Compound stress, informativity and analogy

It has long been claimed (e.g. Sweet 1892, Bolinger 1972, Ladd 1984) that informativity has an influence on the leftward or rightward stress assigned to noun-noun combinations in English. However, the few available empirical studies of this hypothesis have provided contradictory findings (Plag & Kunter 2010, Bell 2012, Bell & Plag 2012). The present paper replicates the effect of informativity using the same measures as Bell & Plag (2012) but with a different set of data. More informative constituents in N2 position tend to be stressed. This result fits with the general propensity of speakers to accentuate important information and our results can therefore be interpreted as further evidence for an accentual theory of compound stress (e.g. Gussenhoven 2004, Kunter 2011). The results also raise the question of the relationship between informativity and constituent identity, which is the strongest known predictor of compound stress pattern (e.g. Plag 2010, Arndt-Lappe 2011). An exploration of this problem shows that the two factors are interrelated. We argue that informativity is best conceptualized as underlying other predictors of prominence, including constituent identity.