This article deals with suprasegmental changes in progress in Japanese and English and demonstrates that the same kind of mechanism is in operation. Data on realtime changes in pitch accent, collected by comparing earlier editions of Japanese pronouncing dictionaries with recent editions, are examined. Analysis of data from present-day speech provides supplementary data on innovative accent. Two types of principle for accent change are proposed: the demand for paradigm coherence and the demand for canonical accent pattern. In the case of nouns derived from adjectives, the two appear to be in conflict. It is suggested that in Japanese the demand for paradigm coherence is the stronger principle. Similar kinds of conflict between two possible types of change are observed in ongoing stress changes in English. It is shown that base (or stem) transparency is a major factor in determining the directionality of accent change.

This article interrogates the line taken in studies of Northern Ireland English that Catholic/Protestant ethnicity is sociolinguistically irrelevant. Using data from Derry/Londonderry English, gathered with the objective of answering the question of whether ethnicity matters in sociolinguistic terms, it examines the relative importance of a set of social factors for language variation. The strength of these factors (ethnicity, class, sex, and age) varies, but where change is occurring, ethnicity has an effect on the adoption of innovations. In particular, changes originating in the (predominantly Protestant) east of Northern Ireland tend to be adopted primarily by Protestants, whereas Catholics tend to be more conservative. This fits well with a general pattern of diffusion of change suggested by a reading of the sociolinguistic literature from the region.

The prodigious expansion of the progressive (be + ing periphrastic form, where be is at the same time the copula and a statement of existence) was a major feature of the modernization of the English verb system in the 19th century, when its frequency quadrupled. A survey (1787–1880) of the collections of private letters from 22 people, most of them famous writers, reveals that linguistic factors played a relatively small quantitative role in this development, whereas a clear correlation is found with two sociolinguistic factors: gender and intimacy. Frequencies are consistently higher for women than for men, and they increase with more intimate correspondents. Some parallels with biographical and literary data suggest that the Romantic vision, in a wide sense, may have contributed to the rise of this concrete, expressive, warmer existential form.

In this article, I describe a new research project on York English (YrkE), a variety in northeast England. In addition to providing the first systematic linguistic documentation of YrkE, I conduct a quantitative analysis of a linguistic feature which not only is well documented in the literature, but also recurs pervasively in varieties of English worldwide—was/were variation in the past tense paradigm. Two separate tendencies are observed, neither of which can be explained by any unidimensional notion of analogical leveling of the paradigm: (1) nonstandard was in existential constructions, and (2) nonstandard were in negative tags. Both trends can be tracked in apparent time in which the contrasting behavior of men and women reveals that women are leading both types of linguistic change. In other contexts, nonstandard was is a synchronic remnant which can be traced to earlier stages in the history of English.

Gondi is a Dravidian language spoken by 2.2 million speakers (Census of India 1981) in the mountains and forests of four adjacent states in central India. Gondi is a chain of several dialects, some of which, at distant points, are perhaps not mutually intelligible. A major dialect division is provided by a two-step sound change: s- > h- in the west, north, and northwest and h- > - in the south and southeast. The present article studies this two-step sound change, which is still in progress, and establishes two facts. First, contrary to the normal expectation that this sound change would be phonetically gradual and lexically abrupt (Neogram-marian type), there is evidence that it has been lexically gradual and perhaps also phonetically gradual (lexical diffusion). Second, phonetic gradualness and regularity in implementation of sound change are properties not incompatible with the mechanism of lexical diffusion. Labov's observation that s > h > has not been reported as a lexically diffused change in many quantitative studies of Portuguese and Spanish (1981) finds a clear exception in Gondi. Under the lexical diffusion model, the regularity of a sound change is defined as the final outcome in a three-stage change of the relevant lexicon: unchanged (u), variant (u ˜ c), and changed (c). If the entire eligible lexicon passed from u to c through u ˜ c, the change would become regular. If all u ˜ c became c and for some reason no item under u became u ˜ c, the sound change would die prematurely, since the variant stage which provided the rule for the innovation would be absent. Since a regular sound change can result from either the Neogrammarian model or the lexical diffusion model, Labov's (1994:542–543) theoretical proposal of complementarity between the kinds of changes resulting from the two mechanisms calls for more studies of sound change in progress to decide the issue.