Spanish phonology and morphology: Experimental and quantitative perspectives. By David Eddington. Amsterdam: John Benjamins, 2004. Pp. xv, 197. ISBN 1588116123. $126 (Hb.)Reviewed by JOSÉ IGNACIO HUALDE, University of Illinois at Urbana-ChampaignThis book is, on the one hand, an exposition and demonstration of a particular approach to the study of language and, on the other, a review of experimental work, by Eddington and others, on different aspects of Spanish phonology and morphology. The emphasis is on morphophonological alternations. The phonetic end of phonology, and even phonemic analysis, are somewhat underrepresented.For E, as for Chomsky, the goal of linguistic analysis is to produce psychologically relevant accounts. The similarity, however, ends here, as E argues for an experimental, quantitative approach with little room for formalism. In Ch.1, E gives four convincing reasons for not accepting claims of psychological reality at face value. Some of these points have been made before by other authors, who are appropriately quoted. One of his objections is that far too many analysesare based on either the uncontrasted intuitions of the authors themselves or perhaps on theopinions of one or two speakers regarding a handful of examples, without any guarantee of replicability. It seems to me that the validity of this point ought to be obvious to all linguists, regardless of theoretical persuasion. We should all agree with E that the practiceof making pronouncements regarding the (un)grammaticality of linguistic structures of doubtful status, without providing any evidence for them, can no longer be accepted. Naturally, the more controversial the claim the more empirical support should be required.This book includes insightful discussion of a great number of experimental studies. In some cases it is not clear that the experiment could distinguish among competing hypotheses. This is the case with several experiments that address the productivity and psychological reality of morphophonological rules. For instance, as discussed in Ch. 3, in Spanish we find alternations of the type illustrated by desdén ‘disdain’, desdenes ‘disdains’ vs desdeñar ‘to disdain’, desdeñoso ‘disdainful’. Furthermore, a phonotactic constraint disallows final -ñ. The two facts are, of course, related. In such families of words, Harris (1983) assumes underlying stem-final -ñ, whichundergoes a rule of depalatalization when final in certain domains. Pensado (1997) tested the reality of this depalatalization rule by providing her subjects with nonce words like the noun sirapén and the related infinitive sirapeñar and asking them to produce the plural noun (expected: sirapenes) and related adjective (expected: sirapeñoso). For the adjective, some subjects produced the expected forms and some did not, and there was a bias toward basing the answer on the last example that was presented (singular noun or infinitive). In regard to this study, E concludes that “[t]he large degree of inconsistency intheir answers, coupled with the fact that many answers were based on the phonological shape of the last nonce word presented to them suggests that depalatalization is not a synchronically active phenomenon” (52). It is not clear that this conclusion follows. A proponent of a rule-based analysis may counter that the behavior of some subjects did show an active application of the rule, whereas those speakers who did not follow the rulemay have treated the relevant words as exceptions. It seems to me that an experiment ofthis type cannot possibly distinguish rule-based from analogy-based accounts. The experiment may show to what extent speakers are willing to extend a morphophonological alternation present in their lexicon to novel pairs of words, but it cannot say anything about the underlying representation of words. For that other techniques would need to be used. A main goal of Ch. 3 is to show that in many cases experimental work has dispelled unfounded notions which were initially proposed with little solid evidence. Thispoint is well taken and is essential to justify E’s experimental approach. Some of the examples, however, are perhaps not the best choices. This is the case, I believe, with the discussion of secondary stress in Spanish. Briefly, although several Spanish phonologists have postulated the presence of noncontrastive secondary stress on certain syllables, acoustic study by phoneticians has generally failed to find evidence for secondary stress. What phonologists have typically failed to state clearly is that the syllables they mark asbeing secondarily-stressed are only potentially stressable. If stressed, they will carry a pitch accent and may show greater duration. Secondary stress, when present, is generally audible. I submit that almost any random sample of Spanish news broadcasting (a style where secondary stresses are particularly frequent) will demonstrate that secondary stresses are almost always placed on one of the syllables that phonologists have marked as being secondarily-stressable. What Spanish phoneticians have failed to do in their attempts to measure the correlates of secondary stress is precisely to elicit speech containing secondary stresses. The case of “Vowel opening in the wake of /s/ deletion” (41-44) is somewhat different. The problem here is that some vowel alternation facts that were originally described for Eastern Andalusian Spanish (the opening of mid vowels and fronting of the low vowel in the context where a word-final /s/ has been deleted), were assumed by several authors to exist also in other Spanish dialects with extensive deletion of /s/, such as Caribbean Spanish, without any real evidence. As E points out, subsequent instrumental work has confirmed the existence of the phenomenon only for Eastern Andalusian, and has found no trace of it in Caribbean Spanish. An error was made by those authors who, making a leap, assumed that Latin American dialects with /s/ deletion would also present this vowel alternation. Spanish dialectologists have known for a long time that the phenomenon in not found in Western Andalusian (Alonso et al. 1950, Alvar 1996). Since Latin American Spanish is much more closely related to Western than to Eastern Andalusian there was never any serious reason to expect that the vowel alternation would be found in Latin American dialects, but, as E points out, this is exactlywhat some authors did. E