Diachronic Change from the Perspective of First Language Acquisition: A Study of the Word Order of Wh-questions in Present-day Norwegian Dialects

In this paper I investigate some child and adult data from an acquisition corpus of spontaneous conversation of some North Norwegian (NN) dialects, focusing on the word order of wh-questions. As is well known, many dialects do not have a strict V2 requirement in this clause type, and this is often interpreted as a process of diachronic change in the direction of non-V2 (e.g. Vangsnes 2004, Westergaard 2005). Vangsnes has also shown that the variation across the different dialects is based on certain microparameters, e.g. the length of the whword. Within a cue-based approach to acquisition and change (Lightfoot 2006), I show that patterns in typical child-directed speech may contribute to changing the statistical frequencies of V2 in the input to children, and in turn causing microparametric changes in the language. The corpus consists of data from eight adults speaking a variety of NN dialects. Six of the adults speak a dialect where there is a choice of word order in questions with monosyllabic wh-words (ka ‘what’, kor ‘where’ and kem ‘who’), while two speak a dialect where the optionality also affects longer wh-phrases. There is also great variation in the proportion of V2 word order produced by the different speakers, from approx. 50% to as little as 2%. Based on a detailed investigation of the subject and verb types used with the two word orders, I identify three distinct V2 grammars, one truly mixed grammar (Tromsø), where the word order choice is based on patterns of information structure, one default non-V2 grammar (most other NN dialects), and one where the optionality has spread to longer wh-elements (Kåfjord). In the two latter grammars, V2 only survives in special cases, notably with the verb være ‘be’. This is briefly compared to the historical development in English declaratives, where the loss of V2 seems to follow similar patterns. As the corpus consists of child-directed speech, certain aspects of the adult production are interesting in terms of the effect on the process of language change. The following two factors seem to cause non-V2 word order to be more frequent in the input to children than in the language in general: Firstly, non-V2 word order is considerably more frequent with the wh-word ka ‘what’ than the other two monosyllabic question words, across all speakers (this has also been attested in elicited production experiments). In child-directed speech, this whword is significantly more frequent than the other two, making up approximately 70% of all wh-questions in the input. Secondly, in child-directed speech, monosyllabic wh-words are significantly more frequent than the longer wh-words, making up approximately 95% of all wh-questions in the corpus. This is argued to contribute to the spread of non-V2 to questions with longer wh-words in certain dialects. Furthermore, the paper argues that the reason why V2 survives with be (also in English) is that this is an extremely early acquisition. This verb makes up between 60% and 85% of all non-subject-initial declaratives with V2 in the child corpus before age 2;2, while a corresponding figure for a sample of the adult data is 23%. Finally, the paper also considers some sociolinguistic variables which contribute to a development in the opposite direction in Tromsø – towards the V2 of the standard language.