Bantu Retention/Innovation Questionnaire (BRIQ)
1. Purpose of this questionnaire
The purpose of this questionnaire is twofold:
First, we are interested in exploring certain unresolved, possibly controversial phonological and morphonological issues arising in the diachronic analysis of the Bantu linguistic group of approximately 500 languages.
Second, we are interested to see if an interactive computer-based questionnaire of this type can be useful to scholars of Bantu in general-and hence be worthy of expansion.
As can be seen in the questionnaire (and further discussion below), our initial intention is to focus on a small number of properties which exhibit unexplained variation. In each case it is not clear what can be attributed to Proto-Bantu vs. what represents later developments in specific languages, language groups, or language areas.
We refer to this enterprise as the Bantu Retention/Innovation Questionnaire, or BRIQ for short.
We hereby welcome you to our website and thank you for your interest, participation, and potential contributions. We think it is especially important for as many of the Bantu languages to be represented in this data base as possible, including those for which only limited (and perhaps unpublished) information is available. As will be discovered, this initial attempt is modest in scope, both in terms of its coverage of issues and in terms of what it requests of participants.
In exchange for contributions, we offer the resulting materials for the benefit of all scholars and interested persons.
2. Personnel and acknowledgements
First-year seed money for this project has been provided by a small grant from the France-Berkeley Fund. The two co-directors and their coordinates are as follows:
| Larry M. Hyman |
Gérard Philippson |
| Department of Linguistics |
Laboratoire Dynamique du Langage, ISH |
| University of California |
14, Avenue Berthelot |
| Berkeley, CA 94720, USA |
69363 Lyon Cedex 07, FRANCE |
| E-mail |
E-mail |
Other participants in the France-Berkeley grant are Lolke van der Veen (DDL, Lyon), Jacky Maniacky (originally from INALCO, Paris and currently at the Musée Royal de l'Afrique Centrale, Tervuren), and Jeffrey Good (originally from UC Berkeley and currently at the University of Pittsburgh).
In anticipation of its usefulness to this project, the Berkeley-based Comparative Bantu On-Line Dictionary (CBOLD) project has been installed at the Laboratoire Dynamique du Langage (Université Lumière Lyon2/CNRS), which will have primary responsibility for its future development. The CBOLD developers are grateful for past support received from the National Science Foundation which made the data base possible.
3. Why these issues?
As is well-known, the Bantu languages exhibit a kind of theme and variations property: Most have the same typological character in their phonology, morphology and grammar, and, although there are sharp discontinuities in certain places, languages which are in geographic proximity tend to show similarities which may or may not be due to contact rather than common inheritance from a proto language. The task of the diachronic Bantuist linguist, therefore, is to sort out these features to account for both similarities and differences.
There is, of course, considerable agreement that certain features were already present in Proto-Bantu, for example: seven vowels, two tones, a full system of noun classes, an agglutinative verb structure consisting of a root and derivational suffixes (or extensions), and so forth. Many of these properties are discussed in the various chapters in the following work, which also provides extensive references:
Nurse, Derek & Gerard Philippson (eds). In press. The Bantu languages. London: Routledge (expected to appear during summer 2003).
On the other hand, there are widespread properties of Bantu languages which may or may not trace back to Proto-Bantu. It is these which have special significance and are of interest to us in this questionnaire. If such properties are innovations, rather than retentions from Proto-Bantu, then they might be of great probative value in determining subgrouping, something which is notoriously difficult within the Bantu family.
With this in mind, the questionnaire initially focuses on the following issues:
i) Vowel harmony.
Bantu languages are known for vowel height harmony (VHH) which is most transparently studied in the verb stem. Most eastern Bantu languages have what Hyman (1999) terms "asymmetric" VHH: the front series shows a lower vowel [I] or [e] after both PB *e and *o, while the back series show a lower vowel [U] only after *o (and not *e). On the hand, many western Bantu languages (and a few eastern Bantu, e.g. E.40) have symmetric VHH where the back series shows a lower vowel after *e and *o. Some southwestern Bantu languages have extended VHH so that a lower front vowel is attested after *a as well as *e and *o. In addition, some Bantu languages have front and/or round harmonies affecting *a in extensions and the final vowel, and some have labial harmony affecting extensions with the vowel /i/. Finally, some Bantu languages have little or no vowel harmony at all.
The question concerns what the situation was in Proto-Bantu? Was there VHH? If so, what shape did it take (symmetric/asymmetric)? Are languages that innovated a different pattern from Proto-Bantu genetically related (i.e. can they be subgrouped), or did the innovations diffuse areally without regard to subgroup?
To study this, our questionnaire asks how the following extensions are realized after each vowel attested in verb roots: causative *-Ic- (> -Is- etc.), applicative *-Id- (> -Il- etc.), reciprocal *-an-. We have also included the general final vowel -a, the more specific (e.g. subjunctive) -e, and the perfective *-ile ending.
We are especially interested in getting information from languages on the wide side of the Bantu geographic area.
ii) NC units
The second issue concerns the nature of what we shall refer to as NC units, widespread within the family, as in the word Bantu itself. There has been considerable interest in determining how CVNCV sequences should be analyzed: Does NC = one vs. two segments? If two, should the syllable break be CVN.CV or CV.NCV? And what about the tonal and length facts that accompany NC units?
Our focus is to try to determine the status of NC units in Proto-Bantu. We note the following variation that appears critical in resolving this status:
- The first issue concerns vowel length: Many Bantu languages that have a long vs. short (VV/V) opposition neutralize the length contrast on vowels that precede an NC unit. The vowel is redundantly long. Other languages have been reported to contrast vowel length before NC (cf. certain of Guthrie's starred forms). Was there a contrast between CVNCV and CVVNCV in Proto-Bantu, or did these latter languages undergo an innovation? If the latter, does this innovation define a genetic subgroup, or did the innovation diffuse areally?
- The second issue concerns tone, especially in languages in which CVVCV counts as three tone-bearing units (TBUs) and CVCV as two. There are some like Ganda where CVNCV counts as three TBUs, while other closely related languages such as Haya, count CVNCV as two TBUs. A number of Bantu languages show a schism depending on the presence vs. absence of an onset. Thus, in Ganda and Yao, CVNCV counts as three TBUs, while VNCV counts as two. Finally, we are interested to know what happens in languages that have lost the Proto-Bantu vowel length constrast. How do NCs function tonally in these languages? Again, we are concerned to know what the situation was in Proto-Bantu: did NC contribute a TBU or not? Whichever it was, do the innovating languages constitute viable subgroups-or are these changes sporadic? areal?
iii) The tone of causative *-i-/*-Ic-i- and passive *-U-/*-IC-U-
The last issue concerns a tonal property which appears sporadically in Bantu languages. Tonal differences have been reported on verb bases in languages such as Ganda, Nande, and Herero, which end in a (vocalic) causative or passive extension. Is this a retention, or are the reported cases innovations (which might then determine small subgroups). We are interested in tone changes which are not attributable to syllable structure (e.g. whether a verb ends in CV or CV-V). Meeussen (1967) speculated that Proto-Bantu causative *-i- and passive *-U- may have had a high tone vs. the other extensions which were toneless. We are interested to establish whether the causative and passive suffixes had a tonality different from the -VC- extensions in Proto-Bantu. We seek to find as many such cases as there may be that bear on this property-which we would think to be less likely to be borrowed (diffused) than the others.
4. Dissemination of results
Data received from contributors to BRIQ will be made available electronically.
View submitted languages
-----------------------------------------------