Morfologia

Morphology ≠ Syntax
Peter Ackema & Ad Neeleman 1. Introduction In this paper we sketch a model of grammar that is developed in detail in Ackema & Neeleman (to appear). We focus on the relation between syntax and morphology, and more specifically on the question whether there is a generative system for word formation separate from syntax. This is quite often denied: various authors argue thatword formation is the result of syntactic operations. We will argue that adopting a morphological module avoids serious complications in syntax.1 The issue we are primarily interested in is the one in (1a). However, this issue tends to get mixed up in the literature with the related issues in (1b-d). (1) a. b. c. d. Are the generative systems that produce words and phrases identical or distinct? Dosyntactic terminals contain phonological material or do they only contain syntactically relevant features, realized ‘later’ through spell-out rules? To what extent are the principles that govern the wellformedness of words identical to the principles that govern the wellformedness of phrases? Is there a special ‘lexical’ aspect to word formation? For example, is word formation never fullyproductive, in contrast to the formation of phrases? Are words perhaps formed in the lexicon?

A title like No escape from syntax – don’t try morphological analysis in the privacy of your own lexicon (Marantz 1997) seems to imply, for example, that morphological structure must either be generated in the syntax or in the lexicon. Thus, the answer to (1a) is linked to the answer to (1d). Similarly,evidence in favour of ‘late insertion’ of phonological forms is sometimes taken as evidence against there being separate generative systems for words and phrases, which amounts to linking (1a) and (1b). Finally, similarities in the principles that govern the structure of words and phrases are sometimes taken as evidence against the ‘two separate generative systems’ view (linking (1a) and (1c)). Infact, the issues in (1a) through (1d) are independent. In principle, any combination of answers is possible, without this resulting in an incoherent model of grammar. We will defend the view that, in addition to the phrasal syntactic generative module, there is a word-syntactic module, which produces the structures of morphological complexes. At the same time, we assume that only syntacticallyrelevant features are inserted in syntactic terminals, not phonological forms. Hence, we show that a positive answer to the question whether there is a distinct morphological module does not imply a negative stance to ‘late insertion’. In fact, we will show that certain difficulties that a model with two generative systems would seem to run into are solved by late insertion.
Overviews of the issuesthat pertain to the relation between syntax and morphology can be found in Borer 1998 and Spencer 2000, amongst others.
1

2 We cannot go into questions (1c) and (1d), but in the next section we will briefly outline why they do not bear on the issue in (1a). 2. The place of morphology in grammar The question in (1d) has, in our view, been answered conclusively by Di Sciullo & Williams (1987)and Jackendoff (1997), amongst others. There is no special relationship between morphology and the lexicon – the relationship between words and the lexicon is not different from the relationship between phrases and the lexicon. Any complex word with fully predictable properties (predictable from the principles of grammar, that is), just like any complex phrase with fully predictable properties,will not be listed in the lexicon.2 Conversely, any phrase with some unpredictable property, just like any word with some unpredictable property, is listed. This, of course, does not mean that syntax is the locus of word formation. More surprisingly, perhaps, even the question in (1c) does not bear on this issue. Consider the architecture of grammar in (2) (compare Jackendoff 1997:113).

(2)...