What does word stress tell us about morphological structure?
- Wednesday 24 May 2023
2311 BD Leiden
Consider the word beer drinker. For a long time, words of this type have been called synthetic compounds and at least three types of analyses have been defended:
(1) a. [[beer drink] er] ‘the suffixation analysis’ (adhering to binarity)
b. [[beer] [[drink] er]] ‘the compound analysis’ (adhering to binarity)
c. [[beer][drink][er]] ‘the ternary analysis’
Competing with the analysis in (1a), whatever the details, is the analysis in (1b) which view beer drinker as a regular NN compound. I will here not consider the ternary analysis in (1c) as a viable solution.
Many arguments to support the suffixation analysis have been provided by various linguists (notably including Peter Ackema and Ad Neeleman), who make their case based on semantic and syntactic considerations. The biggest problem for this approach is that it is not clear what the suffix has been attached to. Is the complex unit [beer drink] a compound or a phrase? Or is it something else?
In this talk, I will suggest that the suffixation analysis in (1a) is the correct one, based on phonological evidence, namely from word stress, which, as far as I know, has never been considered. Consider the example bierbrouwerij ‘beer brewery’. Again, there are two possible (binary) analyses:
(2) a. [[bier brouw] erIJ] ‘the suffixation analysis’
b. [[bIEr] [[brouw] erij]] ‘the compound analysis’
In this case both structures are possible. (2b) is a regular compound, meaning ‘a brewery that has something to do with beer’. Pragmatically we interpret ‘has something to do with’ as beer being what is brewed there. (We do not need a theory of inheritance for that.) (2a) means something different; it means ‘the activity of brewing beer’. What we focus on in this talk is that these two meanings correspond with two different locations of (primary) stress (as indicated with capitalization in 2). What we need to know is that the Dutch suffix -erij (unlike -ery in English) has lexical stress (on ‘ij’). In the noun brouwerIJ, stress is on ij. When we then form a regular compound, as in (2b) the compound stress rule puts primary stress on the first member [bier]. My point is now that we can only explain primary stress falling on the suffix, as in (2a), if the suffix ‘has been added last’. This means that to arrive at the final stress pattern, the morphological structure cannot be that of a regular compound; it has to be the structure in (2a). This then leads to the question what [bier brouw] is? Based on a certain formal analysis of morphological and syntactic structure, I will argue that it is neither a compound word nor a syntactic phrase. Supporting evidence will come from a class of verbs in Dutch, called separable verbs, which, as stored in the lexicon, are neither words nor phrases.
I will then include in the discussion a second class of synthetic compounds: the type broad shouldered, the Dutch equivalent being breedgeschouderd. For this type, by analogy to the suffixation analysis of the synthetic compound bierbrouwerIJ, I will assume that the suffixation analysis in (3b) is the correct one for breedgeschouderd: the affix is attached to the unit [breed schouder]. However, this affix is discontinuous: ge-X-d and as we can see, the prefixal part occurs in front of schouder, rather than in front of the whole unit (*gebreedschouderd). Interestingly, this fact has been used to argue in favor of the compound analysis in (3a):
(3) a. [[breed] [ge- schouder -d]]
b. [[breed schouder] ge-Xd]
By adhering to the suffixation analysis, I will be committed to the proposal that ge-X-d ‘wraps around’ the word schouder (which actually proves that [breed schouder] is not a (compound) word), in the phonological structure, which is:
There are independent reasons (relating to so-called structure paradoxes) for believing that a morphological structure and a phonological structure of complex words are both needed, precisely because these two structures do not have to be isomorphic.