Riika Halme. 2004. A tonal grammar of Kwanyama. Namibian African Studies, Vol. 8. Köln:
Rüdiger Köppe Verlag, i-xii + 299 pp.
Reviewed by Thera Crane & Larry M. Hyman, Department of Linguistics, 1203 Dwinelle,
University of California, Berkeley CA 94720-2650.
Riika Halme’s monograph, A tonal grammar of Kwanyama, is an important contribution to the
study of the Oshiwambo (Guthrie R.20) group of Bantu languages spoken in Angola and
Namibia. Since tonal information on the languages of this group has been scanty, Halme’s broad
and rigorous coverage is particularly welcome. The work is organized into eight chapters:
Chapter 1 “Introduction” situates Kwanyama (R.21) and its neighbors, surveys past
literature, and ends with a discussion of orthography and tone marking.
Chapter 2 “Segmental phonology” provides an overview of consonant and vowel
phonology. Of particular interest is the fate of Proto-Bantu prenasalized consonants: While a
nasal + voiceless stop is realized as a voiceless nasal (e.g. *-ntu > omunhu ‘human being’), the
“Kwanyama Law” denasalizes the second of two successive prenasalized voiced consonants. In
three tables (p.16), Halme organizes NCVNCV inputs by place of articulation and provides the
Proto-Bantu reconstructions, Kwanyama reflexes, and equivalents from neighboring Ndonga
(R.22), e.g. *-jambá > Kw. ondjabá ‘elephant’ (Nd. ondjamba), *gombe > ongobe ‘head of
cattle’ (Nd. ongombe).
Chapter 3 “Tone” gives a brief introduction to tone in Kwanyama, setting the stage for
description of tonal phenomena in the rest of the book. Halme notes that previous works on
Kwanyama describe tonal patterns in quite disparate ways. This is presumably due in large part
to the difficulty of translating unevenly fluctuating surface pitches into what Halme will
ultimtely recognize as a /H/ vs. /L/ system: Sequences of Ls start quite high, each L descending
from the level of the preceding L, as in o-ku-lim-in-afan-a [ _ _ _ _ _ _ _ ] ‘to cultivate for each
other’ (p.23). On the other hand, sequences of Hs start quite low with each H rising in pitch. In a
form such as o-ku-túm-ín-áfán-á [ _ _ _ _ _ _ _ ] ‘to send to each other’, from /o-ku-túm-in-
afan-a/ , the /H/ of -túm- ‘send’ spreads to the end of the word. In addition, by a late rule of
“Tone Shift”, each tone moves one to the right. What this means is that the input to phonetic
pitch assignments is o-ku-tùm-ín-áfán-á, where ` indicates a H which has been displaced one
tone-bearing unit to the right. As seen in the pitch schemas provided by Halme, -tùm- is realized
with the lowest pitch in the word.
Given such complexities, Halme suggests that H and L can be understood in terms of
height relative to the preceding pitch (with H tones higher than or level with, and L tones lower
than, the preceding tones). She goes on to posit three basic tone rules that apply throughout
Kwanyama grammar: The first, Tone Shift has already been mentioned: After all other tone rules
apply, the tones move one mora to the right. This tone shift, similar to that seen in several other
Bantu languages, e.g. Kikuyu (Clements & Ford 1979, Clements 1984), has the interesting
consequence that the underlying identity of the final tone of a word can only be determined by its
effect on the following word. This effect has caused previous analyses of Kwanyama tone to
miss important contrasts that are “clearly identifiable” by native speakers (p. 40). For example,
as seen in (1), bimoraic noun stems elicited in isolation appear to have only a two-way contrast,
L-H or L-L. (Note that noun augments and prefixes in Kwanyama are all L).
(1) a. L-H : o-ka-dilá ‘small bird’
b. L-H : o-ka-fumá ‘small frog’
c. L-L : o-ka-vela ‘bracelet’
d. L-L : o-ka-puka ‘insect’
However, when the initial tone of the following word is taken into account, as in (2), the
expected four-way contrast (characteristic of Proto-Bantu) can be found.
(2)a. /H-H/ : o-ka-dilá ó-ká-wá ‘the small bird is nice’
b. /H-L/ : o-ka-fumá o-ká-wá ‘the small frog is nice’
c. /L-L/ : o-ka-vela o-ká-wá ‘the bracelet is nice’
d. /L-H/ : o-ka-puka ó-ká-wá ‘the insect is nice’
Halme’s exposition and exemplification of the translation of pitch patterns to H and L tones, of
the tone shift rule are nicely laid out and convincing, and she is able to greatly simplify an
apparently messy system.
The other two rules in this chapter deal with tone spread and deletion. Derivations, both
here and elsewhere in the book, are given with each mora having its own H or L tone feature and
without association lines, with the justification that “in Kwanyama, a multiple linked tone seems
to behave like a sequence of equal tones and not like a single tone.” Her rules of tone spreading
are thus shown as tone copying rather than with the dotted line convention of autosegmental
In chapter 4, “Nouns”, Halme gives the requisite information about noun classes (§4.1),
their agreement patterns (§ 4.2), nominal derivation (§4.3), and lexical tone patterns (§4.4)
before introducing rules to account for tonal alternations (§4.5), which make up the bulk of the
chapter. In particular, she deals with “tonal alternation induced by the phonological
environment” (§4.5.1) and “tonal alternation on grammatical grounds” (§4.5.2). She shows that
Kwanyama has what Schadeberg (1986) terms “tone cases”, best known in the Kikongo area
(Blanchon 1998), where noun tones depend on the function and/or position of a noun (phrase) in
the sentence (p.45). Ten heavily interacting tone rules are introduced. The short final section,
“Tonal alternations in nouns without the augment”, (§4.6) makes tentative observations about
class 1a nouns, which do not have the o- or e- augment found on all other nouns and thus behave
quite differently in their tonal patterns. A nice feature of the chapter is the inclusion of numerous
tables detailing, for example, the possible tonal patterns of nouns, their relative frequencies, and
their correlation to Proto-Bantu tones.
Chapter 5 “Verbs” is the longest chapter in the book. It begins with a short description of
verbal derivation (§5.1) and then gives extensive descriptions of verbal inflection rules (§5.2)
and their tonal effects. One new rule (“Floating Contour Simplification”) is introduced, but most
rules affecting verbs are also applicable to nouns and have been defined in the preceding
chapters. In this chapter again, Halme includes a number of clear and useful tables that show the
combinatorial possibilities and tonal effects in Kwanyama’s somewhat complex TAM system,
both in main and independent clauses. Because of its focus on tone, this chapter omits in-depth
descriptions of the semantics and pragmatics of the various verb forms; although this clearly falls
outside the scope of the work, it is still something of a pity due to the lack of other descriptions
of these for Kwanyama. Of particular interest might be the somewhat enigmatic role of the initial
vowel that appears on most, but not all, main-clause positive verb forms. Halme describes the
tonal realization of forms with and without the initial vowel (§5.4.2) and hints at its interesting
semantic features but leaves the final analysis open. This chapter, like the others, contains
numerous examples and detailed derivations of tone from its underlying representation to its
surface form in words and phrases.
Chapter 6 “Minor Word Categories” follows with short descriptions of the morphology and
tone of adjectives (§6.1), pronominal forms (§6.2), numerals (§6.3), and invariables such as
interrogatives, conjunctions, and ideophones (§6.4).
Chapter 7 “On tone in other zone R languages” summarizes the scanty information that is
available for tone in other Wambo (R.20) languages, including her own data for Mbandja, an
Angolan dialect closely related to Kwanyama. Mbandja tone is strikingly different from tone in
Kwanyama, both in its rules (Tone Shift, for example, does not apply) and its phonetic
realization, in which H and L tones are quite discrete and do not show the pitch gradation seen in
Kwanyama. The chapter ends with a description of research on tonal patterns in the well-
documented neighboring language Herero (R.41) and in Umbundu (R.11). The chapter
underscores the paucity of tonal data for zone R, languages and particularly for the Wambo lects.
Chapter 8 “Conclusion” summarizes the arguments made in the preceding chapters and
repeats all posited tone rules, both for reader convenience and to facilitate a discussion of rule
ordering, a somewhat complicated matter. Chapter 8 concludes a short summary placing
Kwanyama phenomena in context with regard to other zone R and K languages, as well as
encouraging future work on several issues raised in Halme’s investigations.
The book ends with a fairly extensive set of appendices. The first two are “Nominal
paradigms” (9 charts with tonal information and examples) and “Verbal paradigms” (49 charts
detailing the combinations and realizations of H and L verb roots with subject concords, object
concords, and clitics for Kwanyama’s various tenses and aspects). Appendix 3 contains three
short glossed texts, with each mora marked for tone. Finally, Halme includes a Kwanyama-
English vocabulary (pp, 228-299, with approximately 6000 entries) based on Turvey,
Zimmermann and Taapopi’s (1977) dictionary, with each entry marked for tone.
Halme’s data and presentation are uniformly rigorous and thorough, and her analysis is
careful. Data that that (appear to be) quite complex, both phonetically and in terms of their
interactions, are clearly presented in a straightforward and easy-to-follow format.
Even so, a reader interested in a specific issue may find herself flipping back and forth
through the pages more than might be desired. This is perhaps inevitable in such a complex array
of data; however, brief examples – or at least page references to where relevant examples can be
found – given with summaries of rules might make for greater convenience. In addition, a
summary of key tonal issues and unique features of Kwanyama (noted throughout the text) might
be of use to typologists not specializing in the language or language area.
A larger issue in the book is Halme’s assumption, stated early in the study, that “Low tones
[are] independent tonemes and not […] default tones assigned at the final stage of derivation”
(p.25). It is widely known that the great majority of Bantu languages have two tones, and the
many of these systems can be analyzed as having a privative /H/ vs. Ø opposition, and that L
tones are assigned at a later stage in the derivation (cf. Hyman 2001, Kisseberth and Odden
2003). The fact that Kwanyama does not exhibit any surface contour tones (p.52) is consistent
with a privative interpretation of Halme’s Hs and Ls.
As justification for her binary analysis, Halme cites several tone rules that appear to
crucially hinge on the presence of a floating L. However, as shown in Crane (2006), it may be
possible to account for all of the tonal effects demonstrated by Halme in a privative-H system. If
this is indeed the case, it may be preferable to the binary system assumed in Halme, as it could
reduce the need for the proliferation of rules Halme posits to account for the Kwanyama data.
For example, Halme states an otherwise identical rule twice (“High Doubling”/ “Augment High
Doubling”) for different contexts to avoid what would otherwise be an “ordering paradox” (144).
In fact, these two rules can be collapsed given several assumptions about the behavior of floating
H and the effects of the OCP in Kwanyama. The salient issues in the elimination of underlying L
would be the assumptions that the OCP functions at word level, that pro- and enclitics do not
function as part of a word with respect to the OCP, and that a floating H tone links, in order of
preference, to the following toneless mora or to the preceding toneless mora, and, in the absence
of a toneless mora to which it can attach, delinks and deletes the following H. If these
assumptions can be accepted, a reanalysis of the underlying tonal system of Kwanyama might be
In general, Halme’s clear and precise, but not overly formal presentation of tonal data for
Kwanyama make this work a very useful reference on a language family very much in need of
Blanchon, Jean. 1998. Semantic/pragmatic conditions on the tonology of the Kongo noun phrase:
A diachronic analysis. In Larry M. Hyman & Charles Kisseberth (eds), Theoretical aspects of
Bantu tone, 1-32. Stanford: CSLI Publications.
Clements, George N. 1984. Principles of tone assignment in Kikuyu. In G.N. Clements & John
Goldsmith (eds), Autosegmental studies in Bantu tone, 281-339. Dordrecht: Foris
Clements, George N. & Kevin C. Ford. 1979. Kikuyu tone shift and its synchronic consequences.
Linguistic Inquiry 10.179-210.
Crane, Thera. 2006. Two tones may not be better than one: Attempting a privative analysis of
Kwanyama tone. Ms. University of California, Berkeley
Hyman, Larry M. 2001. Privative tone in Bantu. In Shigeki Kaji (ed.), Cross-linguistic studies of
tonal phenomena, 237-257. Tokyo: Institute for the Study of Languages and Cultures.
Kisseberth, Charles & David Odden. 2003. Tone. In Derek Nurse & Gérard Philippson (eds),
The Bantu languages, 59-70. London: Routledge.
Schadeberg, Thilo C. 1986. Tone cases in Umbundu. Africana Linguistica X, 423-447.
Tervuren: Musée Royal de l’Afrique Centrale.
Turvey, B.H.C., W. Zimmermann and G. B. Taapopi. 1977. Kwanyama-English dictionary.
Johannesburg: Witwatersrand University Press.