Annotated Swadesh wordlists for the Tegem-Amira group (Kordofanian family).

Languages included: Lafofa [laf-lff]; Amira [laf-ami].


Schadeberg 1981 = Schadeberg, Thilo. 1981. A Survey of Kordofanian. Volume Two: The Talodi Group. Hamburg: Helmut Buske Verlag. // A brief survey of the Talodi languages of Kordofan (including Tegem-Lafofa), based on the author's own fieldwork. Includes brief descriptions of the phonetic inventories and nominal grammar (noun class systems, pronouns, etc.) of ten different languages, as well as 200-item wordlists, a lexicostatistical classification, and a first attempt at the lexical reconstruction of Proto-Talodi.

MacDiarmid & MacDiarmid 1931 = MacDiarmid, P. A.; Mac\-Diar\-mid, D. N. 1931. The langua\-ges of the Nuba Mountains. Sudan notes and records 14: 149-162. // A general survey of the Nuba Mountain region, complete with very brief illustrative wordlists for all the languages covered by the authors.

Blench 2016 = Blench, Roger. 2016. Tegem-Amira: a previously unrecognised subgroup of Niger-Congo. Ms. // A short essay on the history of the study and the current state of knowledge on Tegem and Amira. Includes previously unpublished fieldnotes on these languages, collected by R. Stevenson (Tegem) and the MacDiarmids (Amira).


1. General.

The Lafofa (Tegem) language, spoken by approximately 5,000 speakers in Janub Kurdufan, remains very poorly studied. The majority of the data comes from [Schadeberg 1981], which includes a large lexical wordlist collected directly by the author as well as, in the appendix, a separate wordlist independently collected by Robin Thelwall (discrepancies between Schadeberg and Thelwall may be explained either by inaccuracy of semantic glossing or by idiolectal differences; there is no way to ascertain that). There is, however, almost no grammatical data available, and no texts that would help clarify the actual usage of the words.

Amira (El Amira), usually considered to be a dialect of Lafofa, has been shown by R. Blench [2016] to constitute a separate language, indisputably related to Lafofa but with very significant lexical discrepancies. Unfortunately, due to the scarceness of data, it is impossible to compile a properly representative Swadesh wordlist for Amira. The only source is field data collected by P. and D. MacDiarmids; a small sample of this was published in [MacDiarmid & MacDiarmid 1931], and the rest, preserved in R. Stevenson's archives, was eventually typed up and Web-published by Roger Blench [2016], but even so, this allows to fill in less than 70 positions on the Swadesh list (not to mention the impossibility of verifying the accuracy of the MacDiarmids' glossings; internal morphological structure of many of the forms, most notably verbs, also remains poorly understood). Nevertheless, we have opted to compile a heavily "dented" wordlist all the same, if only to stress the fact that Lafofa is, indeed, not just a language isolate, but a "mini-group" that at least theoretically provides some opportunity for internal reconstruction.

Morphological segmentation for both of these languages, especially Amira, is difficult and largely provisional. It is relatively easy to segment class prefixes for nouns, particularly where the word is attested in both singular and plural forms (for Amira, however, this is done largely by analogy with Lafofa). Verbal forms are much harder, since information on Lafofa verbal morphology is minimal, and on Amira non-existent; for the most part, we segment out some initial sequences where there are strong suspicions that they constitute personal markers.

2. Transliteration.

Transcription in [Schadeberg 1981] is largely based on the IPA, so we have preserved it almost intact, with cosmetic differences concerning IPA > UTS transliteration (ʃ > š, c > ɕ, etc.). Doubled vowels in Schadeberg's transcription, marking length, have been converted to single vowels plus the length sign (aa > , etc.). Tonal systems in the languages usually involve two registers, and the low register in Schadeberg's transcriptions usually remains unmarked; we consistently mark it with the low tone diacritic (V̀).

Transcription in [Blench 2016] is a more difficult affair, since it reflects Blench's re-typing of Stevenson's manuscripts containing data from a third-party source (the MacDiarmids). The degree of phonetic accuracy here is uncertain, and the notation is sometimes unclear; thus, R. Blench makes the plausible assumption that MacDiarmids' actually reflects -ATR ɛ, a hypothesis that is well supported by external data. Consonantal notation th, dh probably reflects alveolar , and is reconverted as such. We also transcribe Blench's , as ɕ, ʓ, as these are probably palatal affricates (ɕ, ʓ; in fact, they are transcribed as c, j in the original source).

Database compiled and annotated by: G. Starostin (last update: June 2017).