(Samenvatting lezing ACCL Seminar, Universiteit van Amsterdam, 2000)
Inflectional suffixes in Dutch have two phonological properties. First, they all consist of a coronal consonant and/or a schwa vowel. Secondly, they are 'invisible' for stress rules. For instance, even though in underived words stress is always on the last syllable if it has a long vowel plus a coda consonant, this is not the case in a word such as kanaries 'canaries', which has stress on the penult, just like the present.
In this talk I argue that these two properties of inflectional suffixes are due to the same fact: inflectional suffixes are not incorporated into the phonnological word of the base, but rather adjoined to it. If the phonological word is the basis for stress, adjunction explains the stress behaviour. If marked phonological features have to be licensed by the phonological word, also the poor feature structure of inflectional suffixes is explained. (Coronals are also the only consonants that can occur extraprosodically in underived words in Dutch.)
The question now obviously arises why inflectional suffixes have to be adjoined. I propose that this is due to a grammatical principle which requires phonological structure to mirror morphosyntactic structure as closely as possible (I call this principle MIRROR). Since inflectional suffixes are adjoined in the morphology, they should also be adjoined in the phonology. Their phonological shape and behaviour follows from this.
Next, we turn to derivational suffixation. I propose a new threefold classification for these suffixes, which in my view is more insightful than the one traditionally used in the morphological literature on Dutch.
There is a problem with satisfying MIRROR for derivational suffixes; these suffixes are morphological heads (they determine the category of the word). This fact should also be reflected in the phonology; I argue that there is a principle (HEAD) which says that morphological heads should also be phonological heads.
MIRROR and HEAD do not conflict for inflected forms, since inflectional suffixes are not heads. Yet they do conflict for derivational suffixes. In order, to satisfy MIRROR these have to be adjuncts to the word. Yet if they are adjuncts they cannot be stressed, hence, not phonological heads, so they violate HEAD. I argue that this conflict has given rise to a twofold distinction within the derivational suffixes: some satisfy MIRROR, others satisfy HEAD.
First, we have a class of suffixes that have the same phonological shape as inflectional suffixes (-@l, -@, -@s, etc). They are also 'stress neutral'; in other words they conform to MIRROR in all respects, but of course they violate HEAD, since they are completely stressless.
In order to satisfy HEAD, a suffix needs a full vowel, and this is true for basically all suffixes except those just mentioned. As a matter of fact, most of these suffixes have a superheavy syllable or are bisyllabic, which causes them to attract stress almost automatically. Hence also in this case the segmental makeup is perfectly suitable to map morphology onto the phonology.
Yet within this class of suffixes we have to distinguish between two types. First, there are the vowel-initial suffixes that are fully integrated into the phonological word of the base in order to attract its stress; second, there are consonant-initial suffixes which make the form behave as a compound.
A draft version of the paper is here.