It has been argued that prominence assignment in triconstituent compounds depends on the compound's branching direction. Left-branching compounds (e.g. [seat belt] law) have highest prominence on the leftmost constituent, whereas right-branching compounds have highest prominence on the second constituent of the entire compound (e.g. [team [locker room]). This assumption is captured, for instance, in Liberman and Prince's (1977) 'Lexical Category Prominence Rule'. A closer look at the relevant literature reveals, however, that this generalization is not the result of large-scale empirical investigations. Instead, the generalization seems to be based on the analysis of small sets of self-selected, mostly isolated examples. In addition, the assignment of prominence to triconstituent compounds has been primarily based on the researchers' own intuition about prominence (e.g. Kingdon 1958; Liberman and Prince1977; Giegerich 1985; Giegerich 1992; Carstairs-McCarthy 2002; Berg 2009). However, this approach to determine prominence in compounds has been argued to be problematic since speakers may vary in their assignment of prominence to one and the same compound (e.g. Bauer 1983a; Plag 2006; Kunter 2011). Last but not least, the LCPR is based on the assumption that Noun+Noun (NN) compounds in English are categorically left prominent (e.g. séat belt; lócker room). Yet, recent experimental and corpus studies (e.g Plag 2006, Lappe and Plag 2007, Plag et al 2008, Plag and Kunter 2010 and Kunter 2011) dealing with prominence assignment in biconstituent compounds have shown that a considerable number of right prominent NN compounds exist in English as well. Given this variable prominence behaviour of NN compounds, which function as complex constituents in triconstituent compounds, we have good reasons to assume that right prominent NN compounds also affect prominence assignment in Noun+Noun+Noun compounds (NNN).
The thesis presents a systematic investigation of the prominence behaviour of English triconstituent NNN compounds analyzing both large amounts of speech corpus data and experimentally obtained data. On the one hand, it addresses the question to what extent the LCPR actually holds for left- and right-branching compounds. On the other hand, it presents the first study that empirically investigates whether the existence of right prominent NN compounds affects prominence assignment in triconstituent NNN compounds. In addition to that, the present thesis differs from previous accounts dealing with prominence assignment in triconstituent compounds regarding the methodology applied to account for the prominence behaviour of NNN compounds; acoustic measurements of pitch are used to determine the prominence patterns of such compounds rather than relying on a few individual speaker judgments.
The thesis shows that the LCPR is wrong and that a considerable number of violations to the rule exist. Furthermore it is shown that the existence of right prominent NN compounds needs to be incorporated in any account that tries to explain prominence assignment in triconstituent compounds. Right prominent NNs occur embedded in triconstituent compounds and they may cause the right member of the complex constituent to be the highest prominent one of the compound, contra the LCPR. Finally, the thesis provides evidence that analogy, a factor found to be responsible for prominence variation in NN compounds, also plays a role in prominence assignment to triconstituent compounds.