Unaccusative Syntax and Verbal Alternations

Abstract

features present on the functional heads in the syntax are spelled out with phonological content in the process of Vocabulary Insertion, which takes place in morphology. Individual Vocabulary Items consist of a phonological exponent and the features that it is associated with.  us, for example, the English plural /-z/ is associated with the following Vocabulary Item: -z []. Vocabulary Items are inserted such that the item with the greatest subset of the features on a node will win out over its competitors. As a result of this, exponents, the morphophonological objects that are inserted into these nodes, may be underspecifi ed with respect to the morphosyntactic context in which they appear.  is type of underspecifi cation will fi gure prominently in the analysis presented in section ..  is grammatical architecture forces a particular approach to the study of verbal alternations.  ere is no extra-syntactic lexicon in which word-formation of any type, or in particular the derivation of one verbal class from another, can take place. What there is to say about verbal alternations is essentially syntactic, and consists in identifying the structures and features underlying particular alternations. In this way there are clear connections with the Hale and Keyser approach to argument structure (Hale and Keyser  and related work), at least to the extent that the structures proposed in that framework are actually part of the syntax and not some other component. Every theoretical framework has to list certain types of unpredictable information, whether the special meaning found with kick the bucket, or the basic sound–meaning connections found in a Root such as √DOG. In this framework, there is a further component of the grammar, the Encyclopedia, in which special meanings of the type found with idioms, light-verb constructions, and, for that matter, simple Roots are listed. Certain aspects of what is sometimes called lexical semantics are therefore stored in this list. Among other things, the fact that certain verbs enter transitivity alternations ( e vase broke and John broke the vase) while others do not ( e books arrived and *John arrived the books) implicates Encyclopedic knowledge—that is, centred on the semantic diff erences between the two roots.  .  .          e idea that unaccusatives and passives do not have external arguments is a familiar one. Here I will review some further assumptions concerning () the licensing of external arguments, and () the unaccusative analysis of refl exives. Based ultimately on arguments that the external argument is not an argument of the verb per se (cf. Marantz ), Kratzer (, ) proposes that such arguments are the specifi ers of   is position will be articulated in greater detail in sect. ..  See also Borer (in this volume) and van Hout (in this volume) for related perspectives.