Morphology and dialectology in the Linguistic Survey of Scotland: A quantitative approach

We present a study of Gaelic dialectal variation based on the unpublished, morphological section of the Linguistic Survey of Scotland (Gaelic). Our findings indicate that a quantitative (dialectometric) approach can yield important insights into the range of internal variation in Gaelic and elucidate certain issues of language structure. Here, we focus on two questions: (1) what patterns of spatial variation are observed in the morphological data, and (2) how are different morphological features in the dialects correlated?

To answer the first question, we fit a logistic, generalised, additive model against a longitude-latitude spline and observe a number of patterns. Some features show a pronounced northwest-southeast cline, usually with more morphologically ‘innovative’ dialects in the east and south. Demographic variables, such as mean age and concentrations of speakers, may be a factor. Others show a clear focus on the Outer Hebrides, particularly from North Uist southwards. This suggests that the Gaelic dialects of the southern Outer Hebrides are the most morphologically ‘conservative’ of the surviving varieties. Our approach allows us to precisely quantify these patterns, providing a firm underpinning for earlier descriptions. In some cases, the data also pose challenges for received pedagogical practice and prescriptive standards.

For the second question, we use Ward clustering on a correlation matrix to identify which features frequently co-occur in the dialects. The results shed light both on the dialectal divisions of Gaelic and, arguably, on its grammatical structure. For instance, we identify a strong correlation between the absence of lenition in returns such as an cas (nominative), cas beag (nominative) and am fear (genitive). These patterns are not strongly correlated with other cases of lenition absence. We interpret this as suggesting that lenition in these contexts may be driven by a single grammatical mechanism (whose loss underlies this consistent absence of lenition), which is not in action in other lenition contexts.

Overall, our results show the potential of a quantitative approach to Gaelic dialectology and its promise for a larger study.