Analogy, Levelling, Markedness: Principles of Change in Phonology and Morphology (review)

Abstract

Reviewed by: Analogy, levelling, markedness: Principles of change in phonology and morphology ed. by Aditi Lahiri Joseph C. Salmons Analogy, levelling, markedness: Principles of change in phonology and morphology. Ed. by Aditi Lahiri. (Trends in linguistics, studies and monographs 127.) Berlin: Mouton de Gruyter, 2000. Pp. viii, 385. ISBN 3110167506. €91.01. While this volume’s title and table of contents suggest a focus on theoretical matters surrounding three tangled knots along the seam between phonological and morphological change, what in fact unifies the papers is their data-rich character and their focus on Germanic data (save one paper on Romance and discussion of Bengali data in another). The contributors are leading scholars in the field, including Paul Kiparsky, to whom the book is appropriately dedicated, and the questions addressed are all important for Germanic historical linguistics, especially prosody. The editor’s introduction is not intended as a full survey of the literature on analogy, markedness, and leveling; rather, it explores the histories of those notions. Lahiri sees all of the papers as agreeing that ‘analogy is not random and that change is constrained by the whole grammatical system’ (11–12), leaving ample terrain to explore. Paul Kiparsky’s contribution, ‘Analogy as optimization: “Exceptions” to Sievers’ Law in Gothic’, builds closely on his earlier work (notably 1998), defining analogy as ‘grammar optimization’. He uses his constraint-based variant of lexical phonology/lexical morphology, wherein optimization can operate either on lexical representations or output (34). While the question of serial derivation in optimality theory (OT) remains contentious (like opacity and such in OT generally), an important, relatively the oryindependent aspect of this paper is the role of Stem-Form in the analysis: Kiparsky traces key Gothic changes (e.g. nom.sg *haris > harjis, gen.sg *riikiis > riikjis) to constraints on stem forms and prosodic structures (syllable, foot). Stem-Form captures the generalization that stems cannot end in a short vowel (thus, *V̆]Stem). This attention to prosodic (and segmental) shapes of various morphological structures turns out to be a major thread running throughout the volume. In ‘Analogical leveling of vowel length in West Germanic’, Elan Dresher revisits Middle English open syllable lengthening (OSL), countering compensatory lengthening accounts and marshaling a broad set of English and West Germanic data. He argues that OSL occurred across all relevant environments but was overridden by later trisyllabic laxing, followed thereafter by leveling in both directions. Ultimately, the complexity of resulting patterns becomes great enough [End Page 783] that ‘learners despair of a rule, and opt instead to choose a consistent vowel quantity on a word-by-word basis’ (59). Broad leveling, he argues, occurs only in response to such crises, so that, like Kiparsky on analogy, he does not treat leveling as a ‘grammatical principle’ or as something ‘easily available at any time’ (60). Whereas learners usually make decisions about assigning words to particular declensional classes, Middle English OSL becomes a larger matter of‘establishing a lexical representation for a stem’ (66). Aditi Lahiri’s ‘Hierarchical restructuring in the creation of verbal morphology in Bengali and Germanic: Evidence from phonology’ treats often-neglected phonological aspects of grammaticalization, arguing that along the cline of grammaticalization forms can become affixes while remaining morphologically complex, but with a reanalyzed (and hierarchical) morphological structure. Along these lines she develops an analysis of Bengali ‘to be’ to a progressive marker; I cannot speak to the details, but it appears plausible. The parallel case made for the Germanic weak preterit assumes that a derivational -j- suffix and a fully inflected form of ‘to do’ were similarly grammaticalized and reanalyzed. The critical junction (99) is when speakers understand a sequence like *dōmida ‘judged’ as [[Root + j + /coronal stop /] + [past infl. suffix]suffix]ω. Here, the coronal stop behaves like a class marker while the suffix marks tense (99–100). Views laid out along the way may leave questions in the minds of Germanists. For some, arguments that the presence ofOld Norse umlaut in heavy i-stem nouns versus its absence in light i-stems was straightforwardly phonological will prove unsettling; others will be troubled to read that Old High German...