PROS investigates variability in stress placement in English complex words. In the first phase we found that main stress in complex adjectives ending in ‑ory and ‑able is systematically, but probabilistically, conditioned by factors that originate, on the one hand, from phonological naturalness, and, on the other hand, from properties of lexical storage. The former can be conceptualised as markedness effects; the latter can be conceptualised as effects of morphological decomposability, productivity, and paradigmatic relationships. Interestingly, the relative weighting of these factors is different for different speakers, with speakers falling into distinct groups. With regard to secondary stress placement in complex words, we found preliminary evidence that, apart from the factors that are postulated in the literature, the informativity of embedded morphological units plays a role. Crucially, it seems that relevant morphological segmentations are not restricted to traditional base-derivative relationships.
In the second phase we will pursue three main questions that have emerged from our work in the first phase, and that constitute critical cases which will enable us to further develop the theoretical models that we have tested in the first phase. These are, on the one hand, models that focus on abstract rule (or constraint) mechanisms and, on the other hand, those that focus on stored representations and linguistic experience.
The first question is whether and to what extent inter-speaker variation is correlated with differences in individual linguistic experience. We will study main stress assignment in production and in comprehension with a heterogenous group of speakers and listeners from different age groups and with different educational backgrounds, and incorporate individual vocabulary size as a predictor variable in the analysis. The second question is about the status of markedness and the role of the prosodic structure of the suffix in stress assignment. We will compare our British English production data on ‑able and ‑ory derivatives with data to be elicited from North American English speakers. ‑able and ‑ory do not differ in prosodic structure in British English, but they do in American English. The third question is how our findings on the role of informativity in secondary stress assignment interact with other factors that have been shown to be relevant in the literature, such as relative frequency and the presence of prefixes. This question will be addressed in a production study that has a twofold goal. One is to replicate our findings from the first phase with a larger dataset; the other is to extend the range of phenomena studied beyond those investigated in the first phase, incorporating (a) more different morphological categories and (b) more prosodic constellations.