Towards a unified account of adjuncts

After Hukari and Levine’s (1995) seminal paper on adjunct extraction and Przepiorkowski’s (1999) discussion on case-marking, a flat construal that treats adjuncts as sisters of complements has established itself as what becomes known as the Adjuncts-as-Complements (henceforth A-as-C) paradigm in HPSG (see Bouma et al. 2001, henceforth BMS01, for a systematic formulation). This type of analysis contrasts markedly with the traditional iterative adjunction analysis, which constitutes a binary configurational tree. Equally important to the flat/configurational contrast is the A-as-C theory’s claim that (at least some) lexical heads select for (at least some) adjunct(s).1This claim is indeed supported by some evidence (as we shall see shortly). However, even the A-as-C advocates do not believe their analysis to be universally applicable to all the head-adjunct phrases. BMS01 say they ‘have no reason to question the traditional wisdom in the case of preverbal adverbs’ (p.38). Also, very little argument for extending the same treatment to adnominals is offered from the A-as-C quarters, presumably because of the dearth of supporting evidence. Thus, in the current state of the theory, two systems co-exist in parallel, forcing an adjunct to receive one analysis or the other, or perhaps both (to be ambiguous). However, it is unclear whether there is evidence for such a sharp boundary or systematic ambiguity. This paper is an attempt towards reconciling the two approaches and find a unifying middle ground. We shall present an analysis that essentially reverts back to the traditional configurational structure, but nevertheless captures the two main phenomena that have motivated the A-as-C analysts, by incorporating adjuncts into the lexical head as valence values as well.