Search

This article compares and contrasts cophonology theory and indexed constraint theory, the dominant current proposals to morphologically conditioned phonology. In cophonology theory, morphologically conditioned phonology is captured by associating each morphological construction or lexical class with its own phonological grammar, or cophonology. All constraints within a given cophonology are purely phonological; no constraint directly refers to morphological context. By contrast, indexed constraint theory assumes a single fixed constraint ranking for the entire language, and captures morphologically conditioned phonology by indexing individual constraints to specific morphological contexts. The article raises three arguments in favor of cophonology theory: greater formal parsimony, the ability to handle free variation, and more accurate predictions about the scope of morphologically conditioned phonological effects. It also evaluates and rejects the primary argument for indexed constraint theory, i.e., Grammar Dependence, the claim that indexed constraint theory is more restrictive in the degree of language-internal diversity allowed. Cophonology theory and indexed constraint theory are equivalent in the range of language-internal diversity they allow; it is argued that the upper limit on language-internal diversity should not be a matter for formal grammar, but instead requires extra-grammatical explanation in terms of the factors influencing language change and variation. [ABSTRACT FROM AUTHOR]

How children learn the category "verb" as a grammatical construct has been the focus of much research over the past two decades. However, less attention has been paid to acquisition of smaller categories within the verb category. This paper uses data concerning the acquisition of causatives in Inuktitut to address the question of how smaller categories of verbs are learned Inuktitut has one class of verbs that permits only a morphological causative, and a second class that permits either a morphological or a lexical causative depending on the semantics of the situation. Data from eight Inuktitut-speaking children aged 1;0 through 3;6 indicate that the distinction between these classes is acquired in three stages. During the first stage, the children use only lexical causatives and there is no evidence that they understand that these structures have a causative meaning component. During the second stage, they use the morphological causative indiscriminately across verb categories to issue commands. In addition, they begin to show productivity in their use of lexical causatives in that they alternate between intransitive and lexical-causative uses of the same verb root. By stage three, children have begun to distinguish the two classes of verbs with respect to causative constructions. Across the three stages outlined above, children apparently learn to use linguistic indicators of causation first in a situation-based way (i.e. for issuing commands) and later verb-by-verb, rather than employing a class-based approach to category formation. These results are analyzed in terms of similar studies in English and Hebrew, and in terms of a general phasal approach to cognitive and linguistic development. [ABSTRACT FROM AUTHOR]

This paper addresses the issue of the relative "naturalness" of affixation and stein change as the expression of morphological categories. According to some accounts, affixation involves symbolic rules that are stored separately from lexically represented morphemes to which they are applied, and that thus enjoy a processing and acquisition advantage over stem changes. By a lexical account, inflected words are stored as lexical units, regardless of whether the inflection is by affixation or stem change, nicking the two morphological processes equally easy to acquire and generalize. We argue that the reason stein changes are less common in the languages of the world is that the grammaticization process by which morphological processes are created forms affixes more readily than stein changes. Stem changes are more difficult for children to acquire because they typically have lower type frequency. When type and token frequency were controlled in an experiment in which subjects learned plural morphology in an artificial language, we found that while type frequency affected the likelihood that suffixes would be generalized to new instances, overall suffixes and stein changes were equivalent in generalizability. [ABSTRACT FROM AUTHOR]