Josef Fruehwald (2015)
In a number of spoken corpora of English, German, Dutch, Norwegian, Danish and Faroese, the filled pause 'um' is increasing in usage frequency relative to 'uh' (Tottie, 2011; Wieling et al, fortcoming). The pattern of the change looks like a textbook language change in progress, even though this is a non-conventional variable for sociolinguists, who typically take pains to demonstrate that sociolinguistic variables are not 'errorlike vagaries of performance.' Additionally, there is a body of work within psycholinguistics suggesting that 'um' and 'uh' signal different information: 'um' indicates the speaker is experiencing a greater processing difficulty than 'uh' (Clark & Fox Tree, 2002). In this talk, I'll address both the question about whether or not 'um' and 'uh' constitute different ways of saying the same thing, and the question of where filled pause selection fits into a model of grammar and speech planning, utilizing data from the Philadelphia Neighborhood Corpus. I'll conclude that 'um' and 'uh' are only partially specialized for indicating processing difficulty, making a language change in their use possible, and that highly articulated sociolinguistic knowledge need not be encoded in the narrow grammar, following on from recent thinking by Mackenzie (2012) and Tamminga (2014). Moreover, further sociolinguistic investigation into filled pauses can benefit from the pre-existing literature on filled pauses, and collaboration with our interested colleagues in psycholinguistics.
Presented at NYU Linguistics Colloquium Series