Kwakum (A91)

Elisabeth Njantcho, Mark van de Velde

To cite this version:

Elisabeth Njantcho, Mark van de Velde. Kwakum (A91). The Bantu Languages, 2019. �halshs-
02504443�

HAL Id: halshs-02504443

https://shs.hal.science/halshs-02504443
Submitted on 10 Mar 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Njantcho, Elisabeth & Mark Van de Velde. Kwakum (A91). In: Van de Velde, Mark; Koen
Bostoen; Derek Nurse; & Gérard Philippson (eds). The Bantu Languages. 2nd Edition.
[Routledge Language Family Series] London and New York: Routledge, p 383-413.

CHAPTER THIRTEEN

Kwakum (A91)

Elisabeth Njantcho & Mark Van de Velde

1 INTRODUCTION1

Kwakum (A91, ISO 639-3 kwu, glottocode kwak1266) is a cluster of Bantu language
varieties spoken in the East Province of Cameroon. Ethnologue distinguishes four dialects:
Baki, Betɛn, Til and Kwakum, which is the focus of this description. There is a high degree
of intelligibility between Kwakum and Til, whereas dialectal variation is stronger between
Kwakum, Baki and Betɛn. The Kwakum variety discussed in this chapter is spoken in the
Doume sub-division. It is referred to by its speakers as Kwàkúm. The speakers’ folk
etymology for the language name is kwày nɛ́ kúm ‘generosity and fame/prosperity’.
According to a 2013 census by Elecam, the Kwàkúm dialect has about 7000 native speakers,
whose villages are spread around the Doume sub-division (Grand Sibita, Petit Sibita, Grand
Paki, Petit Paki, Mendim, Loumbou, Kempong, Kobila, etc.).

Kwakum communities share borders with communities speaking Makaa (Bantu A83), Pol
(Bantu A92), Gbaya (Gbaya-Ngbaka-Manza) and Baka (Ubangi). Moreover, Kwakum
speaking settlements typically have immigrant communities speaking Kako (Bantu A93)
and/or languages from the Grassfields area and northern Cameroon. In this multilingual
setting, Kwakum is mainly used in informal or traditional settings (family, market, cultural
ceremonies, etc.), while communication in formal contexts (administration, school, etc.) is

1
Research for this chapter was carried out as part of the research projects BantuTyp (granted by USPC) and
LC2 Areal phenomena in Northern sub-Saharan Africa of the Labex EFL (ANR-10-LABX-0083). We wish to
thank Simon Charles Ndengue Ndengue, Paul Nargaba and Nicolas Noël Sibita, our main Kwakum
consultants, as well as Larry Hyman and Koen Bostoen for their useful comments.

1
mostly in French. In addition to these two languages, many Kwakum speakers also speak
Ewondo (Bantu A72), Kakɔ (Bantu A91) and/or Gbaya.

The data used for the grammatical analyses in this chapter were collected by the first author
between 2013 and 2017 from Kwakum speakers living in Grand Sibita. We rely on elicited
materials as well as on a corpus of recorded and transcribed spontaneous speech (narrative
and procedural texts, speeches, etc.). The only existing descriptive work on the language is
the grammar sketch written by Belliard (2005, 2007), whose focus was an ethno-
musicological study of the Kwakum people.

Kwakum is in many ways typologically unusual for a Bantu language. In order to

characterise the structure of the language within the available space limits, we had to
concentrate heavily on the morphology and the tone system. The morphological sections do
contain quite a lot of syntactic information. A more thorough analysis of the syntax and
segmental phonology of Kwakum will be provided in the doctoral dissertation of Njantcho
(forthcoming).

2 PHONOLOGY

2.1 Vowels

Kwakum has a seven-vowel system with contrastive vowel length. The (mid-)open vowels
are much more frequent than the (mid-)close vowels in stems.

front back
close i iː u uː
mid-close e eː o oː
mid-open ɛ ɛː ɔ ɔː
open a aː

Table 1: vowel phonemes

The (near-)minimal pairs in (1) illustrate that vowel length is contrastive.

2
(1) a. /i/ vs. /iː/ dʒì° ‘be’ ʤìː ‘excrement’
b. /u/ vs. /uː/ ì-bú ʃɛ̀ ‘ash’ bùːʃɛ̀ ‘gather’
c. /e/ vs. /eː/ kè ‘what’ ì-kèː° ‘egg’
d. /o/ vs. /oː/ lò ‘grow’ lòː ‘shoot’
e. /ɛ/ vs. /ɛː/ ì-ʃɛ̀ ‘raffia’ ì-ʃɛ́ː ‘sand’
f. /ɔ/ vs. /ɔː/ ɲɔ́ ‘snake’ ɲɔ́ː ‘hip bone’
g. /a/ vs. /aː/ bà ‘cut up’ bàː ‘kola nut’

2.2 Consonants

Kwakum has twenty-eight consonant phonemes, including a series of aspirated stops and a
series of prenasalised stops. The aspirated stops are reflexes of historical NC clusters, of
which the initial N tends to be a class 9/10 prefix that has dropped (2).

(2) *n-pígò (9/10) ‘kidney’ > phǐ kí ‘kidney’

*n-bèdì (9) ‘front’ > phɛ̀l° ‘in front’
*n-kákà (9/10) ‘pangolin’ > khà° ‘pangolin’
*n-gáŋgà (1/2, 9/10) ‘medicine man’ > khàː ‘medicine man’

labial alveolar palatal velar labial-

velar
oral stops - aspirated p b t d ʧ ʤ k g k͡p g͡b
+aspirated ph th ʧʰ kh
fricatives - aspirated f (v) ʃ
+aspirated ʃʰ
nasals m n ɲ ŋ
prenasalised stops mb nd ɲʤ ŋg ŋ͡mg͡b
approximants l y w

Table 2: Consonant phonemes.

In our lexical database, the voiced labio-dental fricative is rare. When it occurs before a
front vowel, it is in free variation with its voiceless counterpart /f/: ì-fítlɔ̀  ì-vítlɔ̀

3
‘darkness’, kì-vèklà  kì-fèklà ‘statuette’. The prenasalised labial-velar stop also has a
marginal distribution, as it is found in two stems only: ŋmgbéŋgá ‘fishing net’ and ŋmgbàŋ
‘crow’. The borrowing hámà ‘hammer’ is the only word that contains an [h], which we did
not include in the phoneme inventory. The voiced velar stop /g/ occurs either in front of /w/
(36 occurrences) or a (mid-) close back vowel (8 occurrences) (3). Furthermore, /g/ is found
before an open vowel in one stem (-ŋ̀gàànɔ̀ ‘refuse, deny’, ŋ̀gàtì ‘refusal’) and before a /t/ in
-dùgtàà ‘get tired’.

(3) a. gwɔ̀ɔ ́ mbɔ́ ‘chase’

b. gùʃlɛ̀ ‘pluck’

The lateral approximant /l/ is optionally realised as a trill [r] in nine words in our lexical
database. In each case, it is preceded by an alveolar obstruent /nd/ or /t/ and followed by a
mid-open vowel: -fítlɔ̀ [fítrɔ̀] ‘night’, -fìndlɛ̀ [fìndrɛ̀] ‘knock down’, tʃɛ́ndlɛ̀ [tʃɛ́ndrɛ̀]
‘candle’ and -tándlɛ̀ [tándrɛ̀] ‘attach’, or a low vowel -dʒàndlàà [dʒàndràà] ‘move
(millipede)’ and -ʃàndlàà [ʃàndràà] ‘urine’.

Voiceless stops are in free variation with affricates whenever they occur before /i/ (ì-dí  ì-
dʒí ‘bait’, tìlà  tʃìlà ‘lion’). Likewise, there is a free variation between alveolar and palatal
nasals before /i/ (nìkɔ̀  ɲìkɔ̀ ‘bend’).

2.3 Morphophonology

Nasals assimilate to a following consonant in place of articulation if a morpheme boundary

separates the two consonants. In example (4) assimilation takes place after the deletion of
the last vowel of mɔ̀ɔǹ ɔ́ ‘child’.

(4) mɔ̀ɔm
̀ pʰàâm
mɔ̀-ɔ̀nɔ́ H
-pʰàâm
1-child CON-1.man

‘boy’ or ‘the man’s child’

(5) àphǐ ŋmgbó ndɔ́

à-phìH H
-n-gbóndɔ́L

4
 2-dog PRS-PRS-bark

‘Dogs are barking.’

In non-prepausal context, there is a high tendency for some words ending in /i/, /u/, /ɛ/ and
/ɔ/ to drop their final vowel (6a). Once deletion takes place, the epenthetic vowels iɨ or u
are optionally inserted after obstruents. Their choice is determined by the roundness feature
of the preceding vowel: iɨ occurs after an unrounded vowel (6b), while u is required when
preceded by a rounded vowel or the glide /w/ (6c). Both can occur after /(m)b/ in free
variation, whatever the roundness of the vowel that precedes it (6d).

(6) a. |pʰǐkíL bùláàwɛ̀H|  pʰǐk bùláàwɛ̀° ‘many kidneys’

|à-yéklɛ̀ bùláàwɛ̀H|  àyékl ̀ bùláàwɛ̀° ‘many teachers’
|ì-kààmɔ̀ bùláàwɛ̀H|  ìkààm bùláàwɛ̀° ‘to love very much’
b. |ì-tààkɔ̀ bùláàwɛ̀H|  ìtààk( ̀) bùláàwɛ̀° ‘to take much’
c. |ǹ-bɔ̀ɔʃ̀ ɔ̀ bùláàwɛ̀H|  m̀ bɔ̀ɔʃ̀ (ù) bùláàwɛ̀° ‘much misfortune’
|pʰyààwɔ́ bùláàwɛ̀H|  pʰyààw(ù) búláàwɛ̀° ‘much blood’
d. |ì-dàámbɔ́L bùláàwɛ̀H|  ìdàámb ́ bùláàwɛ̀° ìdàámbú bùláàwɛ̀° ‘to cook much’

This morphophonological process is conditioned by various parameters, such as syllable

structure, grammatical category and tone of the final vowel. For instance, the deletion of
final /ɔ/ can occur in nominal stems with an initial heavy syllable like tʰààlɔ̀ ‘grandchild’
(7), but /ɔ/ cannot drop if the noun stem has an initial light syllable, as in tʃìlɔ̀ ‘gorilla’ (8).
Verbs, in contrast, allow the deletion of final /ɔ/ irrespective of their syllable structure (9).

(7) tʰààl dʒì tɛ́

tʰààlɔ̀ dʒìH tɛ̀H
1.grandchild be there
‘The grandchild is over there.’

(8) tʃìlɔ̀ dʒì tɛ́ (*tʃìl dʒì tɛ́)

tʃìlɔ̀ dʒìH tɛ̀H
1.gorilla be there
‘The gorilla is over there.’

5
(9) áɲtʃìl tàmbyɛ̀
H
-à-n-tʃìlɔ̀ tàmbyɛ̀
PRS-3SG-PRS-write good
‘He writes well.’

2.4 Phonotactics

The syllable nucleus can be a vowel, a nasal or the lateral /l/. Only syllables with a vocalic
nucleus can have a coda. Syllables with a nucleus /l/ can have an onset, as in à.yè.kl ̩́.kòò ‘he
taught’. There is no straightforward way to identify syllable boundaries where two
consonants succeed each other. We decided to assume a complex onset only in those cases
where a succession of consonants can occur in utterance initial position. Defined this way,
complex onsets always consist of a consonant followed by /w/ or /y/. If the glide is /w/, the
initial consonant has to be velar.2 The syllable types attested in the Kwakum lexicon are V,
N, CV(ː), CGV(ː), CV(ː)C, CGV(ː)C and CL. The rule of non-prepausal vowel deletion
described in Section 2.3 gives rise to complex codas with successions of obstruents, as in
/bǎkʃɛ́/  [bǎkʃ] ‘keep!’, /pùʃkɔ́/  [pùʃk] ‘error’ or /fóktɛ̀/  [fókt] ‘listen carefully’.

Only four words on a total of 1900 in our lexical database have a closed syllable with a long
vowel, viz. gbùːŋlɛ̀ ‘plough, turn the soil’, ì-tǎːn ‘five’, mùːŋlɛ̀ ‘uproot’ and pʰàâm ‘man,
male person’.

As is typical in the north-western Bantu languages, the distribution of consonant phonemes

over stems is heavily skewed. Table 3 shows that the occurrence of half of the consonants is
restricted to the onset of stem-initial syllables (O1). Restrictions on the possible occurrence
of consonants become stronger when we move to onsets of non-stem initial syllables (O2),
non word-final codas (CNWF) and word-final codas (CWF). Note that whenever the succession
of symbols <kp> occurs outside of stem initial position, it is the orthographic
representation of a succession of /k/ and /p/, as found in a small number of reduplicated
stems, such as kì-pɛ̀kpɛ́kì ‘end’.

2
The only exception in our lexical database is the verb stem ʃwíjɛ̀° ‘leave’.

6
 O1 O2 Cnwf Cwf
Ch, k͡p, g͡b, f, v, b, d, g, ʤ, ɲ + - - -
ʧ, ɲʤ, ŋg + + - -
p, t, k, ʃ, mb, nd + + + -
m, n, ŋ, l, y, w + + + +

Table 3: Phonotactic distribution of consonant phonemes

On a total of 911 occurrences, there are three exceptions to the generalisation that voiced
stops are restricted to stem-initial onsets: ʃílbà ‘cooking pot’, tɛ̀ndbò ‘spider’ and -dùgtàà
‘become tired.’ The first two of these can be respectively explained as a borrowing (<
English silver) and a historical compound, cf. *-tanda ‘spider, spider’s web’ and PB *-bʊ̀bɩ̀
‘spider’ (Bastin et al. 2002). Whenever the voiceless affricate /ʧ/ occurs outside of O1
position, it is followed by the vowel /ɛ/, for a reason that we do not know. The distribution
of the velar nasal /ŋ/ is exceptional, because it does not occur in O1 position, except in the
verb stem ŋwɛ̀ŋlɛ̀ ‘persist.’

In reduplicated stems, the base forms a prosodic stem: pɛ̀⁓bɛ̀là ‘seed’ (cf. -bɛ̀lɔ̀ ‘plant,
sow’), pà⁓bám ‘disapproval’ (cf. ì-bàá mɔ́ ‘reprimand’). The prosodic status of the
reduplicant is somewhat ambiguous. If the base starts in a voiced oral stop, it is devoiced in
the reduplicant, arguably because voiced stops are restricted to stem-initial position.
However, labial-velar stops are allowed in O1 position of the reduplicant, as in kpà⁓kpá tí
‘scissors’, although they too are normally restricted to stem-initial position.3

The mid-close vowels are absent from final open syllables of polysyllabic stems. The few
exceptions to this generalisation mostly involve borrowings, reduplicated stems or the Past 3
suffix -kòò.

3
A possible explanation for this is that the voiceless labial-velar stop /kp/ is not restricted to stem-initial
position, but that it simply has not been found elsewhere in our lexicon due to its low lexical frequency.

7
2.5 Tone

Syllables can be realised with a low (à), a high (á), a falling (â) a rising (ǎ), or a
downstepped high ( á) tone. In utterance final position, low tones can be realised low-falling
(à) or level low (à°) (10). In certain contexts, utterance final downstepped high tones can be
alternatively realised as level lows and vice versa. In our current analysis, the tone bearing
unit (TBU) is the mora and there is a three way underlying opposition between low, high
and zero. Underlying tones can be floating or attached to a TBU. In underlying
representations, floating low and high tones are respectively represented by the superscript
letters L
and H
(10a) and toneless TBUs are represented by means of the absence of a tone
mark (10b). Example (10) also illustrates the fact that the level realisation of utterance final
low tones is due to a following floating high.

(10) a. |ŋgwɔ̀ɔH̀ |  ŋgwɔ̀ɔ°̀ ‘brain’ (level low realisation)

b. |ʃòo|  ʃòò ‘fish sp.’ (low-falling realisation)
c. |bùpà|  bùpà ‘animal’ (low-falling realisation)

On top of lexical tones, Kwakum also has a low boundary tone L%

that can be optionally
inserted at the beginning of every utterance.

A floating tone attaches to the first TBU to its right, whose underlying tone it delinks (11a).
If it finds a succession of two TBUs of which the second is toneless, it links to both of them
(11b).

(11) a. H
-kù  kúL  [kú] ‘of the hole’
b. L%
tʃóo w-ɛ́  tʃòoH wɛ́  [ʧòò wɛ́] ‘that iron’

Kwakum has a rule of rightward tone spreading across word boundaries. Spreading tones
behave the same as floating tones in the way they attach (12).

(12) a. pú kɔ̀ndù  pú kɔ́ndù ‘the girl’s misfortune’

b. kɔ̀ndù w-ɛ́ mòò mɛ̀-ʃé  kɔ̀ndù wɛ̀ móò mɛ̀ʃé ‘that girl who came’

8
Tone spreading also takes place from prefixes to stems (13a), except in words that are in
utterance final position (13b). Example (13b) also illustrates the fact that tones link to a
following TBU if it is structurally toneless, which is why kìlɛ́wɔ́ surfaces with two high
tones in isolation.

(13) a. kìlɛ̀wɔ̀ wáàmbɔ́

kì-lɛ́wɔ w-ààmbɔ́
7-baby PP1-1SG.POSS

‘my baby’
b. kìlɛ́wɔ́ ‘baby’

There are three contexts in which the attachment of a tone to a following TBU is blocked.
The first is that a low tone cannot delink a following high if the latter is the last tone of an
utterance. Compare the tone on the demonstrative wɛ́ in example (12b) to that in (14).

(14) kɔ̀ndù w-ɛ́  kɔ̀ndù wɛ́ ‘that girl’ *kɔ̀ndù wɛ̀°

The second is the failure of a low tone to attach to a following high tone if the latter is
immediately followed by a low tone within the same word.4 In (15), spreading of the final
low tone of tààkɔ̀ is blocked by the floating low that follows the linked high of fénL
‘handles’. The same floating low prevents the linked high from spreading to the right, so
that the initial low of bùláàwɛ̀° ‘many, much, lots’ is preserved.

(15) |ì-tààkɔ̀ fénL bùláàwɛH|  ìtààk( ̀) fén bùláàwɛ̀° ‘to take many handles’

The third context in which tonal attachment is blocked is the mirror image of the second
one: a floating or spreading high cannot attach to a following TBU with a low tone when the
latter is itself followed by a high tone. However, an extra condition for blocking high tone
attachment is that the LH contour must occur in a verb stem (16a) or it must occur within
one syllable and the high part should not be floating (16b).

(16) a. |ndóm dʒɔ̀ʃɛ́L kòò yɛ̀H|  ndóm dʒɔ̀ʃɛ́ kòò yɛ̀° ‘the husband hid it’

4
The only exception attested so far is the numeral mɔ́tùH ‘one’, of which the high can be replaced by a
preceding low.

9
 b. |H-kɔ̀ɔń dɛ̀|  kɔ̀ɔń dɛ̀ ‘of the fish sp.’

Finally, there are some morphemes, including agreement prefixes, that have fixed tones and
are (optionally?) impervious to tone spreading.

Kwakum also has a number of rules of downstepping. First, downstepping takes place on the
second high in a succession of two high tones in case of an intervening floating low. This
floating low may be the result of delinking due to an incoming high from the left. In (17),
the floating high tone of the connective relator attaches to the first TBU of mòtú ‘head’, of
which it delinks the low tone, which subsequently attaches to the following high. Instead of
delinking this high, it combines with it to create a downstepped high.

(17) |H-mòtú|  mó tú ‘of the head’

In (18), the high of fén ‘handle’ cannot spread, because it is blocked by the final floating
low of this noun. Since this low is trapped in between two highs, it creates downstep.

(18) |fénL wɛ́|  fén wɛ́ ‘that handle’

No downstep formation takes place in (19) for reasons that we have already explained. The
connective high links to the first TBU of ʃùkɛ́ ‘mouse’ of which it delinks the low tone. This
delinked low cannot attach to the following high, because it is itself followed by a low
within the same word.

(19) |H-ʃùkɛ́L|  ʃúkɛ́ ‘of the mouse sp.’

Downstepping also takes place where two TBUs that are linked to a high tone meet. When
this happens across a morpheme boundary, downstep is always optionally possible, as
shown by the second downstep in (20), the one on dú.5 In contrast, within a morpheme two
adjacent high TBUs lead to downstep of the second high only in prepausal position (21a).
We have never heard clear cases of downstep of the high part of a falling tone.

5
The first downstep in (20), on gwí, is due to the low tone of the 3SG subject prefix à- that was delinked from
its TBU by the preceding floating H.

10
(20) áŋ gwí dúŋɛ́ŋ
H
-à-n-gwí dúŋɛŋ
PRS-3SG-PRS-die morning
‘It dies in the morning.’

(21) a. |ʃúlyɛ́L|  ʃú lyɛ́ ‘smoke’

b. |ʃúlyɛ́L nɛ̀ bèetàà|  ʃúlyɛ́ nɛ̀ bèètàà ‘smoke and fire’

The need for positing underlyingly toneless TBUs is demonstrated in three different ways in
examples (22-23). First, the high tone nouns in (22) appear without downstep on the second
syllable in prepausal position. This can be easily formalised by assigning an underlying High
tone to their first TBU only, which surfaces on the next TBU as well.

(22) a. |ʃólɔL|  ʃólɔ́ ‘bench’ (*ʃó lɔ́)

b. |ʤíki|  ʤíkí ‘river’ (*ʤí kí)

Second, when the low tone of nɛ̀ ‘with’ spreads onto the following TBU in (23a-b) and
deletes its high tone, it links to all unattached TBUs. This is why the low links to one TBU
in (23a) versus two TBUs in (23b). Third, in (23c), we see that the high of the first TBU of
ʃólɔ́ ‘bench’ is protected by a its final floating low, meaning that there can be no intervening
tone between them.

(23) a. |nɛ̀ ʃúlyɛ́L bùláàwɛH|  nɛ̀ ʃùlyɛ́ bùláàwɛ̀° ‘with much smoke’
b. |nɛ̀ ʤíki bùláàwɛH|  nɛ̀ dʒìkì búláàwɛ̀° ‘with many rivers’
c. |nɛ̀ ʃólɔL bùláàwɛH|  nɛ̀ ʃólɔ́ bùláàwɛ̀° ‘with many benches’

Finally, two tonal phenomena that cannot be represented in the basic description provided so
far need to be mentioned. The first is that a number of syllables with an underlying low tone
become rising when a high tone attaches to them (versus the expected high or falling
pattern). This is the case of the past tense prefix àà- (Section 6.3.1), the first syllable of the
stem of certain possessive pronouns (Section 4.4) and that of a number of nouns (24).

(24) base noun after a connective H

pʰòŋgò ‘maize’ pʰǒŋgò ‘of the maize’

11
 tʰààlɔ̀ ‘grandchild’ tʰàálɔ̀ ‘of the grandchild’
ʃɔ̀kù ‘elephant’ ʃɔ̌kù ‘of the elephant’

The second concerns a number of morphemes that change the final low tone of a preceding
word to a high. This cannot be represented by means of an initial floating H, because that
would be supposed to attach to the right, rather than to the left. We will mark these
morphemes with an initial upward arrow in their underlying representation. Examples are
the complementiser nàáL, interrogative fɛ̀ ‘where’ and the near-speaker demonstrative nɛ́L.

(25) |H-à-n-kèè nàáL/  áŋkèé nàá ‘He says that...’

3 NOUNS

3.1 The syllable structure of noun stems

The majority of noun stems are disyllabic: sixty-three percent. Two thirds of the disyllabic
stems have a CV.CV pattern. Twenty-eight percent of noun stems is monosyllabic, two
thirds of which have a CVC pattern. Seven percent are trisyllabic, mostly of the CV.CV.CV
type. These figures only contain consonant initial stems. The remaining two percent of noun
stems begin in the vowel a, which adds a syllable to the above patterns, giving rise to di-,
tri- and quadrisyllabic stems.

3.2 Nominal classification

Kwakum has eight morphological classes, defined as sets of nouns that have the same
nominal prefix. There are seven prefixes, mò-, gwò-, kì-, ì-, à-, ǹ-, n- and the lack of a
prefix, symbolised as ∅-. Three of these forms, viz. mò-, n- and ∅- are used exclusively to
mark singular nouns. Three others, gwò-, ǹ- and à-, are restricted to marking plural nouns.
The prefixes kì- and ì- mark the singular of some nouns and the plural of others. Figure 1 is
a somewhat simplified presentation of the singular-plural pairings in morphological classes.

12
Figure 1: Kwakum morphological classes.

Figure 1 excludes a small number of marginal patterns, such as kì-/ǹ- found only in the noun
kìbámbú / mbámbú ‘board/s’. Equally excluded from Figure 1 are the five nouns in our
database with irregular singular-plural pairings, two of which are reflexes of PB class 5/6
nouns with a vowel-initial stem: ʤ-í ʃí / m-í ʃí ‘eye/s’ and dí nɔ́ (⁓ì-dí nɔ́) / m̀ -mí nɔ́
(⁓ǹ-dí nɔ́ ⁓ kì-dí nɔ́) ‘name/s’. We did include pattern mò- / gwò- despite its very low
number of members, because the three nouns it contains are important: mò-mɔ́ ‘person’,
m-ɔ̀ɔǹ ɔ́ ‘child’ and mó-myáá ‘woman’.

The morphological class system shows a high degree of variability, in that many nouns can
have alternative class prefixes, mostly in the plural, but sometimes also in the singular.
During elicitation, speakers often say they do not know the plural of nouns that lack a class
prefix in the singular, or they accept alternative plural forms (26).

(26) a. ∅-mɔ̀ ⁓ ì-mɔ̀ ‘shed’ / m̀ -mɔ̀ ⁓ à-mɔ̀ ‘sheds’

b. ∅-kótú ‘bag’ / à-kótú ⁓ kì-kótú ‘bags’

Six percent of the nouns in our database begin in a syllabic nasal ǹ in the singular. This
nasal is always preserved in the plural, therefore does not commute and should not be
analysed as a morphological class marker according to our definition, even though
historically it is certainly a class marker and synchronically it can be analysed as a
derivational prefix in nouns that have a derivational relation with a verb (27). We will have
more to say on this set of nouns when we discuss gender assignment.

(27) ∅-ǹlàkʃàà / à-nlàkʃàà ‘question/s’ < làkʃɛ̀ ‘ask’

13
Finally, plural markers can be stacked (i.e. two plural markers) or additive (i.e. plural
marker added to the singular marker). We find the optional stacked plural marking à-ǹ- in
ten out of 155 nouns that have a singular in ì- and a plural in ǹ- (28).

(28) ì-ʃúm / ɲ̀-ʃúm ⁓ à-ɲ̀-ʃúm ‘bracelet/s’

Stacking is more generalised among younger speakers, who add an à before all other plural
markers. Additive plural marking can be found in nine out of 136 nouns with a singular
prefix kì- and a plural ì-. These can have an alternative plural in à-, which is then added to
the singular prefix kì- (29). Both phenomena point to a certain tendency for à- generalising
as a plural marking.

(29) kì-bɛ̀kɔ̀ / ì-bɛ̀kɔ̀ ⁓ à-kì-bɛ̀kɔ̀ ‘shoulder/s’

Kwakum has five noun classes, defined as sets of nouns that trigger the same agreement
pattern. Agreement in noun class is restricted to a relatively small set of adnominal
modifiers, almost each of which has a separate paradigm of agreement markers, given in
Table 4. These modifiers are possessive pronouns (I), demonstratives (I-V) and the
connective relator (VI). For ease of reference, we have given the Kwakum noun classes a
Bantu-style number, chosen somewhat arbitrarily (but see below for a partial justification).
The agreement prefixes of the six paradigms will all be glossed as PP in this chapter. Nouns
are assigned to noun classes on the basis of their morphological class and number.

noun morphological I II III IV V VI

class # class & number
1 mò-, n-, ∅- w- - yí- - yí- -
2 à-, gwò-, ì- (PL) y- yí- yí- yí- yí- -⁓ yì-
5 ì- (SG) ly- -  lí- lí-  yí- - yí- -
6 ǹ(-) m- mí- mí- mí- mí- -
ǹ-
7 kì- tʃ- tʃí- tʃí- tʃí- tʃí- - ⁓ kì-
Table 4: paradigms of noun class agreement markers.

14
Class 6 has a subclass for the plural of five nouns, viz. ì-tɔ́ / ǹ-tɔ́ ‘ear/s’, ì-lɔ́ / ǹ-lɔ́
‘intestin/s’, ɲ̀-ʃɛ́ɛ ́ ‘work’, ì-tú / ǹ-tú ‘day/s’ and mbɔ́ / m̀ -mbɔ́ ‘hand/s’, defined by the
agreement prefix ǹ- in paradigm VI (connectives). The brackets around the hyphen in ǹ(-) in
the second column of Table 4 are meant to show that nouns that begin in a non-commuting
syllabic nasal consonant in the singular tend to be assigned to class 6, as are the plural nouns
that begin in ǹ-, where this nasal can be straightforwardly analysed as a class prefix, because
it commutes with a different singular prefix. This could be used as an extra argument to
recognise ǹ- as a class marker in singular nouns anyway. However, a minority of thirteen
percent of singular nouns that start in ǹ trigger agreement of class 1, as if they had no prefix.
Singular ǹ- is therefore somewhere in between a canonical class prefix and the initial
syllable of a possibly derived stem.

Some of the Kwakum class markers are easily identified as cognate to the class markers of
languages with a more typical Bantu noun class system. Among the morphological class
markers, mò- corresponds to PB class 1, ì- (SG) to PB class 5, kì- (SG) to PB class 7 and ì-
(PL) to PB class 8. Many instances of Kwakum ǹ- correspond to PB class 6, as ǹ- marks the
plural of nouns that take ì- in the singular and the near totality of nouns for liquids take ǹ-.
The à- prefix may be a reflex of the PB class 2 prefix bà- that generalised to become a
default plural marker. The form gwò- that currently marks the plural of nouns that have the
prefix mò- in the singular must be an innovation.

3.3 Derivation

Slightly more than seven percent of the nouns in our lexical database are clearly in a
derivational relation with a verb. In many instances nouns are derived from verbs by the
addition of a class prefix to the verb stem. This prefix is most often ì- (30), sometimes also
kì- (30b) or ǹ- (30c). Some nouns are derived from verbs through reduplication of the verb
stem (31).

(30) a. ì-fɔ̀ɔ ́ lɔ́ ‘peace, coolness’ < fɔ̀ɔ ́ lɔ́ ‘be cool’
b. kì-lɔ́ ɔ́ ‘depth’ <lɔ́ ɔ́ ‘be deep’
c. ɲ̀-dʒòòlàà ‘bath’ < dʒòòlàà ‘bathe’

15
(31) a. pɛ̀⁓bɛ̀là ‘seed’ <bɛ̀lɔ̀ ‘plant’
b. pà⁓bám ‘disapproval’ < bàá mɔ́ ‘reprimand’
c. kà⁓kàn° ‘story’ < kàɲɛ̀ ‘tell a story’

In the other types of derivational relations, there is no straightforward way to decide which
is derived from which. Sometimes the addition of a prefix is accompanied by a change in
the stem, which may be segmental (32a), tonal (32b) or both (32c). Sometimes, the noun
does not have a prefix and the related verb and singular noun do not differ (33a) or only in
the shape of their stem (33b).

(32) a. ì-dʒɔ̀ʃù ‘hiding place’ - dʒɔ̀ʃɛ̀ ‘hide’

b. ì-yépyɛ́ ‘religion’ < yé pyɛ́ ‘believe’
c. ì-byàndʒí ‘disobedience’ < byàndʒɛ̀ ‘disobey’

(33) a. yéklɛ̀ ‘teacher’ - yéklɛ̀ ‘teach’

b. tʃèw ‘game’< tʃèwɔ̀ ‘play’

The only more or less regular derivational pattern we were able to identify thus far is the
derivation of instrument nouns involving the prefix kì- and the suffix –kà, on top of other
formal changes.

(34) a. kì-pètkà ‘plug, lid < pètɔ̀ ‘close’

b. kì-ʤìwùkà ‘key’ < ʤítlɛ̀ ‘open’
c. kì-tʃálkà ‘sharp weapon’ < tʃàá lɔ́ ‘be sharp, be fast’

4 PRONOUNS, ADNOMINAL MODIFIERS AND NOUN PHRASE STRUCTURE

4.1 Personal pronouns

Table 5 provides the paradigms of person markers in Kwakum. We distinguish between free
forms (independent pronouns) and bound forms, analysing a person marker as an affix when
its occurrence is restricted to immediate pre- or postverbal position. Some person/number
combinations lack a subject and/or object affix and are therefore always represented by
means of an independent pronoun. Kwakum does not have object prefixes.

16
 free forms subject prefixes object suffixes
1sg ɲì ǹ- -ɛɛ
2sg gwɛ̀ ɔ̀- -ɔɔ
3sg dʒɛ̀H à- -yɛ̀H
1pl ʃɛ̀H
2pl dínɛ́ nɛ̀H-
3pl dʒàH yɛ̀H- -yàH
Table 5: Person markers.

The first and second person singular object suffixes are restricted to verbs that lose their
final vowel in non-prepausal position, i.e. they attach only to verb forms that end in a
consonant (35a). Other verbs are followed by an independent pronoun (35b).

(35) a. ám bíwɛ́ɛ ̀
H
-à-n-bíwɔ́L-ɛɛ
PRS-3SG-PRS-beat-1SG

‘He beats me.’

b. ámbèè ɲì
H
-à-n-bèè ɲì
PRS-3SG-PRS-follow 1SG
‘He follows me.’

The 2nd person singular subject prefix ɔ̀- is in free variation with an independent pronoun
(36).

(36) a. ɔ́m bíwɛ́ɛ ̀

H
-ɔ̀-n-bíwɔ́L-ɛɛ
PRS-2SG-PRS-beat-1SG

‘You beat me.’

b. gwɛ́ mbíwɛ́ɛ ̀
H
-gwɛ̀ n-bíwɔ́L-ɛɛ
PRS-2SG PRS-beat-1SG

‘You beat me.’

17
In the presence of a nominal subject, the use of an additional subject marker is optional in
the plural and ungrammatical in the singular (37). However, when the nominal subject is in
focus, the verb always takes a subject marker.

(37) a. pʰàâm ɲ́ʃè°

pʰàâm H-n-ʃèH
1.man PRS-PRS-come

‘The man comes.’

b. àpʰàâm (yɛ́-)ɲ́ʃè°
à-pʰàâm H
-yɛ̀H-n-ʃèH
2-man PRS-3PL-PRS-come

‘The men come.’

Kwakum also has a set of four dual number pronouns, presented in Table 6, which can take
any position in the clause: subject (38a), primary object (38b), secondary object or
prepositional complement (38c).

1sg+2sg díʃɔ̀ɔH̀
1sg+3sg díʃɛ̀ɛH̀
2sg+3sg díɲɛ̀ɛH̀
3sg+3sg yáɲɛ̀ɛH̀
Table 6: Dual number pronouns.

(38) a. díʃɔ̀ɔH̀ mɛ́ dʒí támbyɛ̀

díʃɔ̀ɔH̀ H
-mɛ̀-dʒí tàmbyɛ̀
1SG+2SG PST2-PST2-eat.PST2 good
‘We (you and I) ate well.’
b. pʰàâm mɛ́ dʒɛ́ɛ ́ díʃɔ̀ɔ°̀
pʰàâm H-mɛ̀-dʒɛ́ɛ ́ díʃɔ̀ɔH̀
1.man PST2-PST2-see.PST2 1SG+2SG
‘The man saw us (you and me).’
c. pʰàâm mɛ́ ʃɛ́ɛ ́ nɛ́ díʃɔ̀ɔ°̀

18
 pʰàâm H-mɛ̀-ʃɛ́ɛ ́ nɛ̀ díʃɔ̀ɔH̀
1.man PST2-PST2-work.PST2 with 1SG+2SG
‘The man worked with us (you and me).’

Where a dual pronoun can be used, the corresponding plural pronoun can normally be used
too (39b), as well as the two relevant singular pronouns linked by the preposition nɛ̀ ‘and,
with’ (39c).6

(39) a. pʰàâm mɛ́ dʒɛ́ɛ ́ díʃɔ̀ɔ°̀

pʰàâm H-mɛ̀-dʒɛ́ɛ ́ díʃɔ̀ɔH̀
1.man PST2-PST2-see.PST2 1SG+2sg
‘The man saw us (you and me).’
b. pʰàâm mɛ́ dʒɛ́ɛ ́ ʃɛ́
pʰàâm H-mɛ̀-dʒɛ́ɛ ́ ʃɛ̀H
1.man PST2-PST2-see.PST2 1PL
‘The man saw us.’
c. pʰàâm mɛ́ dʒɛ́ɛ ́ ɲí nɛ̀ gwɛ̀
1.pʰàâm H
-mɛ̀-dʒɛ́ɛ ́ ɲì nɛ̀ gwɛ̀
1.man PST2-PST2-see.PST2 1SG with 2SG
‘The man saw you and me.’

4.2 Connectives

The connective construction is used to link two, typically nominal constituents. Because the
dependency relations between these constituents are ambiguous in Kwakum, we will
designate them by means of the neutral terms first relatum (R1) and second relatum (R2).
The connective relator in Kwakum is an optional floating high tone prefixed to R2, as in
(40), where the connective construction is used to express a possessive relation. The
connective relator optionally agrees in gender with the head noun (41). Only two classes
have a connective agreement prefix, viz. class 2 and class 7. Surprisingly, these agreement
prefixes, respectively yì- and kì-, have a low tone.

6
There are some co-occurrence restrictions in topic constructions between anteposed topical pronouns and
resumptive pronouns. Both cannot be dual pronouns, for instance.

19
(40) kàkàn mɔ́ɔǹ ɔ́
kàkànH H-mɔ̀-ɔ̀nɔ́
1.story CON-1-child
‘the story of the child’

(41) kìdʒìmá kìfyâl

kì-dʒìmá kì-H-fyàl
7-beautiful PP7-CON-daughter_in_law

‘a beautiful daughter in law’

Some of the adnominal modifiers discussed in the remainder of this section involve a
connective relator.

4.3 The nominaliser-linker mòò / gwòòm

Kwakum has a marker mòò (PL gwòòm) that originates in the noun mòmɔ́ ‘person’ and that
is used as a nominaliser or a linker used to introduce adnominal modifiers. It is used to
nominalise adnominal demonstratives, e.g. mòó nɛ́ ‘this one’ (see section 4.5 for
demonstratives). Moreover, it can head R2 in connective constructions that express a
possessive relation (optionally) (42) or an ordinal number (obligatorily) (43). The element
that follows mòò takes the connective relator H
-, optionally if it is a number (43),
obligatorily elsewhere, including in the expression of ‘first’ and ‘last’ (44). Mòò can itself
optionally be linked to the preceding element by the connective prefix H-.

(42) bùpà mòò mɔ́ɔǹ ɔ́

bùpà mòò H
-mɔ̀-ɔ̀nɔ́
1.animal NLNK.SG CON-1-child

‘the child’s animal’

(43) a. mòò íbá á  mòò ìbá á ‘the second one’

b. mɔ̀ɔǹ mòò íbá á  mɔ̀ɔǹ mòò ìbá a
mɔ̀-ɔ̀nɔ́ mòò H
-ì-báàH mɔ̀-ɔ̀nɔ́ mòò ì-báàH
1-child NLNK.SG CON-2-two 1-child NMLZ.SG 2-two
‘the second child’

20
(44) mɔ̀ɔǹ móò pʰɛ̌ ĺ
mɔ̀-ɔ̀nɔ́ mòò H
-pʰɛ̀lH
1-child NLNK.SG CON-1.front

‘the first child’

The nominaliser-linker mòò is also used to introduce relative clauses (see Section 7.3). Its
current distribution strongly suggests that mòò was initially grammaticalised as a
nominaliser of adnominal modifiers that were used in apposition to their head noun and
subsequently reintegrated in the noun phrase, leading to the further evolution of mòò from
nominaliser to linker in some of its uses. This is a common scenario in the Bantu languages,
responsible for the typologically unusual word order patterns in the noun phrase structures
of the family (see Section 5.3 in Van de Velde, this volume). In contrast, the origin of the
Kwakum nominaliser-linker in a noun for ‘person’ is a departure from the much more
common demonstrative origin of this element.

4.4 Possessive pronouns

Possessive pronouns follow the noun and take an agreement prefix of paradigm I (see Table
4). There are segmental and tonal differences between the stem of possessive pronouns that
agree with class 1 controllers and the stem of those that agree with nouns from the other
classes. Since these differences cannot be described by means of the synchronic rules of the
language, we treat them as being suppletive. Segmentally, the stems for 1st and 2nd person
plural possessors have vowels that are identical to those of the agreement prefix, viz. /u/ in
class 1 and /i/ elsewhere. Tonally, the 1SG possessor stem of the classes 2-7 has a rising tone
on its first syllable, the high part of which must historically originate in the high tone of
their prefix.

1sg 2sg 3sg 1pl 2pl 3pl

class 1 wààmbɔ́ ɔ̀ɔ̀H ɛ̀ɛH̀ ùʃú ùnH àaH
other classes -àámbɔ́L -ɔ́ɔ -ɛ́ɛL -íʃíL ínL áa
Table 7: Possessive pronouns.

21
INSERT TABLE 7 HERE

As shown in example (45), singular nouns can trigger agreement of their own class (45a) or
of class 1 (45b), in free variation. Note that a high tone that links to the 1SG possessor
pronoun of class 1 links to its first TBU (45b).

(45) a. kìlɛ̀wɔ̀ tʃàá mbɔ́

kì-lɛ́wɔ tʃ-àámbɔ́L
7-baby PP7-1SG.poss

b. kìlɛ̀wɔ̀ wáàmbɔ́
kì-lɛ́wɔ w-ààmbɔ́
7-baby PP1-1SG.POSS

‘my baby’

Two nouns have an alternative inalienable possessive construction, in which the possessive
modifier is merged with their stem. The inalienable construction is the most frequently used.

(46) a. kòtʃ-ɛ́ɛ°̀ ⁓ kòkù w-ɛ́ɛ°̀ ‘his/her maternal uncle’

b. ɲótʃ-ɛ̀ɛ°̀ ⁓ ɲótú wɛ̀ɛ°̀ ‘his/her body’

In order to express dual possessives, the noun is followed by a connective relator and a dual
independent pronoun (47).

(47) gwɔ̀ɔǹ (yì)díʃɔ̀ɔ°̀

gwɔ̀-ɔ̀nɔ́ yì-H-díʃɔ̀ɔH̀
2-child pp2-con-1sg+2sg
‘our children’

Independent possessive pronouns agree with their possessee in number, but not in noun
class. In the singular they take the prefix gú-, which may be a reflex of the Proto-Bantu class
17 marker, and the stem of the adnominal possessive pronouns of class 1. In the plural, their
prefix is ʤ- and their stem that of the adnominal form of classes 2-7.

22
 SG possessee PL possessee
1sg gwáàmbɔ́ ʤàámbɔ́L
2sg gwɔ́ɔ ́ ʤɔ́ɔ
3sg gwɛ́ɛ ́ ʤɛ́ɛL
1pl gúʃú ʤíʃíL
2pl gúń ʤíńL
3pl gwáá ʤáa
Table 8: Independent possessive pronouns.

4.5 Demonstratives

Kwakum has five series of demonstratives, one is used to identify referents as being close to
the speaker (-nɛ́), two are used to identify referents as being close to the hearer (-ɔ́ ɔ́ and -ɛ́),
one for items far from speaker and hearer (-kɛ́) and one for anaphoric use (-ɛ́). The
difference in use between the two near-listener demonstratives is not clear yet. In elicited
utterances they are interchangeable. The roman numbers in the headers of Table 9 refer to
the paradigms of agreement prefixes provided in Table 4.

near speaker near listener near listener far anaphoric

(II) (III) (IV) (V) (I)
1 nɛ́L y-ɔ́ɔĹ kɛ́L yí-kɛ́L (w-)ɛ́
2 yí-nɛ́L y-ɔ́ɔĹ y-ɛ́L yí-kɛ́L y-ɛ́
5 lí-nɛ́L ly-ɔ́ɔĹ  y-ɔ́ɔĹ kɛ́L yí-kɛ́L ly-ɛ́
6 mí-nɛ́L m-ɔ́ ɔ́L m-ɛ́L mí-kɛ́L m-ɛ́
7 tʃí-nɛ́L tʃ-ɔ́ɔĹ tʃ-ɛ́L tʃí-kɛ́L tʃ-ɛ́
Table 9: Demonstratives.

4.6 Quantifiers

Numbers do not agree in class with the noun they quantify. Numbers from 1 to 5 have the
formal characteristics of nouns and are assigned to morphological class kì- (SG) / ì- (PL).
Numbers from 6 to 10 do not have a class assignment.

23
1 mɔ́tùH 6 tówo
2 ì-báàH 7 tàmbályɛ̀H
3 ì-tátí 8 ʃál
4 ì-néèH 9 bùyɛ́
5 ì-tàán 10 káamɔ
Table 10: Adnominal numbers.

In cardinal adnominal use, numbers are postposed to the noun they modify (49-50). In this
construction, the numeral nouns 2-5 are always in their plural form, but the quantified head
noun can take either its singular or plural form (48), unless if they belong to morphological
class mò-/gwò-, in which case they have to be plural.

(48) a. ìtóó ìbá á

ì-tóóL ì-báàH
5-house 2-two
b. ǹtóó ìbá á
ǹ-tóóL ì-báàH
6-house 2-two
‘two houses’

(49) kìfyètì tówó

kì-fyètí tówo
7-tree six
‘six trees’

The nominal prefixes of numbers 2-5 have in common with agreement prefixes that they can
optionally prevent a preceding high from attaching.

(50) kìfyètì ìbá á ⁓ kìfyètì íbá á

kì-fyètí ì-báàH
7-tree 2-two
‘two trees’

24
Numbers from 1 to 5 can also be the first relatum in a connective construction, in which
case the resulting NP has a definite interpretation. The stem of number 2 is -bàlá in this
construction (51).

(51) ìbàlá kífyètí

ì-bàlá H
-kì-fyètí
2-two CON-7-tree

‘the two trees’

The numbers 2-5 can be used in the singular or in the plural, in free variation (52). When
such an NP is extraposed in a topic construction, the resumptive subject pronoun is always
in the plural, showing semantic agreement with the subject NP (the semantic head of which
can be a singular noun) (53). The quantified noun too can be either in its singular or plural
form, except, again if it belongs to morphological class mò-/gwò-.

(52) ɲàádʒɛ́ɛ ́ íbàlá yífyâl ⁓ kíbàlá kífyâl

ɲì-H-àà-dʒɛ́ɛ ́ ì-bàlá yì-H-fyàl
1SG-pst-PST2-see.PST2 2-two PP2-CON-1.daughter_in_law

‘I saw the two daughters-in-law.’

(53) kìbàlá kìpʰàâm wɛ̀ yɛ́ɲʃè°

kì-bàlá kì-H-pʰàâm w-ɛ́ H
-yɛ̀H-n-ʃèH
7-two PP7-CON-1.man PP1-ANAPH PRS-3PL-PRS-come

‘As for those two men, they are coming.’

The number 1 is used to mean ‘the only’, with a prefix kì- in the singular and ì- in the
plural.

(54) a. kìmɔ̀t kífyâl

kì-mɔ́tùH kì-H-fyàl
7-one PP7-CON-1.daughter_in_law

‘the only daughter-in-law’

b. ìmɔ̀t yìgwɔ́ɔǹ ɔ́
ì-mɔ́tùH yì-H-gwɔ̀-ɔ̀nɔ́

25
 2-one PP2-CON-2-child

‘the only children’

There is a set of four quantifiers that have the same grammatical behaviour as the cardinal
numbers, viz. kì-pyàpyá ‘very little, very few’, kì-ɲàá ʃí ‘little, few’, kòŋàá ʃí ‘little’ and
ʧìndí ‘all’. The first two belong to class kì-/ì-, the other two have no class prefix, but trigger
agreement of class 2. The special tonal behaviour of numbers 2-5 (optional blocking of
H
-attachment) does not apply to these quantifiers.

(55) a. ɲ̀ʧìk kípyàpyá

ǹ-ʧìkí kì-pyàpyá
6-water 7-little
‘a little water’
b. ìtàw ìpyàpyá
ì-tàw ì-pyàpyá
5-sheep 5-few
‘few sheep’

(56) a. àbùpà ʧìndí

à-bùpà ʧìndíL
2-animal 2.all
‘all the animals’
b. tʃìndí yìbúpà
tʃìndíL yì-H-à-bùpà
2.all PP2-CON-2-animal

‘All the animals’

(57) gwɔ̀ɔǹ kóŋàá ʃí

gwɔ̀-ɔ̀nɔ́ kòŋàáʃíL
2-child little
‘few children’

(58) pʰàâm mɛ́dʒí kípyàpyá kìpǒŋgò

26
 pʰàâm H-mɛ̀-dʒí kì-pyàpyáL kì-H-pòŋgò
1.man PST2-PST2-eat.PST2 7-few PP7-CON-maize

‘The man ate a small quantity of maize.’

When ʧìndí is used prenominally to mean ‘every’, it is not linked to the quantified noun by
means of a connective relator (59).

(59) tʃìɲdʒí bùpà dʒì nɛ́ kùl àɲdʒí

tʃìɲdʒíLbùpà dʒìH nɛ̀ kùl à-n-dʒíL
2.every1.animal COP with strength 3SG-CSC-eat.CSC
‘Every animal can eat.’

4.7 Qualifiers

Nominal qualification is particularly interesting in Kwakum, due to the wide range of

constructions in this domain. From a lexical perspective, Kwakum qualifiers can be divided
into those whose use is restricted to qualification (dedicated qualifiers), and those who are
also used to refer (referring-qualifying nouns). The former group can be divided into a set of
twelve underived qualifiers and an open class of derived qualifiers. Qualifiers are derived
from verbs by means of one of three suffixes, shown in Table 11.

referring- ʃómtu ‘wise, wisdom’

qualifying
dedicated underived tàmbyɛ̀ ‘good’
to derived -áá dèt-áa ‘hard’
qualification -áàwɛ̀H dèt-áàwɛ̀H ‘hard’
-ɛ́ŋ/-áŋ dèt-ɛ́ŋ ‘hard’
Table 11: Lexical types of qualifiers, with an example of each.

The suffix -áàwɛ̀H most probably originates in a possessive form consisting of the Proto-
Bantu connective stem a and a third person pronominal form. It is an instance of the
possessee-like qualifiers that are common in Northern Sub-Saharan Africa.

27
Qualifiers are similar to quantifiers in that they can be used in a variety of constructions, as
illustrated for dètáàwɛ̀H ‘hard’, which is simply preposed in (60a), R2 in a connective
construction in (60b) and R1 in a connective construction in (60c).

(60) a. dètáàwɛ̀ kídʒà

dèt-áàwɛ̀H kì-dʒà
be.hard-ADJ 7-chair
‘a hard chair’
b. kídʒà kìdétáàwɛ̀°
kì-dʒà kì-H-dèt-áàwɛ̀H
7-chair PP7-CON-be.hard-ADJ
‘a hard chair’
c. dètáàwɛ̀ yìkífyètí
dèt-áàwɛ̀H yì-H-kì-fyètí
be.hard-ADJ PP2-CON-7-tree

‘hard trees’

The underived qualifier tàmbyɛ̀ ‘good’ is simply postposed to the semantic head in the
singular (61a) and R2 in a connective construction in the plural (61b). It cannot occur in
front of the semantic head. There is no space here for a full description of the
morphosyntactic behaviour of qualifiers.

(61) a. kɔ̀ɔǹ dɛ̀ tàmbyɛ̀

kɔ̀ɔǹ dɛ̀ tàmbyɛ̀
1.plantain in.good.condition
‘a plantain in good condition’
b. àkɔ̀ɔǹ dɛ̀ yìtámbyɛ̀
à-kɔ̀ɔǹ dɛ̀ yì-H-tàmbyɛ̀
2-plantain PP2-CON-in.good.condition

‘plantains in good condition’

28
The underived qualifier tàmbyɛ̀ and the ones derived by means of the suffix -áàwɛ̀H can also
be used as adverbs, whereas those derived by means of the suffix -ɛ́ŋ/-áŋ can be used as
secondary predicates.

4.8 Agreement and word order in the noun phrase

Agreement in complex noun phrases is often determined by proximity, rather than syntactic
structure or semantic scope. The demonstrative in (62), for instance, modifies the noun ǹtóó
‘houses’, but agrees in noun class and number with the quantifier kìpyàpyá ‘few, little’.

(62) ǹtóó kìpyàpyá tʃɛ́

ǹ-tóóL kì-pyàpyáL tʃí-ɛ́
6-house 7-little pp7-anaph
‘those (aforementioned) few houses’

Elsewhere, speakers volunteer agreement with either the head noun (63-64a), or the
immediately preceding nominal form (63-64b).

(63) a. ǹtóó ìbáà mí kɛ́

ǹ-tóóL ì-báàH mí-kɛ́L
6-house 2-two PP6-DEM

‘those two houses’

b. ǹtóó ìbáà yí kɛ́
ǹ-tóóL ì-báàH yí-kɛ́L
6-house 2-two PP2-DEM

‘those two houses’

(64) a. kìbàlá kìntóó màá mbɔ́

kì-bàlá kì-H-ǹ-tóóL m-àámbɔ́L
7-two PP7-CON-6-house PP6-1SG.POSS

‘my two houses’

b. kìbàlá kìntóó ʧ-àá mbɔ́

kì-bàlá kì-H-ǹ-tóóL ʧ-àámbɔ́L

29
 7-two PP7-CON-6-house PP6-1SG.POSS

‘my two houses’

Class 1 serves as a default agreement class for all singular controllers, which can optionally
be used instead of the controller’s lexically determined agreement class. Occasionally, we
have found examples of this default agreement with a plural controller too, mostly in noun
phrases that contain a numeral.

Word order in complex noun phrases is schematised in (65). QUAL and QUANT are formal-
functional notions, used for the quantifiers (including numbers) and qualifiers that are
juxtaposed to the noun they modify, i.e. excluding the use of these lexemes in connective
constructions.

(65) HN - {QUAL, QUANT, POSS, ANA, CON} - REL – DEM

The abbreviation HN stands for the head noun from a morphosyntactic point of view: the
nominal element that is not used to modify any other element in the noun phrase. Needless
to say, this is not necessarily the semantic head, which happens to come at the very end of
the NP in (66).

(66) ìbàlá yìdétáá tʃìndí wɛ̀ɛ ̀ dʒó wɔ́

ì-bàlá yì-H-dèt-áa H
-tʃìndíL w-ɛ̀ɛH̀ H
-dʒòwɔ́
2-two PP2-CON-be.hard-ADJ CON-all PP1-3SG.POSS CON-day

‘all his two difficult days’

5 ADPOSITIONS

Kwakum has six prepositions and three “ambipositions”. The latter are postposed with
nominal complements and preposed with pronominal complements.

(67) a. prepositions
nɛ̀ ‘with, by’
pɔ̌mbú ‘for’
pʰɛ̀lH ‘in front of’

30
 ʃìmɔ́ ⁓ ʃǐm ‘behind’
lémɛ́ lémɛ́ ‘between’
pákláá ‘among’
b. ambipositions
ʃìH ‘under’
kóóL ⁓ kólL ‘on’7
téé ‘in’

Except for the comitative-instrumental-agentive preposition nɛ̀, they all originate from nouns
in a connective construction, which is synchronically evidenced by the fact that they take
possessive pronouns as pronominal complements, rather than personal pronouns. On their
pronominal complements, they take the agreement pattern of the nouns from which they
originate, which is in all cases class 1. However, the pronominal complements of the
prepositions pɔ̌mbú ‘for’, pʰɛ̀lH ‘in front of’, kólL ‘on’ and ʃìmɔ́ ‘behind’ can alternatively
have the y- prefix of class 2, in which case they do still have the tone pattern of possessives
of class 1 (68).

(68) a. pɔ̌mb yɛ́ ɛ́  pɔ̌mb wɛ́ ɛ́ ‘for him’

b. ʃǐm yɛ́ ɛ́  ʃìmɔ́ wɛ́ ɛ́ ‘behind him’

In order to disambiguate between the adpositional and the nominal use of these lexemes, the
nominaliser-linker mòò can be used in the nominal use.

(69) pʰɛ̀l (móò) tóó

pʰɛ̀lH mòò H
-ì-tóóL
1.front NLNK CON-5-house

‘the front of the house’

The complements of all Kwakum adpositions are accessible to relativisation, but only nɛ̀ can
be stranded. The other adpositions require a resumptive pronoun.

7
The form kólL is obligatory in prepositional use, i.e. with a pronominal complement.

31
6 VERBS

6.1 The structure of verb stems

Table 12 summarises the most frequent syllable structures of underived verb roots. As can
be seen, ninety per cent of the verb stems are disyllabic, which is unusual for a Bantu
language.

CV scheme % example
CV.CV 25 bè.nɔ̀ ‘deny’
CVC.CV 21 yék.lɛ̀ ‘teach’
CVV.CV 17 bà.àndɔ̀ ‘peel’
CV.CVV 10 dò.wáà ‘call’
CVC.CVV 5 lùk.làà ‘buzz’
CV.CCV 3 fí. myɛ́ ‘wipe’
CVV 5 bèè ‘follow’
Table 12: The most common syllable schemes of underived verb stems.

Ten verb stems in our database of 614 verbs are trisyllabic. There is only one verb stem that
ends in a consonant, viz. kɛ̀n ‘go’. There are strong phonotactic constraints on the last vowel
of verb stems: /ɛ, ɔ, aa/ in disyllabic stems and /ɛ, aa/ in trisyllabic stems.

6.2 Derivation

Kwakum has four verb-to-verb derivational suffixes. One is valency increasing

(causative -ʃɛ̀), two are valency reducing (-yɛ̀ and reciprocal -àà) and one can be either (-lɛ̀).
These suffixes almost always attach to a base with CVC-shape, usually obtained by the
addition of /y/ to CV roots8 and /ŋ/ to CVV roots9 or by deletion of the last vowel of
CV(V)CV roots and the shortening of the vowel of their initial syllable if that happens to be

8
The only examples we have in our lexical data base consist of CV verbs derived by means of -àà. We don’t
know whether /j/ will also occur in coda position if the verb is derived with other suffixes.
9
Some exceptions include ʃáá ‘do’ and ʃɛ́ɛĹ ‘work’ where the inserted consonant is /l/ (e.g. ʃááL ‘do’> ʃàláà,
‘be done’; ʃɛ́ɛĹ ‘work’ > ʃɛ̀láà ‘be processed’) and dʒɛ́ɛ ́ ‘see’ where it is /n/ (dʒɛ́ɛĹ ‘see’ > dʒɛ̀náà ‘see each
other’). These “inserted” consonants are most probably retentions of root consonants that have eroded in other
contexts.

32
long.10 CVCV roots with an alveolar second consonant drop their last syllable and lengthen
their first vowel in front of the suffix -ʃɛ̀. We have no examples of verbs derived from a
trisyllabic root.

Verbs derived from roots that are not entirely low have a fixed tone pattern determined by
the derivational suffix, as illustrated in (70). Entirely low roots remain low when a
derivational affix is added.

(70) a. -ʃɛ̀: HL.L

lííL ‘get black > lîŋ-ʃɛ̀ ‘blacken’
bélɔ́L ‘be cooked’ > béè-ʃɛ̀ ‘cook’
dʒàálɔ́L ‘give birth’ > dʒáà-ʃɛ̀ ‘deliver’
b. -lɛ̀: H.L
béʃɛ́L ‘rise’ > béʃ-lɛ̀ ‘lift’
c. -àà: L.HL
fɛ́ ‘give’ > fɛ̀y-áà ‘receive’
dʒɛ́ɛĹ ‘see’ > dʒɛ̀n-áà ‘see each other’
bíwɔ́L ‘beat’ > bìw-áà ‘beat each other’
d. -yɛ̀ : H.HL
dʒàálɔ́L ‘give birth’ > dʒál-yɛ́L ‘be born’

There is only one verb in which the suffix -ʃɛ̀ is not clearly causative, viz. ʃèk-ʃɛ̀ ‘shake’
(< ʃèkɔ̀ ‘sieve’). Some typical causative examples are provided in (71).

(71) lîŋ-ʃɛ̀ ‘darken’ < líí ‘get dark’

pûp-ʃɛ̀ ‘clean’ < pú pɔ́ ‘recover’
ɲîŋ-ʃɛ̀ ‘let in’ < ɲîŋlɛ̀ ‘enter’
dáà-ʃɛ̀ ‘put to bed (sp.)’ < dàá lɔ́ ‘sleep’
dèè-ʃɛ̀ ‘strengthen’ < dètɔ̀ ‘be strong’

10
The only exception in our database is dʒòòlàà ‘bathe (intr.)’ (cf. dʒòòlɛ̀ ‘bathe (tr.)’) where shortening of the
initial long vowel fails to occur.

33
The suffix -lɛ̀ has a causative meaning in one verb, namely béʃ-lɛ̀ ‘lift’ (<bé ʃɛ́ ‘rise’). In
dít-lɛ̀ ‘open’ (<dí tɔ́ ‘close’), it encodes reversal. Its function is decausative in tʃík-lɛ̀ ‘stay’
(<tʃí kɔ́ ‘abandon, leave something’) and perhaps intensive or repetitive in bòm-lɛ̀
‘hammer’ (<bòmɔ̀ ‘break’).

(72) ámbéʃl ̀ bùpà

H
-à-n-béʃɛ́L-lɛ̀ bùpà
PRS-3SG-PRS-rise-CAUS animal
‘He lifts up the animal.’

The suffix -àà can be used to derive reciprocal, reflexive, passive and/or decausative verbs.
Some examples are provided in (73-74). In addition, in two verbs, -àà serves to express
collective actions (dùlyáà ‘shout together’< dú lɔ́ ‘shout’ and kàw-àà ‘share among many
people <kààwɔ̀ ‘share’). The reciprocal meaning is often reinforced by the phrases yɛ̀ nɛ́
dʒɛ̀° ‘each other’ (SG, i.e. the subject and one other participant) and yà nɛ́ dʒà° ‘each other’
(PL, i.e. the subject and more than one other participants), while a pronoun preceded by a
floating high tone and followed bytʃítʃɛ́ ‘oneself’ is used to specify a reflexive meaning.
Example (74) shows that passive constructons with a verb derived by –àà can have an
agentive complement flagged by means of the preposition nɛ̀.

(73) a. Reciprocal
dʒɛ̀n-áà ‘see each other’ < dʒɛ́ɛ ́ ‘see’
kàm-áà ‘love each other’ < kààmɔ̀ ‘love’
b. Reflexive
kɛ̀ŋ-àà ‘shave’ (intr.) < kɛ̀ɛ ̀ ‘shave’ (tr.)
fìʃ-áà ‘cover oneself’ < fí ʃɔ́ ‘cover’
c. Passive voice
bày-àà ‘be operated’ < bà ‘operate’
ʃàl-áà ‘be done’ < ʃáá ‘do’
kɛ̀k-áà ‘be circumcised’ < kɛ́ kɔ́ ‘circumcise’
màŋ-àà ‘be gathered’ < màà ‘gather (tr.)’
d. Decausative
màŋ-àà ‘gather (intr.)’ < màà ‘gather (tr.)’

34
 bànd-áà ‘bend (intr.)’ < bàá ndɔ́ ‘bend (tr.)’
pàɲdʒ-àà ‘scatter (intr.)’ < pàɲdʒɛ̀ ‘separate’

(74) àdɔ̀ɔḱ ḿmàŋàà nɛ̀ pʰàâm

à-dɔ̀ɔḱ ɔ́L H
-n-màŋ-àà nɛ̀ pʰàâm
2-mango PRS-PRS-gather-PASS with 1.man
‘The mangoes are gathered by the man’

Our lexical database contains four verbs derived by -yɛ̀.

(75) ʃá n-yɛ́ ‘split (intr.)’ < ʃàá nɔ́ ‘split (tr.)’

pém- yɛ́ ‘change (intr.)’ < pémlɛ̀ ‘change (tr.)’
dʒál- yɛ́ ‘be born’ < dʒàá lɔ́ ‘give birth’
bòm-yɛ̀ ‘burst (intr.)’ < bòmɔ̀ ‘burst (tr.)’

We have found only one example of stacked derivational suffixes, viz. the verb bɔ́mʃáà
‘sell’ (<bɔ̀ɔ ́ mɔ́ ‘buy’), which is formed by means of the suffixes-ʃɛ̀ and -àà.

Verbal derivation is not productive in Kwakum. Causation can be expressed by means of a

complex sentence that has ʃáá ‘do’ as its main verb. The lexical verb occurs in a
complement clause introduced by the complementiser nàáL (76).

(76) àfɛ̀ɛ ́ ʃáá nàá ɔ̀ʃûm ìtóó

à-fɛ̀ɛĹ ʃááL nàáL ɔ̀-ʃúmɔ̀ ì-tóóL
3SG-FUT1 do COMP 2SG-build.SBJV 5-house
‘He will make you build a house.’

Periphrastic passives are productively formed with a copula and a participial form of the
verb derived by means of the suffix -ɛ́ŋ/-áŋ (see Section 4.6, table 11). As with derived
passive verbs, the agent is introduced by the preposition nɛ̀.

(77) ìtóó dʒì ʃú mɛ́ŋ nɛ́ pʰàâm

ì-tóóL dʒìH ʃùm-ɛ́ŋ nɛ̀ pʰàâm
5-house COP build-ADJ by man

35
 ‘The house is built by the man.’

There is no applicative derivation. Beneficiaries are dependent marked by means of the

preposition pɔ̌mbú.

(78) áɲʃùm ìtóó pɔ̌mb pʰàâm wɛ́

H
-à-n-ʃùmɔ̀ ì-tóóL pɔ̌mbú pʰàâm w-ɛ́
PRS-3SG-PRS-work 5-house for man PP1-ANAPH

‘He is building a house for the man.’

6.3 Inflection

6.3.1 Tense and aspect in the indicative mood

In the indicative mood, Kwakum distinguishes 15 tense-aspect combinations, summarised in

Table 13. It has four past tenses, a present and three future tenses, as well as a distinction
between perfective versus imperfective aspect. Past 1 and 2 are not distinguished in the
imperfective.

perfective temporal reference imperfective

Past 4 remote Past 4
Past 3 yesterday Past 3
Past 2 today Past 2
Past 1 immediate
Present now Present
Future 1 immediate/today Future 1
Future 2 tomorrow Future 2
Future 3 remote Future 3
Table 13: temporal and aspectual distinctions

More research is needed on the exact use of the different past and future tenses. Basically,
Past 4 is a remote past, Past 3 is typically used for events that took place the day before
utterance time. Past 2 and Past 1 are today’s pasts, with Past 1 being used for events that

36
took place right before the time of utterance and Past 2 for earlier events. The distinctions in
the future are similar.

Table 14 gives an overview of the verb forms that express the eight tenses in the perfective.
By way of an example, the last column gives the surface representation of the inflected verb
ì-dàá mbɔ́ ‘to cook’ in utterance final position, with a third person singular subject prefix à-.
Verb forms that can only be used with nominal subjects are illustrated with the noun pʰàâm
‘man’ in subject position.

PST4 (SM-)STEM-mɛ àdámb( ́)mɛ́

PST3 (SM-)STEM kòò àdàmb( ̀) kóò
PST2 H
-mɛ̀-STEM pʰàâm mɛ́dáámbɛ́
SM-H-àà-STEM àádáámbɛ́
PST1 mɛ̀-STEM pʰàâm mɛ̀dáámbɔ́  mɛ̀dáámbɛ́
SM-àà-STEM ààdáámbɔ́  ààdáámbɛ́
PRS H
-(sm)-n-stem ándàá mbɔ́
FUT1 (SM-)fɛ̀ɛĹ -STEM àfɛ̀ɛd́ àá mbɔ́
h
-sm-stem ádàá mbɔ́
FUT2 (SM-)ʃɔ̀ɔ-̀ H-STEM àʃɔ̀ɔd̀ àá mbɔ́
FUT3 (SM-)ʃɔ̀ɔ-̀ ŋgɛ̀-H-STEM àʃɔ̀ɔŋ̀ gɛ̀dàá mbɔ́
Table 14: Kwakum tenses (perfective forms)

The schemes in Table 14 show that tense is expressed by means of prefixes, suffixes and
postverbal particles or clitics (in PST3) and that tense prefixes can be tonal morphemes. Past
2 and Past 1 have different tense prefixes depending on whether their subject is nominal or
pronominal. An interesting feature of the initial floating high tone in the Present and Future
1 is that it is realised on the subject whenever it is pronominal, whether it is a prefix or an
independent pronoun, as shown in (79), repeated from (36).

(79) a. ɔ́m bíwɛ́ɛ ̀

H
-ɔ̀-n-bíwɔ́L-ɛɛ
PRS-2SG-PRS-beat-1SG

‘You beat me.’

37
 b. gwɛ́ mbíwɛ́ɛ ̀
H
-gwɛ̀ n-bíwɔ́L-ɛɛ
PRS-2SG PRS-beat-1SG

‘You beat me.’

Table 14 also shows that the future has as many formal distinctions as the past. Since we
were not able to find any difference in the use of the two hodiernal future tense forms, we
treat them as allostructs of Future 1. More research might identify a functional difference
between both verb forms.

Finally, the forms in the third column of Table 14 show that the verb stem in the past tenses
differs tonally and segmentally from the basic allomorph dàá mbɔ́ found in the other tenses
and in the infinitive. Every verb stem has three or four non-basic allomorphs. Their shape is
predictable and determined by tense and mood, by the position of the verb in the utterance
(final versus non-final), as well as by the tone, the syllable structure and the final segment of
the basic allomorph. A full description of the patterns of allomorphy will be provided in the
doctoral dissertation of the first author (Njantcho forthcoming). In Table 14, the use of a
non-basic allomorph of the verb stem is signalled by curly underlining.

Tense suffixes are treated as part of the verb stem with respect to the application of fixed
tone schemes in non-basic stem allomorphs. This observation justifies the distinction in
morphological status of the post-stem tense marker between the Past 4 suffix -mɛ and the
Past 3 unbound morpheme (or clitic?) kòò. The difference in their behaviour is illustrated in
the verb forms in Table 15. In the second column, F is short for (utterance) final and NF for
non-final. The third column contains the tone scheme of the non-basic allomorph of the verb
stem. The examples in the last column have a 3SG subject prefix à-. The verb bà means ‘cut
up’ and ʤɔ̀ʃɛ̀ means ‘hide’.

Stem Conjugation Tone scheme Example

bà PST4 F LH àbàmɛ́
PST4 NF LH àbàmɛ̀H
PST3 LH àbà kóò
ʤɔ̀ʃɛ̀ PST4 LHH àʤɔ̀ʃɛ́ mɛ́

38
 PST3 LHL àʤɔ̀ʃɛ́ kòò
Table 15: The difference in morphological bonding between -mɛ and kòò

INSERT TABLE 15 HERE

The verb ‘be’, whether used as a copula or to express existence or location, has a suppletive
paradigm, with the stem ʤìH (most probably cognate with ʤìlɔ̀ ‘stay, live’) in the Present
indicative and bɛ́ in the other verb forms, including non-indicative moods.

Imperfective verb forms are construed by means of an auxiliary and the stem of the lexical
verb, as shown in Table 16. The last column is again an example with the verb dàá mbɔ́
‘cook’.

PST4 (SM-)ʤìH/yìH-mɛ STEM àʤìmɛ́ dàá mbɔ́

PST3 (SM-)ʤìH/yìH kòò STEM àʤì kóò dàá mbɔ́
PST1/2 H
-mɛ̀-yìH STEM pʰàâm mɛ́yì dàá mbɔ́
SM-H-àà-yìH STEM àáyì dàá mbɔ́
PRS (SM-)yɔ̀kùH STEM àyɔ̀kù dàá mbɔ́
(SM-)ʤìH/yìH STEM àyì dàá mbɔ́
FUT1 (SM-)fɛ̀ɛĹ -bɛ́ SM-ʤìH/yìH STEM àfɛ̀ɛ ́ bɛ́ àʤì dàá mbɔ́
FUT2 (SM-)ʃɔ̀ɔ-̀ H-bɛ́ SM-ʤìH/yìH STEM àʃɔ̀ɔb̀ ɛ́ àʤì dàá mbɔ́
FUT3 (SM-)ʃɔ̀ɔ-̀ ŋgɛ̀-H-bɛ́ SM-ʤìH/yìH STEM àʃɔ̀ɔŋ̀ gɛ̀bɛ́ àʤì dàá mbɔ́
Table 16: Kwakum tenses: Imperfective forms

INSERT TABLE 16 HERE

Interestingly, past imperfective forms involve the Present tense stem ʤìH of the auxiliary
‘be’ with past tense morphology. In some TA-forms this stem has the reduced form yìH,
suggesting that it is prosodically evolving towards affix status. In the Present tense, the

39
auxiliary yìH/ʤìH is used when the verb is in focus, the ʤìH-form being obligatory with
nominal subjects. When the verb is not in focus, the auxiliary yɔ̀kùH is used. Imperfective
forms in the future tenses involve a succession of two ‘be’ auxiliaries. The first one is
inflected for tense and may take a pronominal or nominal subject. The second one is the
Present tense stem preceded by a subject pronoun.

6.3.2 Non-indicative moods

The three non-indicative moods of Kwakum - Imperative, Subjunctive and Consecutive - all
involve non-basic allomorphs of verb stems. The Imperative singular has no extra marking,
except with CV-stems, where it takes the suffix –kɛ́L (81). The suffix of the Imperative
plural is -kínL (82). Two verbs have a suppletive Imperative form (83).

(80) Imperative verb forms

a. CV-kɛ́L (SG)

b. stem (sg)
c. STEM-kínL (PL)

(81) bàkɛ́ bùpà

bà-kɛ́L bùpà
cut_up.IMP-2SG.IMP 1.meat
‘Cut up (SG) the meat!’

(82) bàkín bùpà

bà-kínL bùpà
cut_up.IMP-2PL.IMP 1.meat
‘Cut up (PL) the meat!’

(83) a. ʤìH ‘eat’  ʤìkɛ́ ⁓ ʤòkɛ́ ‘Eat!’

b. ʃèH ‘come’  ʃɔ̀ɔ ́ kɔ́ ‘Come!’

The Subjunctive mood is used in subordinate clauses and to express hortative modality (85).
It is formed by means of a non-basic allomorph of the verb stem and a subject prefix (84).

(84) the Subjunctive verb form

40
 (sm-)stem

(85) ʃɛ̀báà ìtàá mbɔ́

ʃɛ̀H-báà ì-tàámbɔ́L
1PL-set.SBJV 5-trap
‘Let us set the trap.’

The Consecutive mood is formed by means of the prefix n- and a non-basic allomorph of the
verb stem (86).

(86) the Consecutive verb form

(SM-)n-STEM

It has a wide array of uses. In its consecutive use it functions as a relative tense form
indicating that an event takes place after a previously mentioned event. Second, it can
function as a general present, typically in proverbs (87). Third, it can be used in either or
both the apodosis and the protasis of conditional clauses. Finally, it can be used in
subordinate clauses instead of the Subjunctive.

(87) mòmyàà mbɛ́ tʰà kìbàpù

L%
mó-myaa N-bɛ́L tʰà kì-bàpù
1-woman CSC-COP.CSC CMP 7-horsefly
‘A woman is like a horsefly.’

6.3.3 Negation

The negative form of the verb ʤìH ‘be’ is fètɛ́ɛ́ tɛ́ɛ ́ (88).

(88) pèŋgɛ̀ fètɛ́ɛ ́ tʃàk bɔ́ɔ ́ ʃɛ́ŋ

pèŋgɛ̀ fètɛ́ɛĹ tʃàkí bɔ̀ɔʃ-ɛ́ŋ
1.money NEG.COP 1.thing be.bad-ADJ
‘Money is not a bad thing.’

In indicative verb forms, the negative marker is wɛ́ɛ.́ It originates in a third person singular
possessive pronoun, a path of grammaticalisation that is not unusual in the Bantu languages

41
(Devos and van der Auwera 2013). When the subject is a first person pronoun, the 1SG
possessive pronoun wàá mbɔ́ can be used as well and some speakers use the 2SG possessive
pronoun wɔ́ɔ ́ with 2SG subjects. In the present and future tenses the negative marker is in
between the Tense marker(s) and the stem (89). In the past tenses it follows the stem of the
conjugated verb immediately (90).

(89) àfɛ̀ɛw
́ ɛ́ɛd́ ʒì búpà
à-fɛ̀ɛĹ -wɛ́ɛĹ -dʒìH bùpà
3SG-FUT1-NEG-eat 1.meat
‘He will not eat meat.’

(90) àádʒìwɛ́ɛ ́ bùpà

à-H-àà-dʒìH-wɛ́ɛĹ bùpà
3SG-PST2-PST2-eat.PST2-NEG 1.meat
‘She did not eat meat.’

In non-indicative moods, negation is marked by means of the marker bɛ̌k (91).

(91) bɛ̌kdʒì búpà

bɛ̌kL-dʒìH bùpà
NEG-eat.2SG.IMP 1.meat
‘Don’t eat meat!’

7 CLAUSAL SYNTAX

7.1 Simple clauses

As in most Bantu languages, the subject precedes the verb and the object follows it. In
clauses with two unmarked complements the Goal precedes the Theme (92), an order that
can be optionally reversed if and only if the Theme is pronominal and the Goal a noun with
human reference (93). Both the Theme and the Goal are accessible to relativisation, but only
the Theme is accessible to passivisation.

(92) pʰàâm ḿ fɛ́ mɔ́ɔǹ á máŋgòlò

42
 pʰàám̀ H
-n-fɛ́ mɔ̀-ɔ̀nɔ́ à-máŋgòlò

1.man PRS-PRS-give 1-child 2-mango

‘The man is giving the child mangoes’

(93) pʰàâm ḿ fɛ́ já mɔ́ɔǹ ɔ́

pʰàám̀ H
-n-fɛ́ jàH mɔ̀-ɔ̀nɔ́

1.man PRS-PRS-give 3PL 1-child

‘The man is giving them to the child.’

7.2 Questions

Polar questions are marked intonationally by a rising tone on the last syllable of the
sentence. Information questions are formed by replacing the questioned constituent by an
interrogative, such as kè ‘what, why’ or tà ‘who’, which remains in situ (94). Alternatively,
interrogatives can occur in clause initial position in a focus construction.

(94) gwɛ́ ndàámbú kè?

H
-gwɛ̀ n-dàámbɔ́L kè
PRS-2SG PRS-cook what
‘What are you cooking?’

7.3. Subordinate clauses

Subordinate clauses are optionally marked by means of the clause-final subordinator jí.
Relative clauses can additionally be marked by means of a connective relator, which may or
may not be preceded by the nominaliser-linker mòò/gwòòm (95a). The connective relator
that introduces relative clauses is not an agreement target. Alternatively, relative clauses
with a plural head noun can be introduced by the copula ʤìH (95b). Since both the
subordinator and the relativiser are optional, relative clauses can be totally unmarked (96).
All positions in the clause are accessible to relativisation.

43
(95) a. àpʰàâm (gwóòm) bá kóò bùpà (yí)
à-pʰàâm H
-gwòòm bàH
H-
kòò bùpà yíL
2-man CON-NLNK.PL CON-cut_up.PST3 PST3 1.animal SUB

b. àpʰàâm dʒì bá kóò bùpà (yí)

à-pʰàâm ʤìH bàH kòò bùpà yíL
2-man COP cut_up.PST3 PST3 1.animal SUB

‘the men who cut up the meat’

(96) àpʰàâm bà kóò bùpà

à-pʰàâm bàH kòò bùpà
2-man cut_up.PST3 PST3 1.animal
‘the men who cut up the meat’ or ‘The men cut up the meat.’

Complement clauses are introduced by the complementiser nàáL, which has an initial í
when introducing a clause functioning as subject.

(97) ínàá pùlɔ̀ mɛ́ nɔ́ búláàwɛ̌ (yí) ńdɛ̌lkáà ɲì

ínàáL pùlɔ́ H
-mɛ̀-nɔ́ bùl-áàwɛ̀H yíL n-dɛ̌lkáà ɲì
COMP 1.rain PST2-PST2-fall.PST2 be_numerous-ADJ SUB PRS-surpass 1SG
‘It surprises me that it rained so abundantly.’

REFERENCES

Bastin, Yvonne & Coupez, André & Mumba, Evariste & Schadeberg, Thilo C. (eds.). 2002.
Bantu Lexical Reconstructions 3. Tervuren: Royal Museum for Central Africa, online
database: linguistics.africamuseum.be/BLR3.html (last accessed November 28, 2017).
Belliard, François. 2005. Instruments, chants et performances musicales chez les Kwakum
de l’arrondissement de Doume (Est-Cameroun). Etude ethnolinguistique de la conception
musicale d’une population de langue bantoue A91. Paris: Université Paris 7; LLACAN.
Belliard, François. 2007. Parlons kwàkúm, langue bantu de l’Est Cameroun : langue et
culture. Paris: L’Harmattan.
Devos, Maud & van der Auwera, Johan. 2013. Jespersen Cycles in Bantu: Double and
Triple Negation. Journal of African Languages and Linguistics 34(2). 205-274.

44
Njantcho, Elisabeth. forthcoming. A Grammar of Kwakum. Paris: INALCO.

45