Phonological acquisition and phonological theory Ed. by John Archibald (review)

REVIEWS Phonological acquisition and phonological theory. Ed. by John Archibald. Hillsdale , NJ: Lawrence Erlbaum, 1995. Pp. xxi, 204. Reviewed by Charles Reiss, Concordia University, Montréal* As stated in the preface, this volume is designed to address the fact that '[tjhere has been much less work on the acquisition of phonology' and phonological learnability than on the acquisition of syntax. The wide selection of topics discussed in the nine papers should appeal to phonologists whose interests range from metrical theory to feature geometry and tonal phonology. General questions of learnability and acquisition are also raised, and after absorbing A's useful introductory sketch of recent phonological theory, more syntactically-oriented researchers should be able to follow those aspects of the papers which intersect with their general learnability interests. The firstfourandthe sixthpapers are concernedprimarily with L1 acquisition. The fifthcompares the tasks confronting Ll and L2 learners. The last three study phonological disorders, L2 acquisition , and recognition offoreign accents, respectively. In Ch. 1, 'Global determinacy and learnability in phonology', B. Elan Dresher and Harry van der Hülst discuss the learnability ofgrammars in which 'the phonological representation is underdetermined by the local phonetic [i.e. surface phonological] properties of the relevant unit and can be established only by taking into account a widerrange ofdata' such as alternations. The authors discuss this phenomenon ofglobal determinacy in the context ofstructuralistphonemics, linear generative phonology, and autosegmental and underspecificationtheories ofvarious sorts. One oftheirconclusions, tacitlybased on the reasonable assumption that acquisition and learnability studies should guide general linguistic theory and not just attempt to make children conform to the musings of the theoreticians, is basically negative: 'modem multilinearphonological theories raise all the same questions regarding learnability as did older SPE-style models . . .Too often, new feature geometries oralternative applications ofexisting ones are proposed without serious consideration of the learnability issues that are implicitly raised . . . [W]e mustbe careful about advancing analyses that crucially appeal to global determinacy without specifying a learning path' (20). Ch. 2, 'Variability in a deterministic model of language acquisition: A theory of segmental elaboration', by Keren Rice and Peter Avery, is concerned with the acquisition of the feature geometric representations which constitute the phonological inventory ofa language. Their central hypotheses are (1) 'Minimality: Initially the child has minimal structure' and (2) 'Monotonicity: Inventories are built up in a monotonie fashion', by which they mean that 'addition of structure proceeds a single step at a time' ' (35). The chapter also provides useful discussion of several kinds of inter- and intraspeaker variability, thus addressing, at least, a problem which plagues so much of the phonological acquisition case study literature—the fact that children endowed with (the same) UG who are learning (basically) the same target phonology produce incredibly nonuniform output. This chapter follows the Jakobsonian view of ever-growing segment inventories and thus appears to be a straightforward application of the subset principle, though the authors do not refer to this principle. For a radically different interpretation of the subset principle as applied to phonological inventories, see Hale & Reiss 1997b for a view which is based on features rather than segments, according to which all feature distinctions must be present at the initial state of the grammar. In Ch. 3, 'Segments and syllables in early language acquisition', Jane Fee 'attempts] to integrate the melodic and prosodie components of current nonlinear phonological theory into a theory of phonological acquisition' (59). Some of the arguments, however, appear somewhat circular: 'we predict that children's word forms should be minimally bisyllabic . . . Forms that * Thanks to Mark Hale for comments and discussion. 1 This use of monotonie is slightly inconsistent with its use in mathematics, where it describes curves or sequences whose rate of change does not vary in sign—the rate is always nonnegative or nonpositive. 838 REVIEWS839 are simply lexically learned [i.e. precede the stage of 'phonological organization'] may be monosyllabic , because they do not have to adhere to the constraints of UG. Words that are constrained by the phonological system should, in contrast, be bisyllabic' (60). Reference to linguistic forms which violate UG appears contradictory to the present reviewer. Fee also argues that acquisition evidence supports the existence of the minimal word as a category of...