reviewed by Bem & Bem, supports Chomsky's approach. Lenneberg treats the organism as a set of biologically determined mechanisms that define competence for manipulating structured word sequences. The reviewers discuss how the contradictory agendas of the two approaches override any potential complementarities, asserting that the plausibility of one necessarily precludes the plausibility of the other. Bem & Bem are persuaded by the "new linguistic observations" that concern mainly the structure rather than the function of language, and succinctly characterize the manner in which nativistic linguists and philosophers have challenged empirically oriented theorists such as behavior analysts.

The target of Segal's review, Roger Brown’s A First Language: The Early Stages,is another book taken by the community of linguists as undermining a behavioral approach to language, because Brown found little evidence of parent-mediated differential reinforcement in his longitudinal study of language acquisition in three children. Nevertheless, Segal shows how the categories of Brown's analysis are translatable into Skinnerian terms without appreciable loss of detail: "intentional contours," which affect whether the listener reacts to utterances as declarative, interrogative or imperative, are taken as elementary autoclitic relations; semantic complexity translates as the difficulty of environmental discriminations that control an utterance; and so on. To support the validity of her behavioral translations, Segal discusses the role of conditional discriminations in the behavioral account of word order. The importance of this discussion is evident in the failures of psycholinguists to understand it. For example, Fodor (1981) portrayed the radical behaviorist approach as inadequate precisely because he saw it as unable to handle conditional discriminations, even though behavior analysts had been studying such relations systematically for at least two decades at the time of his writing.

Brown's book was a primary basis for assertions about the hopeless inadequacy of behavioral accounts of language in the face of the speed and ease with which children were said by linguists to master their native languages. As asserted by Bem & Bem and others, the behavioral failing was a failing in principle, not merely one of technical resources. Ernst Moerk, however, (who characterizes himself as a skill-learning theorist rather than a behaviorist) exhaustively reanalyzed Brown's data, presenting the results in First Language: Taught and Learned,reviewed by Salzinger. Moerk identifies in extensive detail instances of parents' prompting and then reinforcing imitative vocalizations, shaping by successive approximations, and other massive evidence of behavioral process in Brown's data. Moerk's analysis shows that claims about the ease of first language learning grossly underestimate the time required for a normal child's language learning.

While fundamental differences separate linguistic/cognitivist and behavior analytic viewpoints and keep them in tension, attempts have been made to identify their complementarities. Segal's review is one example; Hayes' review of Ericsson & Simon's Protocol Analysisis another. Hayes summarizes Ericsson & Simon's information-processing model as it relates to verbal reports of ongoing nonverbal performance, and identifies ways in which Ericsson & Simon and behavior analysts similarly distinguish among verbal categories and have similar reservations about what can be taken at face value. The explicitness of Ericsson & Simon's assumptions about verbal reports as indexing events in memory and their acknowledgment that verbal reports can affect performance permit fairly detailed comparisons between their analyses and the behavior analytic distinction between rule-governed and contingency-shaped behavior. On the other hand, the necessary role of awareness in their account distances it from a behavior analytic one (though learning without awareness has become widely accepted among cognitivists; e.g., Kihlstrom, 1987). The ensuing discussion of specific methods, measures and their implications shows how data viewed as highly relevant to cognitivist formulations can be addressed by contemporary developments and strategies in behavioral theory. Rapprochement on methods and questions can follow from constructive recognition of differing assumptions and philosophies.