To what extent can we meaningfully deem a dialect area to be “conservative”? Remote, isolated communities are relatively cut off from linguistic contact, and thus stand a chance of staying untouched by the innovations that bubble up in busy, high-population cultural centres. On the other hand, small, isolated, close-knit social groups are also known to evolve distinctive features of their own.

In this report we examine the tension between innovation and conservation in Japanese dialectology. We look at a number of linguistic and extra-linguistic traits that select two areas of Northeastern Japan as candidates for “conservativeness”: The eastern shores of the Shimokita peninsula, and the Rikuchū coast in Iwate. Their dialects show some distinctive features that are demonstrably old, probably preserved by linguistic isolation. However, some other local features turn out to be innovative. We look at two such innovations: Vowel lowering, and early compound noun accent. For the latter, preliminary fieldwork data is presented for discussion.

It is remarkable that these innovations seem to share the same limited geographical distribution as some well-known ar-chaisms. We discuss processes by which this might happen, and what it can mean for the historical reconstruction of dialect families. The conclusion is that, even if a dialect area can be generally considered conservative, its specific features cannot be assumed to be old; they must be examined case by case.