The dative alternation poses a learnability paradox: when children hear give money to him and give him money, they could formulate a rule deriving the double object from the prepositional form, but the rule would allow overgeneralization from donate money to him to donate him money. Children are not corrected for speaking ungrammatically, so how do they avoid overgeneralizing? The ‘conservatism’ hypothesis proposes that children do not generalize at all; the ‘criteria’ hypothesis holds that children learn to constrain their rule to apply to monosyllabic verbs denoting possession changes. In a questionnaire study, adults rated double-object forms with novel verbs as sounding better if they met these criteria. In an analysis of speech transcripts, children were found to produce ungrammatical double-object sentences (though not very frequently). In two experiments children were taught novel motion verbs; they extended them to double-object structures, and did so more often for monosyllabic than for polysyllabic verbs and more often to denote a possession transfer than motion to a location. However, children also had a bias to use each verb in the construction they heard it in. Thus children are not invariably conservative but show conservative tendencies, and their generalizations are influenced by morphophonological and semantic criteria. We propose that speakers acquire a dative rule that operates on two levels: a broad-range rule defines the possibility of a verb meaning ‘cause to move’ to be changed into one meaning ‘cause to have’, and narrow-range rules license such extensions to be made for subclasses of semantically and morphologically similar verbs.

This paper argues that theories of phonological underspecification requiring the underlying absence of all predictable values of features are doubly problematic: too radical for certain cases, not radical enough for others. Relevant evidence in favor of a Restricted Theory of Underspecification is found in the palatalization prosody operative in the mimetic (sound-symbolic) vocabulary of Japanese. It is shown that mimetic palatalization is the surface manifestation of an independent autosegmental micromorpheme mapped from right to left onto the mimetic root. We argue furthermore that cases calling for radically underspecified representations are more adequately accounted for by a Theory of Privative Features, which alone correctly predicts the desired underspecification not only underlyingly but also throughout the phonological derivation.

In recent work Noun Incorporation has been argued to result from head movement, in which the head of an object noun phrase moves into the verb, creating a complex verb. This paper argues instead that NI derives from word formation rules applying in the lexicon, presyntactically. On the basis of clusters of grammatical properties associated with NI, it is apparent that there are two separate word formation processes that languages may choose. In one, when a noun root combines with a verb root, the argument structure of the verb is altered such that the complex verb takes one less argument. In the other form of NI, when a noun root combines with a verb root, the argument structure of the complex verb is unaltered. It is shown that the predicted grammatical properties associated with the change or lack of change in argument structure follow.

This paper offers an account of the distributional possibilities of Hungarian preverbal particles in the autolexical framework proposed in Sadock 1985. We show how a complex range of facts can be predicted from the interaction of a few independently motivated assumptions once autonomous modules of the grammar are allowed to assign partially noncongruent representations to the same linguistic entity. In the course of the discussion the Incorporative Principle of Sadock 1985 is extended and turned into a general constraint on allowable discrepancies between representations contributed by various components of the grammar.

Complementary distribution between SVO and clitic-VSO word orders in Yagua receives a natural explanation from independent principles of Case and X'-theories of the Principles-and-Parameters theory of grammar. Word-order variation in Yagua results from the fact that both clitics and arguments compete for a single Case. This conflict is resolved by moving the head of the phrase to the phrase-internal AGR[eement] position and allowing the clitic to acquire morphological visibility via incorporation into the head. Independent justification for this analysis is adduced from reflexives: all and only those arguments whose heads raise to AGR (subject, genitive, and oblique, but not direct objects) may serve as antecedents for reflexive clitics. This is explained if we require that the indices of the head, AGR, and the argument be connected. This analysis sheds new light on phrase structure, Case theory, and clitics in Universal Grammar.