The Anong people of China, a small ethnic minority, live surrounded by speakers of Lisu, Bai, and Chinese. Many of them no longer speak the Anong language, having switched to one of those three languages, usually Lisu; of those who still speak it, it’s practically only used when all participants in the conversation speak Anong.
The Anong language has been studied for forty years. In those forty years, it’s moved dramatically closer to language death, and what remains of it has been radically restructured. The phonology has converged with that of Lisu, the verbal agreement and nominal case-marking systems have been mostly lost (although the agentive suffix su55 has been borrowed from Lisu and “now looks to be at least semi-productive”), and, most interestingly, the causative system has completely changed.
In the Anong of 1960, causatives were formed by prefixing sɯ-, or ɕi- before palatals, to the verb root. (A causative is a derived form of a verb: where the basic verb means ‘X’, the causative means ’cause to X’. English has no morphological causative, but Proto-Germanic did; etymologically, raise is the causative of rise.) Within forty years, the prefix fused with the root, and causatives became suppletive. For older speakers, the causatives of ŋ́ ‘be broken’, ɲí ‘know’, dím ‘collapse’, and lím ‘bury’ are ɕîdʑɯ́ŋ, ɕîɲí, ɕîdím, and ɕîlím; for middle-aged speakers, they are tɕʰɯ́ŋ, ɲ̥í, tʰím, and ɕím. Some middle-aged speakers claim to recognize the older forms, but some don’t.
Graham Thurgood, whose paper I rely on as the source for this post, asks how these changes could have operated so quickly. (The changes themselves are unremarkable, but normally they wouldn’t occur in such a short time.) Here’s his answer:
The shift to Lisu was well underway before the founding of the People’s Republic of China in 1949; in fact, it must have been well underway at the time of Barnard’s 1934 (1934:89) grammatical sketch and glossary of closely related Rawang which lists as the Rawang term for Lisu the word Anung. However, it is only in the last 30 years or so that the decline in the numbers of speakers and the increasingly restrictive usage patterns have produced the massive restructuring of Anong, not just in its causatives, but in all its systems.
Thus, it is obvious that the changes are a response to language contact and subsequent changes in usage patterns, however, to say this is to point out a correlation rather than to explain anything. Nor is it much of an analysis to invoke ‘assimilation to Lisu’ as an explanation. Where Lisu differentiates causative and non-causatives, it is with suppletion. If so, what would Anong speakers be said to be assimilating from Lisu? Certainly not the specific words. Certainly not the process.
It is, however, true that Lisu is exerting an influence on the phonology of Anong, but while Lisu influence might account for the directionality of the phonological changes, it does little to account for the pace and timing of the restructuring. The bulk of the explanation for the rapidity of the changes lies outside of the phonology of Lisu.
It is clear that the older generation of Anong speakers is the last generation to successfully learn Anong. The middle-aged and younger speakers have not done so. And, as already observed, this failure to learn Anong correlates with the rapid restructuring and with the increasingly restricted usage and access. In this sense, access to Anong has become too restricted to be successfully passed on; the middle-aged and the younger Anong do not seem to be learning it.
None of this seems controversial. However, none of this directly accounts for the rapid restructuring either. The key to understanding lies in the nature and direction of the restructuring. Our examination of the Anong causative restructuring reveal a series of changes that, although they occurred unusually rapidly, were quite natural. For instance, all of the following developments are fairly natural changes: a prefixal sɯ- losing its unstressed vowel, fusing with the following root, devoicing a root initial stop or nasal, and then dropping. Complicating the picture, however, is the side-by-side existence of an older pattern often accompanied by more than one newer form.
In fact, the changes look to natural changes—the only thing requiring an explanation is the rapidity with which they took place. Thus, returning to a suggestion made earlier in this paper, these changes are so natural, it is necessary to explain not only why the changes have not occurred elsewhere, for instance, in the strikingly similar Trung data, but also why they haven’t occurred as rapidly.
The answer is that normal transmission undoes both phonetic and morphological changes in children and, at least, in part in adult learners. Young and old learners regularly ‘repair’ their own phonetically motivated changes to make them conform to what they perceive as more desirable pronunciations. The repairs in morphology are a little more sophisticated but produce a similar result. Initially, the learner seems to simply acquire morphology as a series of individual tokens. At a later stage, learners often recognize a morphological pattern, and, in part restructure earlier forms on the basis of their generalizations—in the case of causatives by putting the prefix before the basic root. Later, some phonetically motivated changes that occurred quite naturally in the token stage of acquisition are at least partially undone when the learner recognizes the morphological pattern, in the case of causatives putting the prefix before the basic root—thus, connecting in some sense the unprefixed root with the prefixed root. This second stage, generalizing the pattern, has a tendency to undo the phonetic effect of the prefix on the root initial: the causative is reanalyzed each generation as a prefix plus the root. In Anong, however, the younger learners of Anong seem to have never gotten beyond the stage of learning individual tokens-essentially because they lacked sufficient access to language. The result was the connection between the basic root and the causative root was lost-along with the constraint that that connection imposed on phonetic and morphological restructuring. The fact that Lisu uses a lexical approach too probably played a minor role, but the major impetus to restructuring came from increasingly restricted access to the language—a restriction that made the morphological patterns too rare and thus too obscure to learn. The phonetic tendencies explain how the changes took place; the loss of sufficient access to the language explains why the changes occurred in the last 40 years.