‘Phoneme Fluctuation’ in low-resource languages: Theoretical problems and practical prospects

In the description of a number of under-resourced languages, there is a tradition of observing patterns of so-called ‘phonemic fluctuation’. These are cases where a phonemic contrast appears not to hold in particular lexical items, such that two or more otherwise-contrasting segments are permitted in the same environment without creating a minimal pair. An example of this type of alternation can be found in Arabela (ARL, Zarapoan, Peru), as described in Rich (1963) and exemplified in (1). Here we see that, although minimal pairs contrasting /n/ and /r/ are common (1a-b), there are also lexical items where the phonemes alternate in what appears to be a free manner (1c-d).

1. ‘Phoneme fluctuation’ in Arabela

a. /ninju/ ‘to come’

b. /ˈrinju/ ‘to breathe’

c. /ˈnjurjuku/ ‘egg’

d. /ˈrjurjuku/ ‘egg’

In this paper, we survey a series of phenomena that have been described as ‘phoneme fluctuation’ in languages ranging from North to South America, and from Africa to Australasia. We also trace the intellectual roots of the concept, from its main practitioners — the SIL missionary/language-documentation tradition — to its origin in the work of Kenneth Pike (1946), and its elaboration by Mary Ritchie Key 1968, 1979} and the French functionalists (Martinet 1969,1983, Clairis 1991). Throughout this initial review, we point out a number of the empirical and theoretical perils of the notion and look into reasons that have conspired to keep it in place. These include the fact that the theoretical availability of such free variation likely produced confirmation biases, precluding further investigation into the potential conditioning factors of the alternation, or into other theoretical frameworks that would account for them. We ultimately argue that the alleged unconditioned alternations we find in these languages are descriptively inadequate and can be accounted for by more parsimonious means, compatible with analyses of better-understood languages. We conclude that phoneme fluctuation is epiphenomenal and pernicious insofar as it creates a veneer of ‘exoticism’ or ‘simplicity’ around the languages so described.

Beyond the intrinsic interest this concept may hold for the history of linguistic ideas, we hope to show that, inasmuch as it is still part of the descriptive tradition of many languages, ‘phoneme fluctuation’ provides a gateway to some of the more revealing phonological patterns of said languages. Our main case study is the phonology of Mapudungun (ARN, isolate, Chile/Argentina) which has a long tradition highlighting these kinds of unconditioned alternations (Key 1978, Croese 1980, Salas 1992, Smeets 2008, Salamanca and Urrea 2021}, among others).