TS507-2008 Relative Clauses In Kikamba

PDF version: 08ts507c-rowbory-relative-clauses-kikamba

Relative Clauses in Kikamba

David Rowbory • 20 June 2008

Nairobi Evangelical Graduate School of Theology: Advanced Morphology & Syntax

Introduction

Kikamba is classified by Guthrie (1948) as Bantu language E55, related quite closely to Kikuyu. See Rowbory (2008) for more detail. Here we study morphological operations used in Kikamba to relativise nominals so as to produce a multi-clause (multi-predicate) sentence. The study has been somewhat hindered by many writers (whether casual writers, or in books such as Mbiti 1966) neglecting to use the full orthographic means to distinguish tone and some vowels. So, what should be written ĩ representing IPA [e] is often not distinguished from i (IPA [i]), though there is a definite phonological difference. The situation is similar with u / ũ (IPA [o]). The standard word order is SVO.

We began this study not with elicitation but by examining some texts from transcribed speech and translated stories and it became clear that Kikamba makes frequent use of restrictive relative clauses. However we found no unambiguous evidence of non-restrictive relatives, which merely comment on a noun phrase rather than delineating its reference. So for this paper we narrow our focus to treat only restrictive relative clauses, following Andrews (2006:207) who considers non-restrictive relatives to behave somewhat differently to relative clauses. Although Andrews considers questions and adverbial clauses to fall outside the scope of a typology of relative clauses, we touch on these briefly here since they seem to be related to relative clauses in Kikamba.

We explore the most common and obvious relativising strategy, the situations in which it may be used, and other relativisation strategies. We examine the grammatical categories (subject, direct object, oblique, indirect object etc) which may be relativised, and briefly mention the use of relative clauses in questions and the related use of headless relative clauses.

Natural texts were sourced from Mbiti (1966) and an interview with an elderly man recorded and conducted by a fellow student Michael Mwaka. Subsequent clarification, translation and elicitation was carried out with the help of Michael and our house help.

The Most Common Strategy: -la

The most common strategy for relativising noun phrases in Kikamba is to use the relativiser la with a noun class prefix corresponding to the relativised noun phrase:

(1)

Kuya

kitwii

mundu

ula

wambie

utangiiwa

mbee

ni

chief.

kuya

kitwii

mu-ndu

u-la

w-amb-i-e

u-tangiiwa

n-vee

ni

chief.

In

Kitwii

1-person
(NPmat)

1-rel
(NPrel)

1-become-perf-fv

14-saved

9-first

was

chief.

“In kitwii, the man who got saved first was a chief.”

(2)

Twineena

iulu

wa

musimamo

wa

ikanisa

iulu

wa

iveti

ili

na

mathina

tu-i-neen-a

i-ulu

wa

mu-simamo

wa

i-kanisa

i-ulu

wa

i-veti

i-li

na

ma-thina

1p-talk-fv

8-matter

of

3-position

of

5-church

8-matter

of

8-wives

8-two

and

6-problems
(NPmat)

“We are now talking about the position of the church on two wives and the problems…”

 

ala

mookie

nundu

wa

musimamo

usu

wa

ikanisa

a-la

ma-ok-i-e

nundu

wa

mu-simamo

usu

wa

i-kanisa

6-rel
(NPrel)

6-arise-perf-past2

because

of

3-position

that (demonst)

of

5-church

“…which arose because of that position of the church.”

 

In the first example (1) above, the subject “man” of the matrix clause “In Kitwii the man… was a chief.” is class 1 (see class list in the appendix). It is relativised with ula in the relative clause ula wambie utangiiwa mbee “who got saved first” where the relativised noun phrase (NPrel) is also the subject of the verb wambie. The relativiser la takes a prefix u- corresponding to class 1 (human, singular). This combination could be considered a relative pronoun, though it never inflects for anything except the class of the relativised noun phrase. So it behaves quite differently from English relative pronouns which inflect for the grammatical category of the NPrel. In further examples below we see that the same form can be used even when the NPrel is the direct object, indirect object, oblique etc. in the relative clause.

The relativiser appears directly after the noun phrase in the matrix clause which is being relativised (NPmat), that is, the noun phrase which is coreferential with the NPrel. The relativiser (rel) marks the start of the NPrel but no morphology indicates the end of the NPrel; that must be deduced from context. In both examples above the relative clause is indicated in bold. We note in (1) that the relative clause is embedded within the matrix clause, but the relative clause in example (2) is postposed to the main clause, since the NPmat is an object (which comes normally at the end of the main clause). However in example (2) the NPrel functions as the subject in the relative clause.

One surprising observation of the relativiser’s concord marking is that when the NPmat is class 6 (prefix ma- for nominals and verb subject marking) the rel takes the prefix a- rather than ma-. However, given in class 2 the prefixes a- and ma- both occur in verb subject marking, this may just be a variant class marker which Kioko (2005:22,49) has not included. In the section on multiple-embedding below, examples (12) and (13) show that the class of the NPmat governs the relativiser’s prefix, but other evidence suggests that the relativiser may take slightly different prefixes:

(3)

Ngombe

ila

yina

mukwato

ni

yiva?

Ø-ngombe

i-la

yi-na-Ø

mukwato

ni

yi-va

9-cow

9-rel

9-pres-be

on.heat

qmark

9-which.of

“Which cow is on heat?”

(4)

Ngombe

ila

syina

mukwato

ni

syiva?

Ø-ngombe

i-la

syi-na-Ø

mukwato

ni

syi-va

10-cow

10-rel

10-pres-be

on.heat

qmark

10-which.of

“Which cows are on heat?” (a plural number expected)

This is not very surprising, since nominals and verb subjects use slightly different sets of prefixes. Class 9 nominals often have a null prefix, but just like a verb, the relativiser requires a non-null prefix. The question words yiva and syiva takes the same class prefix as the verb, but in each case the prefix for the relativiser is i-. (However in this class the nominals share the same null prefix for singular and plural, and it is not so surprising that the relativiser prefix reflects this invariance.)

Non-Subject Grammatical Roles

This –la relativising process is very productive and can apply to noun phrases throughout the continuum of grammatical roles from subject to genitive. That is, the NPmat (noun phrase co-referential with the NPrel in the matrix noun phrase) may exhibit any grammatical function in the matrix clause and the corresponding NPrel may fit into any grammatical role.

When the NPrel is a direct object in the relative clause, the main clause and relative clause behave similarly to when the NPrel is subject:

5)

Nininaiye

liu

ula

kiveti

kyakwa

kinauie

Ni-ni-Ø-nai-i-e

Ø-liu

u-la

ki-veti

ky-akwa

ki-Ø-nau-i-Ø-e

foc-1s-past2-eat-perf-fv

3-food

3-rel

7-wife

7-1s.poss

7-past2-cook-perf-do-fv

“I ate the food which my wife cooked.”

6)

Ninaiye

kila

kiveti

kyakwa

kinauite

Ni-ni-Ø-nai-i-e

ki-la

ki-veti

ky-akwa

ki-Ø-nau-i-t-e

foc-1s-past2-eat-perf-fv

7-rel

7-wife

7-1s.poss

7-past2-cook-perf-do-fv

“I ate what my wife cooked.”

Example (5) shows a simple matrix clause where the direct object liu has been relativised. Somewhat surprisingly the relativiser takes the prefix u-, though this cannot be class 1 since that class is for humans, so we deduce it is class 3 like liu which apparently has a null class marker. Food is generally class 3, together with ‘tree’ since so much food comes from trees. Example (6) demonstrates a headless relative clause which we discuss below. Aside from the presence of liu ula the NPmat and NPrel, the other difference between these two examples of a direct object NPrel is a resumptive pronoun suffix –t on the relative clause verb in (6) which does not appear in (5). So either it may be that where a headless relative is used a resumptive pronoun is required. However since liu has a null class prefix, it seems reasonable that whether headless or not, a direct object NPrel requires a resumptive pronoun before the final vowel.

Obliques (such as locative, temporal) and indirect objects are sometimes hard to distinguish in Kikamba and can be relativised with –la too:

7)

Ikanisa

ila

mathina

maumilile

yaii

kitwii

i-kanisa

i-la

ma-thina

ma-u-m-i-il-il-e

ya-i-i

kitwii

5-church

5-rel

2-problem

2-?-2.refl-past4-arise-iter-fv

9-past4-be

kitwii

“The church in which the problems arose was in kitwii.”

8)

Nyumba

ila

mundu

usu

wekalaa

yaii

nini

Ø-nyumba

i-la

mu-ndu

usu

w-e-kala-a

ya-i-i

Ø-nini

9-house

9-rel

1-person

that

1-past4-live-fv

9-past4-be

9-small

“The house where that man had lived was small.”
(Note: The sense of ‘in’ seems to disappear or be absorbed into the ‘relative pronoun’.)

9)

mundu

usu

wekalaa

nyumbani

nini

mu-ndu

usu

w-e-kala-a

Ø-nyumba-ni

Ø-nini

1-person

that

1-past4-live-fv

9-house-in

9-small

“That person lived in a small house.”

(10)

Nyumba

ila

mundu

usu

wekalaa

niyavalukie

Ø-nyumba

i-la

mu-ndu

usu

w-e-kala-a

ni-ya-Ø-valuk-i-e

9-house

9-rel

1-person

that

1-past4-live-fv

foc-9-past2-collapse-perf-fv

“The house where that man had lived collapsed.”

These four examples demonstrate various ways in which a a subject nominal from the matrix clause can be relativised as a locative indirect object or oblique. Example (9) contrasts with (8) revealing that the expected location suffix –ni is lost in the relative clause. It may be implied by the class 9 prefix to the relativiser ila which is associated with a place. There is nothing distinctive about the final vowels or any other feature of the verb in the relative clause. Unsurprisingly for a restrictive relative, the time of the relative clauses is either simultaneous with the matrix verb time or precedes it in these examples. (Past 1 to 4 move back from recent to distant time.) The analysis of the somewhat complex verb in (7) may be insufficient, but does not seem to contradict this relationship of temporal grounding of the verbs.

Not only can NPmat subjects be relativised in this way, but objects (and likely all other roles) can be relativised as locative obliques or indirect objects:

(11)

Ninyie

naakie

nyumba

ila

mundu

usu

wekalaa

ni-Ø-ny-i-e

n-Ø-aak-i-e

Ø-nyumba

i-la

mu-ndu

usu

w-e-kala-a

1s-past2-build
-perf-fv

1s-past2-build
-perf-fv

9-house

9-rel

1-person

that

1-past4-live-fv

“I built the house where that person lived.”

Here, assuming the tense analysis of the verb pair Ninyie naakie is correct, it does seem surprising that wekalaa ‘live’ is grounded in an earlier time.

In all of these examples the NPmat is identified unambiguously regardless of grammatical role since the relativiser immediately follows it, and takes a matching prefix.

Multiple-embedding

The main clause may include more than one relative clause as appropriate. Often a question (see below) might involve a relative clause, but as in the examples below, another part of the question might include a relative clause. As the two following examples show, it is perfectly reasonable to embed multiple relative clauses in one matrix clause, with the relativisers taking the appropriate (different) class-concord prefixes:

12)

Andu

ala

maii

iveti

ili,

maundu

ala

mataiimeka

ikanisani

ni

mau?

a-ndu

a-la

ma-ii

i-veti

i-li

ma-undu

a-la

ma-t-aii-m-ek-a

i-kanisa-ni

ni

mau?

2-person

2-rel

2-past4

8-wives

8-two

6-thing

6-rel

2-neg-permit
-2.refl
-do-fv

5-church-in

qmark

what?

“Those people who had two wives, what things could they not do in the church?”

13)

Mundu

ula

maii

iveti

ili,

maundu

ala

utaiieka

ikanisani

ni

mau?

mu-ndu

u-la

wa-i-i

i-veti

i-li

ma-undu

a-la

u-t-aii-Ø-ek-a

i-kanisa-ni

ni

mau?

1-person

1-rel

1-past4
-
have

8-wives

8-two

2-thing

2-rel

1-neg-permit
-1.refl-do-fv

5-church-in

qmark

what?

“This person who had two wives, what things could they not do in the church?”

Although the forms of the relativiser are identical in the first example, in the second the change of matrix clause subject (which is also relativised) brings a change in the prefix for the first relativiser. The second relativiser operates on ‘things’ maundu so is unaffected whether the matrix subject is plural or singular because it is relativising the complement of an implied ‘be’ verb, possibly incorporated in the question word mau.

Other Strategies in brief

Three other relativising strategies are evident: omitting the relativiser, using a copular with nominalised verb and using a demonstrative. It seems that certain strategies are preferred by particular verbs or combinations of grammatical role in the NPmat and NPrel.

Null Relativiser

When the NPmat follows the verb (for example as a direct object) the relativiser is frequently omitted:

14)

Ninisi

mundu

wai

na

nzwii

nene

muno

Ni-ni-s-i

mu-ndu

Ø

wa-i-Ø

na

n-zwii

n-ene

muno

1s-pres-know-perf

1-person

rel

1-past4-be

with

9-hair

9-long

very

“I know a man who has very long hair.”

15)

Kamundu

kai

na

nzwii

nene

muno

Ka-mu-ndu

ka-i-Ø

na

n-zwii

n-ene

muno

12-1-person

12-past4-be

with

9-hair

9-long

very

“A small man had very long hair.”
(Class 12 is a diminuitive when prefixed to an existing
class 1 noun with prefix.)

16)

Tene

muno

nduani

imwe

vai

kamundu

kai

na

nzwii

nene

muno.

tene

muno

n-dua-ni

Ø-imwe

va-i-Ø

ka-mundu

ka-i-Ø

na

n-zwii

n-ene

muno

past.time

very

9-village-in

9-one

16-past4-be

12-man

12-past4-be

with

9-hair

9-long

very

“A very long time ago, in one of the villages there lived (was) a small man
who had very long hair.”

If we compare (13)(14) and (15), which are elicited extracts of (16) we see in the place after mundu where we would expect ula there is nothing before the verb in the relative clause. Were it not for the longer example (16) we might suspect that the verb ‘know’ was merely taking a complement or that this is two sentences. But here we see the same structure as for a main clause, with matrix and relative clauses sharing kamundu. In the matrix clause it is a direct object, but in the relative clause a subject. The unusual focus of the sentence – not making the man the subject, but introducing a dummy ‘place’ subject “there was…” – may be a result of rearranging the sentence so that kamundu can be in the right position to function in both the matrix and the relative clause.

Copular and nominalised verb

Stative-type verbs such as ‘marry’ or ‘be given in marriage’ may be particularly susceptible to nominalisation. When such verbs (which in nominalising have become like an adjective) are combined with the copular uu something equivalent to a restricted relative clause.

17)

Mataii

na

undu,

ningi,

mundu

muka

uu

utwaitwe

umunthi

uu

tikau

Ma-t-a-i-i

na

Ø-undu,

n-ingi,

mu-ndu

mu-ka

uu

u-twait-w-e

u-munthi

uu

ti-kau

2-neg-be
-perf-FV

with

9-problem

9-no

1-person

1-female

cop

1.nom-marry
-pass-past

14-today

cop

it.is-fight

“They had no problem, this woman who has been married, today it’s… it’s a fight.”

18)

Mundu

muka

uu

nutwaitwe.

Mu-ndu

mu-ka

uu

n-u-twait-w-e

1-person

1-female

cop

foc-1.nom-marry-pass-past

“This woman is married.”

Comparing examples (17) and (18), the major difference is the prefix n- in (18) which may indicate focus. The nominalised verb ‘marry’ combines with the copular to act as a restrictive relative clause in exactly the same position as a –la relative clause would be expected. The copular with nominalised verb requires a verb subject marker, coreferential in these cases with ‘the woman’ who is also subject in the matrix clause. A transitive verb or an interansitive verb with oblique may also be used. The NPrel is not restated except being referred to in the relative verb subject marker:

(19)

Ninaneena

na

mundu

muka

uu

mutwae

ni-na-neen-a

na

mu-ndu

mu-ka

uu

mu-twa-e

1s-past1-talk-fv

with

1-person

1-female

cop

1.nom-marry-fv

“I talked with a woman who is married.”

(20)

Ninaneena

na

mundu

muka

uu

utwaitwe

ni

fugoyo

ni-na-neen-a

na

mu-ndu

mu-ka

uu

u-twait-w-e

ni

fugoyo

1s-past1-talk-fv

with

1-person

1-female

cop

1.nom-marry-pass-fv

with

fugoyo

“I talked with a woman who is married to Fogoyo.”

This strategy is most usually observed when the NPrel functions as subject in the relative clause and NPmat functions as object, so that the nominals are suitably positioned.

Demonstrative

The last and somewhat rare strategy is to use a demonstrative pronoun where we might expect a relativiser. Similarly the demonstrative takes a class-concord prefix:

(21)

Kwoou

kila

mundu

niwonaa

kyeni

kyu

twitaa

utisi.

Kuwoou

kila

mu-ndu

ni-w-ona-Ø-a

ky-eni

ky-u

tu-ita-Ø-a

u-tisi

Therefore

every

1-person

foc-1-see-cont-fv

7-light

7-that

1p-call-cont-fv

14-lightning

“Therefore, everyone sees the light which we call lightning.”
(Mbiti 1966:227)

My informants insisted that this is not two sentences, but that the relative clause is only correct when in the context of the main clause. This example from the end of a fable could possibly be a case of a non-restrictive relative, but in its context, explaining why people see lightning the relative clause seems most likely to restrict kyeni ‘light’ rather than just supplying additional information. That is, the story is about the light called ‘lightning’, not light called ‘sunlight’.

Etymology of the Relativiser

Although relativisers (such as English ‘that’) often are related to demonstratives, the common Kikamba relativiser looks very different from all the demonstratives, so it seems unlikely that it has derived at all recently from a demonstrative. Kila in the above example my informants insisted was not ki-la but a simple, non-inflecting adverbial. However it bears some resemblance to the relativiser so the relationship between these words may be worth exploring.

Headless Relative Clauses and Questions

The la strategy is used for headless relative clauses:

(22)

Ninaiye

kila

kiveti

kyakwa

kinauite

Ni-ni-Ø-nai-i-e

ki-la

ki-veti

ky-akwa

ki-Ø-nau-i-t-e

foc-1s-past2-eat-perf-fv

7-rel

7-wife

7-1s.poss

7-past2-cook-perf-do-fv

“I ate what my wife cooked.”

The generic class 7 ki coincidentally has the same class as kiveti but there is no occurrence of the NPrel in the matrix clause, so this is a headless relative clause. Otherwise it is just the same as other –la constructions. Where the NPmat is obvious or non-referring (vague), a headless relative clause uses the clause itself to specify the omitted NPmat.

Headless relative clauses are sometimes (but not exclusively) used in questions, with additional question-marking words:

(23)

Ngombe

ila

yina

mukwato

ni

yiva?

Ø-ngombe

i-la

yi-na-Ø

mukwato

ni

yi-va

9-cow

9-rel

9-pres-be

on.heat

qmark

9-which.of

“Which cow is on heat?”

A question mark ni always appears as the penultimate word before the question word which is linked to the relativiser, but also specifies the type of question ‘who’, ‘what’, ‘why’, ‘when’ etc.

Multiple embedding

Relative clauses may also be embedded in questions so that la occurs twice with slightly different functions and often quite different translations of la into English:

(24)

Musungu

ndaii

na

mavityo,

nundu

maandiko

measya

Mu-sungu

n-ti-a-ii

na

ma-vityo,

nundu

ma-andiko

me-a-sya

1-european

1-neg-be-past4

with

2-wrong

because

6-words

6-say-past

“The white man was not wrong because the scriptures say…”

 

ila

kila

kianiu

kiuka,

kila

kitekianiu

kikavetwa.

i-la

ki-la

ki-aniu

ki-uka,

ki-la

ki-te-ki-aniu

ki-ka-vet-w-a.

5-rel

7-that

7-perfect

7-come

7-that

7-neg-7.refl-is.perfect

7-fut-take.away-pass-fv

“when the thing that is perfect comes, the thing that is imperfect will pass away.”

However in terms of Kikamba grammar this multiple use of the same form is unsurprising, since unlike English the relativiser complex is not inflecting for grammatical role. The second line is simultaneously a complement of a speech verb (measya) and a time relative (‘when’) clause. Within the time clause are two headless relative clauses, using the generic class marker ki-la to as a non-referring restrictive relative clause.

Question markers and the relativiser

We analyse ni mau as a question marker followed by ‘what’. However, it is possible that ni is more of a question form of the copular verb and it is clear that mau (which does not inflect) at most specifies the type of question (‘what’, ‘when’, ‘why’ etc) and so only in conjunction with the preceding la is it equivalent to the English translation ‘what’. This is outside the scope of this paper.

Conclusions

The major strategy for relativising nominals of a main clause is to use the relativiser –la with a concord prefix agreeing with the class of the NPmat. As in a main clause, if the NPrel is subject, the verb of the relative clause requires the appropriate class-concord prefix. It appears any grammatical role within the main clause can be relativised in this way, and there is evidence that the NPrel may take any grammatical role in the relative clause except a possessive/genitive. If the NPrel is a direct object in the relative clause then a resumptive pronoun may be added as a suffix to the relative clause verb root. Where the NPrel is an indirect object or oblique, a resumptive class-concord prefix may appear in the usual place for an indirect object within the relative clause verb. ‘What’, ‘When’ and ‘Who’ questions may use one of the relativising strategies – most often using –la. The relativiser is quite different from the common demonstrative usu so would not appear to be related to that. We were unable to find concrete clues as to its etymology.

Bibliography

Andrews, Avery D (2006) Relative clauses, in Shopen, Timothy (ed) Language Typology and Syntactic Description, Cambridge: Cambridge University Press, pp 206-236

Guthrie, Malcolm (1948) Classification of the Bantu Languages, London: Oxford University Press for the International African Institute
summarised at http://www.linguistics.berkeley.edu/CBOLD/Lgs/LgsbyGN.html

Guthrie, Malcolm (1967) Comparative Bantu, Farnborough: Gregg Press Ltd

Kioko, Angelina Nduku (2005) Theoretical Issues in the Grammar of Kikamba, München: LINCOM GmbH

Mbiti, John S. ed. (1966) Akamba Stories, Oxford: Clarendon Press / 1983 Nairobi: Oxford University Press

Payne, Thomas E (1997) Describing MorphoSyntax, Cambridge: Cambridge University Press

Rowbory, David (2008) Complements of Speech in Kikamba, Unpublished paper


    

Note that orthographic e, o represent IPA [ɛ], [ɔ]

Notes on the interpretation of the interlinear text:
The permissive sense here may be difficult. It seems the person being permitted is
embedded as a reflexive indirect object. Other forms of the verb shedding light on the permissive ‘mood’ are here:
They could do: m[a]-aii-m-ek[a]-a He could not do: u-t-aii-Ø-ek[a]-a
They did
not do: ma-t-eka-a

Notes on the interpretation of the interlinear text:
(Utwaitwe is clearly derived from a verb root -twait- and takes a verb-style subject
marker u- but here appears to be nominalised and used with the copular
uu.)

When the thing which is perfect comes, the thing which is imperfect will be removed.
(i- is used for a relative of time, so that i-la = “when”)

 

Tags: , , ,

One Comment