Ok, I’ve finally had to concede defeat with this point. I’ve reserved it a place in Chapter 6 for as long as I could, but it’s just too detailed to fit. But I still like it too much to ditch it, so here it is. Context: should segmental phonology be regarded as a qualitatively separate domain from suprasegmental phonology? Someone who says they should, says so because he thinks suprasegmental features are both paradigmatic and syntagmatic. I say:
There seems to be some confusion in the literature when it is claimed that suprasegmentals are distinguished from segments by being both syntagmatic and paradigmatic (Fox (2000); this differs from Lehiste’s (1970) view that prosody is syntagmatic rather than paradigmatic). Whereas paradigms are lists of interchangeable options, syntagms are collocations; they are as different as the vertical and horizontal axes on a graph, for example, and there does not seem to be a way in which any particular linguistic phenomenon could coherently be described as both syntagmatic and paradigmatic simultaneously.
This seems to be the position of a wide variety of theorists. A very strict separation is maintained between syntagmatic and paradigmatic kinds of analysis among Firthian prosodists, for instance (Lyons 1962, Ogden & Local 1994, Waterson 1987). Although this distinction between syntagmatic ‘prosodies’ and paradigmatic ‘phonematic units’ is admittedly unique to Firthian analyses in many of its elements and implications, the incompatibility of paradigms and syntagms (or rather, more accurately, the incoherence of characterising something as both syntagmatic and paradigmatic) is shared by other very different schools of thought in phonology too. In Trubetzkoy’s case, to pick just one example, it is the phonemes and phonemic relations in a language’s inventory which are paradigmatic, while rules are syntagmatic, but again this formulation seems to clearly preclude the possibility for some phonological feature to be described as both paradigmatic (belonging to the inventory) and simultaneously syntagmatic (a rule); see Cairns (1971).)
The specific example which is used in support of the ‘both-and’ claim for prosody is time, or duration (Fox 2000): duration is said to be both a segmental property (in which case it is called ‘length’ in phonology) and also a suprasegmental property (in which case it is called ‘weight’ or ‘quantity’). But this example does not provide evidence that prosodic features can be both syntagmatic and paradigmatic – what it shows is that some acoustic property of the speech stream can be put to use in a language in either or both of these ways. That a single acoustic property of the speech stream can be multi-purpose in a language system is of course not a particularly controversial claim, but it is not a claim which contributes to the argument for or against a qualitative distinction between segmental and suprasegmental features.