+ All Categories
Home > Documents > Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz

Date post: 11-Oct-2016
Category:
Upload: colleen
View: 216 times
Download: 1 times
Share this document with a friend
45
Noun incorporation and predicate classifiers in Gumuz 1 COLLEEN AHLAND JALL 31 (2010), 159–203 0167–6164/10/031-00159 DOI 10.1515/JALL.2010.008 ©Walter de Gruyter Abstract Gumuz is a Nilo-Saharan language spoken in western Ethiopia and south- eastern Sudan. Much dialectal variation exists among the Gumuz but all lects within Ethiopia show evidence of noun incorporation, including that of Gilgel Beles, the focus of this paper. In the past, this noun incorporation in verbs had been overlooked and the verb-noun morpheme pair had been analyzed as one morpheme with infixing inflectional morphology. Thus, verbal ‘roots’ without incorporated nouns had been described as suffixing vs. those with incorpo- rated nouns as infixing verbs (Irwin 1966: 5), or unsplit vs. split (Uzar 1989: 371). Upon further investigation, it has become evident that these putative split or infixing verbs do not comprise merely a single root after all. Rather, these verbs have incorporated nouns which create a complex verbal stem. Such in- corporated nouns serve many functions in Gumuz and, in some cases, they have grammaticalized as classifiers similar to the process described by Mithun (1986:385). To date, no system of predicate classifiers has been documented in the lan- guages of Africa, let alone Ethiopia. This paper argues for the existence of such a system in Gumuz and describes the extent to which these classifiers are a productive part of the verbal morphology. 1. This research has been partially funded by the National Science Foundation award #0723404. Special thanks to Gumuz language consultants Work’u Mekash, Lemmi Wendu, Moges Andi and my SIL colleagues in Ethiopia for their help in this research as well as for invaluable feed- back from Doris Payne, Eric Pederson, Spike Gildea, Michael Ahland and two anonymous reviewers regarding the final drafts of this paper. Brought to you by | University of Virginia Authenticated | 128.143.23.241 Download Date | 9/22/12 8:23 PM
Transcript
Page 1: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiersin Gumuz1

COLLEEN AHLAND

JALL 31 (2010), 159–203 0167–6164/10/031-00159DOI 10.1515/JALL.2010.008 ©Walter de Gruyter

Abstract

Gumuz is a Nilo-Saharan language spoken in western Ethiopia and south-eastern Sudan. Much dialectal variation exists among the Gumuz but all lectswithin Ethiopia show evidence of noun incorporation, including that of GilgelBeles, the focus of this paper. In the past, this noun incorporation in verbs hadbeen overlooked and the verb-noun morpheme pair had been analyzed as onemorpheme with infixing inflectional morphology. Thus, verbal ‘roots’ withoutincorporated nouns had been described as suffixing vs. those with incorpo-rated nouns as infixing verbs (Irwin 1966: 5), or unsplit vs. split (Uzar 1989:371).

Upon further investigation, it has become evident that these putative splitor infixing verbs do not comprise merely a single root after all. Rather, theseverbs have incorporated nouns which create a complex verbal stem. Such in-corporated nouns serve many functions in Gumuz and, in some cases, theyhave grammaticalized as classifiers similar to the process described by Mithun(1986:385).

To date, no system of predicate classifiers has been documented in the lan-guages of Africa, let alone Ethiopia. This paper argues for the existence ofsuch a system in Gumuz and describes the extent to which these classifiers area productive part of the verbal morphology.

1. This research has been partially funded by the National Science Foundation award #0723404.Special thanks to Gumuz language consultants Work’u Mekash, Lemmi Wendu, Moges Andiand my SIL colleagues in Ethiopia for their help in this research as well as for invaluable feed-back from Doris Payne, Eric Pederson, Spike Gildea, Michael Ahland and two anonymousreviewers regarding the final drafts of this paper.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 2: Noun incorporation and predicate classifiers in Gumuz

160 Colleen Ahland

1. Introduction

Predicate classifiers (or verbal classifiers) are classifiers found in the verb stem.These classifiers do not classify the verb itself but rather an argument of theverb (Grinevald 2000: 67). While classification systems are known to existin Africa, a system of classification in which an incorporated noun/verbal af-fix classifies an argument of the predicate is virtually unknown on the conti-nent. Previous literature on predicate classifiers only includes languages of theAmericas and Australia (Mithun 1986, Grinevald 2000). In addition, Aikhen-vald further asserts that “there are no verbal classifiers in the languages ofAfrica or Eurasia or in the Austronesian family” (2000: 171). Thus, such asystem is not documented and at best rare in Africa, let alone Ethiopia whereGumuz is spoken.

The Gumuz language varieties exhibit a fair degree of variation but all lectsbehave similarly in terms of noun incorporation (NI) and predicate classifica-tion. However, the focus of this paper will be on the form of Gumuz spokenin and around the town of Gilgel Beles, north of the Blue Nile. Occasionally,reference will be made to other Gumuz lects where relevant.

Focus will also be on the following nominal roots/suffixes: -Vk’w ‘head’,-Vts ‘body’, -Vc ‘eye/seed’, -k’wós ‘tooth’, and -ts’ê ‘ear’ which in certainverbs exhibit a classificatory function. Other noun roots/suffixes of the sameposition class within the verb will be mentioned to demonstrate the variety offunctions of these morphemes.

1.1. Outline of the article

Section 1 of this paper includes previous analyses of Gumuz verbal morphol-ogy, background information about the Gumuz language, its structure, andpertinent typological patterns as well as the more general phenomenon of NIin Africa. Section 2 defines NI and demonstrates its behavior in the Gumuzlanguage, synchronically as well as diachronically. As NI historically leadsto a system of predicate classifiers in Gumuz, it is important both to estab-lish its existence as well as to describe its various functions in the language.Section 2 also discusses the effects of NI on verbal valence. Section 3 in-troduces two current typologies of predicate classifier systems and describesthe classifier system that exists in Gumuz. Section 4 demonstrates the behav-ior of predicate classifiers in Gumuz with regard to two syntactic operations:promotion of arguments and coordination of O arguments. Both operationslend further support to the existence of such a classifier system. Section 5discusses and refutes a possible alternative analysis for the Gumuz predicateclassifier system, while Section 6, places the system within the two previ-

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 3: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 161

ously described typologies. Finally, Section 7 offers a summary of these find-ings.

1.2. Background

1.2.1. Gumuz. Previous analyses of the Gumuz language assumed that Gu-muz had a class of verbs which were infixing (Irwin 1966:5), which were laterlabeled split roots (Uzar 1989:371). According to Uzar (1997:27), these splittype verbs are to be analyzed as given in example (1). That is, f@rts ‘get up’ isanalyzed as a single root in the imperative form (2), but this same root is splitin the non-future tense (1).

(1) b-aa-f@[email protected] (root 1)-tense-optional.vowel-root 2-tense.suffix 2‘You got up.’

(2) [email protected]‘Get up!’

However, these roots are not split; they are merely complex verbal stems thathave come about via noun incorporation. In nearly all cases, the single verb rootis also possible without the incorporated noun (3). Thus, the verb stem com-prises both a verb root and a noun root. In some cases, the incorporated nounhas grammaticalized to such a degree that it should be considered a deriva-tional suffix (5). In either case, these putative split verb stems should be an-alyzed as comprising two morphemes as given in (4) through (6) below. Dueto phonological reduction and a certain degree of grammaticalization of thesemorphemes, it is understandable that such incorporated nouns were not readilyidentified as such in previous descriptions of the language.

(3) b-aa-f@r-k@2

AFF-2S-rise-NFUT‘You rose.’

2. Uzar analyzed the b- prefix as a mood marker denoting ‘affirmative’, as this marker does notshow up in negatives or questions. The analysis in this paper will maintain the “affirmative”(AFF) label for the corresponding d- prefix in the Northern Gumuz varieties as well, eventhough this designation is lacking (for all varieties) when considering all constructions. Notealso that Gumuz has two tones: high and low. While the Uzar data remains unmarked for tone,the remaining Gumuz data in this paper is marked for high tone and unmarked for low tone,but if the syllable carries a contour, both are marked. Also note that the data is phonemicizedusing IPA characters; however further phonological research on phonemic vowels is underway.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 4: Noun incorporation and predicate classifiers in Gumuz

162 Colleen Ahland

(4) b-aa-f@r-k@-tsaAFF-2S-rise-NFUT-body‘You got up.’

(5) b-aa-f@r-k-ok’w

AFF-2S-rise-NFUT-head‘You lifted (something).’

(6) f@r-tsrise-body‘Get up!’

In northern Gumuz varieties, the complexity of the verbal morphology is notreadily apparent in the non-future tense as this tense is unmarked (7). However,the complexity of certain stems becomes apparent in the 2P imperative. Thus,near minimal pairs like ‘Breathe!’ (8) and ‘Kill!’ (9) are quite different in the2P imperative as ‘Breathe! (2P)’ (10) consists of a single root while ‘Kill!’surfaces as a complex verbal form interrupted by the 2P marker (11).3

(7) d-á-f@r-@tsAFF-3S-rise-body‘He got up.’

(8) Sókw

breathe‘Breathe!’

(9) Só-k’w

die-head‘Kill!’

(10) Sókwí-cabreathe-2P‘Breathe!(2P)’

(11) Sí-có-k’w

die-2P-head‘Kill!(2P)’

1.2.2. Noun incorporation in Africa. Noun incorporation is not unknownto languages of Africa but “systematic incorporation of non-referential objectsis not common in African languages” (Creissels et al. 2008: 97). Such a pro-cess has been documented in Mande of the Niger-Congo family (Hutchison2003) and exists across Bantu in complex verb stems meaning ‘sit’ (Botne1993). Furthermore, there is a set of reconstructed verbs in Proto-Bantu whichhave incorporated nominal roots – more precisely locative nominals – mean-ing ‘on the head’ or ‘in the earth’, for example (Meeussen 1967: 88). As forEast Africa, the Cushitic languages of Somali (Tosco 2004), Iraqw (Mous1993), Alagwa (Kiessling 2007) and Boni (Sasse 1984) have noun incorpora-tion. Within Ethiopia, noun incorporation has been documented in the EasternCushitic languages of Dullay and Daasanach (Sasse 1984). In addition, Berta,a Nilo-Saharan language which is not deemed to be closely related to Gumuz

3. These two roots are also interrupted by person agreement in the future tense.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 5: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 163

but is spoken in the same geographic area, appears to have a process of nounincorporation that functions in a way similar to that of Gumuz. However, the“incorporated” nouns are not tightly integrated with the verb root as other con-stituents of the predicate can intervene (Neudorf 2008). Thus, more researchis needed before one can definitively label the process in Berta noun incorpo-ration. As for languages closely related to Gumuz, there is no known litera-ture describing either noun incorporation or predicate classification. The mostclosely related languages, those of the Koman subgroup, need further researchin order to discover whether such processes exist.

1.2.3. Gumuz alignment. In order to better understand noun incorporationin Gumuz, it is necessary to understand its system of case and word order.Transitive word order in Gumuz tends to vary between AVO and AOV. How-ever, other word orders are also possible depending on the discourse context.Likewise, word order of intransitives varies between SV and VS. Core argu-ments in Gumuz remain unmarked for case unless the A/S argument followsthe verb, in which situation the A/S is marked with a nominative case markerand the O remains unmarked; hence Gumuz exhibits what König has labeled atype 1 marked nominative language with a split system (2006: 658). Examples(12)–(16) below demonstrate various attested word orders.

(12)Awo-bám@taM-Bamita

Vd-@-dák’w

AFF-3S-grab

Od@gw@nábeautiful

Ng@fawoman

‘Bamita married a beautiful woman.’

(13)AáXo3S

Os@Nasalt

[email protected]

‘He found salt.’

(14)Om@ts’áhouse

Vd-a-sú-k’w

AFF-3S-eat-head

Aá-m@néaNOM-fire

‘The fire consumed the house.’

(15)Sá@gaperson

Vd-á-âúAFF-3S-be.sick

‘The person is sick.’

(16)Vd-á-âúAFF-3S-be.sick

Sá-á@gaNOM-person

‘The person is sick.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 6: Noun incorporation and predicate classifiers in Gumuz

164 Colleen Ahland

Furthermore, peripheral arguments such as indirect objects are marked withthe bound preposition k@-, as shown in (17).

(17) étíz1b@Xatrader

d-a-kud-éé-k’w

AFF-3S-buy-CENT-headk@-gú-mato-place-1S

ópábean

‘The trader sold beans to me.’

2. Noun incorporation

2.1. Incorporated nouns

The body part nouns, head, eye, tooth, mouth, and stomach are regularly in-corporated into verbs to form new verbal compounds. This is not surprising asbody part terms are some of the most commonly incorporated nouns (Mithun1986: 383). These body part nouns exist both as noun stems4 as well as a partof complex verbal stems. Examples (18) through (22) demonstrate these bodypart nouns as noun stems (a) as well as their corresponding incorporated forms(b).5 It is important to note that the free noun forms of ‘head’ (18a) and ‘eye’(19a) and ‘mouth’ (21a) are inherently possessed and thus never occur as thesimple noun root.

(18) a. áX1má3S

obí-ts1-mábe.big-body-NMLZ

l1-k’ú-máPOSS-head-3S

d-á-wít-1nAFF-3S-exist-LOC‘He/she has a big head.’

b. d-é-wíâ-@r-ók’w

AFF-FUT-see-1S-head(á@ga)person

‘I will visit (someone).’

(19) a. éEndáother

á@gaperson

l1-cí-máPOSS-eye-1S

d-a-á@tSAFF-3S-hit

‘Another person hit me in the eye.’b. d-é-p@-r@-c

AFF-FUT-go.out-1S-eye‘I will escape.’

4. In the northern Gumuz dialects, many body parts are inherently possessed. Therefore, they donot exist as free morphemes in the language.

5. Example (22b) in which ‘stomach’ is incorporated into the verb is an example of Mithun’sType II noun incorporation (NI which promotes peripheral arguments) (1984: 856); themetaphorical ‘stomach of the house’ was likely promoted to O argument status, e.g., ‘I willwash the stomach of the house’ → ‘I stomach-washed the house’.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 7: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 165

(20) a. k’wós-ímátooth-1S

d-a-âuAFF-3S-hurt

‘My tooth hurts.’b. d-é-Síl-@r@-k’wós

AFF-FUT-pick/pluck-1S-tooth‘I will choose.’

(21) a. lí-lí-s1-main-POSS-mouth-1S

d-@r-t’o-tsAFF-1S-do-body

N"gá

food‘I put the food in my mouth.’

b. d-é-Ng1S-ár@-sAFF-FUT-speak-1S-mouth‘I will whisper.’

(22) a. íl-ímastomach-1S

d-a-âuAFF-3S-hurt

‘My stomach hurts.’b. d-é-Paf-@r-íl

AFF-FUT-wash-1S-stomachm@ts’áhouse

‘I will wash mop/inside the house.’

The body part nouns, ts’éa ‘ear’ and tSogwa ‘foot/leg’ are also incorporatedinto verbs. However, they are not incorporated into verbs as regularly as theother body part nouns listed in this section. Examples (23) and (24) demon-strate these body part nouns as free nouns (a), and as incorporated nouns (b) inGumuz.

(23) a. Ng1fawoman

d-a-t’ooAFF-3S-do

anc’íagold

n@-ts’é-maLOC-ear-3S

‘The woman put gold (jewelry) in her ears.’b. d-é-t’o-@r@-k@-ts’ê

AFF-FUT-do-1S-APPL-ear‘I will be quiet.’

(24) a. d-a-á@tSAFF-3S-hit

n@-tSogo-máLOC-foot-3S

‘He hit (him) in the foot.’b. d-é-t’o-@r@-tSogw

AFF-FUT-do-1S-foot‘I will drive away (someone).’

One other (historically) incorporated noun, the morpheme -Vts ‘body’, nolonger exists as a noun stem in the language. However, native speakers stillrecognize this morpheme as ‘body’ when incorporated into a verb. A relic ofthis noun stem remains in expressions denoting ‘oneself’ as in examples (25)–(26) below (see also Section 2.3.2.2).

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 8: Noun incorporation and predicate classifiers in Gumuz

166 Colleen Ahland

(25) duwachild

d-a-âambAFF-3S-try

m@-nts’áNMLZ-crawl

k@-tsí-máwith-body-3S

‘The child tried to crawl by himself.’(26) áma-lí

2S-EMPHk@-ts-úáwith-body-2S

á-gáX-aNgó2S-be.good-NEG

‘You yourself are not beautiful.’

Lastly, the morpheme, -S does not exist as a free morpheme in Gumuz, noris it identifiable by native speakers of Northern Gumuz as a nominal root (bodypart or otherwise) within a verbal compound.6 However, this morpheme func-tions in much the same way as the other body part morphemes which have beenincorporated into the verb and it occupies the same position structurally withinthe verb stem. Beyond verbal stems, this morpheme can be found as the boundobject of a preposition in an associative expression (27) as well as in a locativeexpression (28). Thus it is assumed to be of nominal origin.

(27) k@-Swith-?

áma2S

‘with you’(28) n@-S

at-base?poXwamountain

d-@r-íi-SAFF-1S-be/live-down?

‘I sat at the base of the mountain.’

2.2. Evidence for noun incorporation

According to Mithun (1986: 379), “Noun incorporation is a lexical processwhereby a noun stem and verb stem are compounded to form a derived verbstem” but once these nouns are incorporated, “they have no syntactic roles asarguments of the clause.” As AVO word order is quite common in Gumuz andincorporated noun roots are incorporated following the verb root, it is neces-sary to demonstrate that the incorporated noun is not an argument of the verb.Furthermore, it was shown in Section 1 that the O argument in Gumuz remainsunmarked for case, thus potentially making it difficult to distinguish an incor-porated noun from an O argument.

Following are five arguments for noun incorporation using both phonologi-cal and syntactic evidence. Phonological evidence for incorporation in Gumuzis that, in most cases, the incorporated noun is phonologically reduced. Thisis one indication that the incorporated noun forms a phonological unit with

6. The morpheme -S likely came from the body part noun Sa ‘hip, pelvic region’ which is anoun form attested in the southern Gumuz area. Gumuz language consultants from Yaso andKamashi were readily able to identify this incorporated noun in certain verb stems.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 9: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 167

the verb. In example (18b) above, the incorporated noun root, k’wa ‘head’,has been phonologically reduced to -Vk’w which is unpronounceable in iso-lation. In addition, Gumuz exhibits anticipatory harmony (also known as re-gressive harmony or left spread) in which the vowel previous to the incor-porated noun/affix must agree in rounding with the consonant/vowel of theincorporated noun/affix; vowel harmony in Gumuz does not operate acrossword boundaries. Thus, when k’wa ‘head’ is incorporated, a central vowelis epenthesized which is rounded to harmonize with the labialization on theconsonant. The epenthesized central vowel must also agree in height with theprevious vowel. As there are only back rounded vowels in Gumuz, the choiceof vowel must be [o] in the case of (18b) as it also agrees in height with schwa.

Further evidence for incorporation is found in argument structure. Again,using example (18), if ‘head’ were counted as a core argument, the verb ‘eat’would need to be ditransitive with ‘head’ and ‘house’ as its objects, a seman-tic unlikelihood given the English translation. Moreover, there is no evidencethat Gumuz verbs can take two bare objects. As shown in (17) above, indirectobjects are marked with the prepositional proclitic, k@-.

Other syntactic evidence for incorporation in Gumuz can be found in bothword order and nominalization. As demonstrated above, various word ordersare attested in Gumuz. That is, the O argument is not required to follow theverb. However, the incorporated noun must follow the verb root (and be phono-logically reduced, if applicable) in order to maintain the same meaning. In (29)and (30) below, the meaning of the verb changes if the incorporated noun isuttered before the verb or if it is uttered in its non-reduced form.

(29) âamb-@ctry-eye/seed‘Try a little (water).’

(30) cí-máseed-3S7

âambtry

∼ âambtry

cí-máseed-3S

‘Try/taste the seed(s).’

Lastly, when a verb is nominalized, m@- is prefixed to the verb and if thenominalized verb is transitive, an additional -má suffix is required. An incor-porated noun, in such cases, should therefore be followed by the transitive suf-fix, -má, if it is truly part of the verb stem. Core arguments of the verb cannotbecome part of the nominalized verb and therefore remain outside the nomi-nalized form. As shown in examples (31) and (32), the O argument ‘hole’, doesnot become part of the derived verbal noun.

7. It appears as though cá ‘seed’ requires a possessive marker in northern Gumuz dialects. How-ever, in the southern dialects, cá ‘seed, eye’ can be uttered without the possessive marker.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 10: Noun incorporation and predicate classifiers in Gumuz

168 Colleen Ahland

(31) k’w@S

digXwahole

‘Dig a hole!’

(32) d-@r-p@N

AFF-1S-wantm@-k’w@S-ímáNMLZ-dig-TR

Xwahole

‘I want to dig a hole.’

On the other hand, when a noun is truly incorporated, it remains within thenominalized verb. In example (20), the noun -Vk’w ‘head’ is incorporated intothe verb and remains part of the derived stem after it has been nominalized.

(33) bambájásweet.potato

m@-k’w@S-ók’ú-máNMLZ-dig-head-TR

d-@r-p@N

AFF-1S-want‘I want to dig sweet potatoes.’

For intransitive verbs, the incorporated noun is clearly not the O argumentof the verb. In many cases the incorporated noun functions more like an in-strument or location as in ‘stand up’ of example (34). However, instrumentsin Gumuz are typically expressed in an oblique phrase with the prepositionk@- ‘by, with’ as shown in example (35), and locations are expressed with thepreposition n@- ‘in, on’ (23a). The fact that a bare noun (without a preposition)cannot denote an instrument or location coupled with phonological reductionof the noun supports an incorporation analysis. Furthermore, the incorporatednoun ‘foot’ in (34) occupies the same position in the verb as the bound mor-pheme -S ‘down’ which marks a location/direction as shown in example (36)and which is also assumed to be an incorporated noun historically.

(34) íi-tSogw

be-foot‘Stand up!’

(35) k@-tSogú-maby-foot-1S

d-@r@-Píí-tSogw

AFF-1S-be-foot‘I stood on my feet.’

(36) íi-Sbe-down‘Sit down!’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 11: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 169

2.3. Valence of complex verbal stems

2.3.1. Intransitive verbal roots

2.3.1.1. No change in valence. Intransitive verbs in Gumuz will either main-tain intransitivity when a noun is incorporated or may become transitive as aresult of incorporation. When intransitivity is maintained, the resulting com-plex verb may further qualify the S argument. For instance, the verb Xa ‘besmall’ incorporates various body part morphemes in order to create more spe-cific verbs. In (38) below, if ‘eye/seed’ is incorporated, the meaning changesto ‘be small in number.’ Likewise, if ‘head’ is incorporated, the verb will mean‘be small in stature’ (39) and if ‘body’ is suffixed to the verb, the verb willmean ‘small in girth’ (40). Lastly, when ‘stomach’ is incorporated (41), theverb will mean ‘be small in breadth.’ All of these verbs can take similar S ar-guments which are merely limited due to the types of arguments which can be‘small in girth’ or ‘small in number/amount’, e.g., a seed cannot be ‘thin’, anda mango cannot ‘be short’, etc. Regardless of the noun being incorporated, theverb remains intransitive.

(37) m@ts’áhouse

d-á-XaAFF-3S-be.small

‘The house is small.’

(38) á@gaperson

d-a-Xa-cAFF-3S-be.small-eye

n@-gú-maleaat-place-crop

‘There are only a few people in the field.’

(39) mámatsí-mabrother-1S

d-a-Xá-k’wAFF-3S-be.short-head

‘My brother is short.’8

(40) l@m@

1S.Possduachild

d-a-Xa-tsAFF-be.small-body

‘My child is thin.’

(41) nâéaground

l-ú-k-ícREL-3P-BEN-give

k-árato-1S

k@-m-tuXofor-NMLZ-build

míts’áhouse

d-a-Xa-ílAFF-3S-be.small-stomach‘The land that they gave me is narrow (too small) for building a house.’

8. At least one language consultant preferred to gloss this as ‘My brother became short’ andtherefore preferred a nominalized version of the verb stem in order to obtain the meaning ‘isshort’.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 12: Noun incorporation and predicate classifiers in Gumuz

170 Colleen Ahland

Another intransitive verb which maintains its valence when part of a complexverbal stem is the verb ‘to be/live’ which was mentioned at the end of Sec-tion 2.2. When either tSogw ‘foot’ ((34) and (35)) or -S ‘down’ (36) is incorpo-rated, the verb changes in meaning, yet maintains its intransitive status.

2.3.1.2. Increase in valence. Occasionally incorporated nouns can increasethe valence of an intransitive verb. This increase in valence can simply add anargument as is the case for ‘speak’ (42) and ‘tell’ (43).

(42) bab-éafather-1S

zialánow

d-@-Ng1S@

AFF-3S-speak‘My father is speaking now.’

(43) d-@-Ng1S1-tsAFF-3S-speak-body

k-árato-1S

á-bab-éaNOM-father-1S

N"gíSá

thingd-á-káAFF-3S-say

k@-m1-gakw-aNgoto-NMLZ-steal-NEG‘My father told me things (like) not to steal.’

With other verbs, this process can produce a causative. Most often, if acausative reading is necessary for the verb, the body part morpheme -Vkw

‘head’ is suffixed to the verb. Bender was the first to hypothesize this pro-cess in his analysis of Gumuz but admitted it was problematical (1979: 43).Certainly, this function as a causative is not productive, only interpretable assuch for a few verbs.

(44) duwachild

d-á-f@rALAFF-3S-rise

n@dí-mí-Píi-Saon-NMLZ-be-down

k@-gozato-sky

‘The child rose from sitting to standing.’

(45) f@r-ók’w

rise-headduwachild

‘Lift the child!’

(46) duwachild

gwásbecause

m@-g@X

NMLZ-goodm@-s1-gwaNMLZ-eat-because?

d-á[email protected]‘Because the child ate well, he grew.’

(47) duwachild

gwábecause

kí-Síl-1ts’é-Z1n@

to–cut-ear-PERFób1-máfather-3S

f@g-ók’w

grow.up-head‘Because his father had died on him, raise the child!’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 13: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 171

Not all causatives in Gumuz are formed with the -Vk’w suffix. At times,other body part morphemes increase the valence and produce a causative read-ing. For instance, when -Vc ‘eye’ is incorporated in the verb ‘fall’ (48), theverb ‘cause to fall’ is derived (49).

(48) wo-bámítáM-Bamita

d-á-bEP

AFF-3S-fall‘Bamita fell.’

(49) wo-bámítáM-Bamita

bEP-ícfall-eye

á@ga-nínáperson-DEM

‘Bamita, make that person fall!’

However, when -Vk’w is suffixed to this same verb, a transitive (perhaps causa-tive) yet distinct verb is derived. In (50) below, the newly formed verb means‘lower’.9

(50) bEP-ók’w

fall-headsi@rope

dílittle

Pé-amp@X-@r-ók’w

FUT-jump-1S-head‘Lower the rope a little and I will jump over it.’

2.3.2. Transitive verbal roots. Depending on the verb, the valence of a tran-sitive verbal root may or may not change when a noun is incorporated. Someverbs maintain their valence while others may decrease in valence.

2.3.2.1. No change in valence. Similar to intransitive verbal roots, there aretransitive verbal roots whose valence does not change after forming a complexverbal stem with an incorporated noun root. Likewise, this incorporated nounmay further qualify the O argument (51b), or the two roots combine to create anew but related meaning (52b).

(51) a. mítáchicken

sáeat

‘Eat the chicken.’b. ítSa

wormd-a-s-ílAFF-3S-eat-stomach

k’ósí-matooth-1S

‘The worm ate (the) inside of my tooth.’

(52) a. wiâsee.2S.IMP

á@gaperson

‘See the person!’

9. Note that the suffix -Vk’w on the verb ‘fall’ may also be functioning as a predicate classifierhere as is the case for ‘jump’. See Section 3.5.1 for comparison.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 14: Noun incorporation and predicate classifiers in Gumuz

172 Colleen Ahland

b. d-@r-tsAFF-1S-go

d-@´AFF-1S-see-CENT-head

r-uâ-e-k’w

personá@gaLOC-house

l@-m1ts’áLOC-DEM

l@-t’a

‘I went (and) visited people at that house.’

2.3.2.2. Decrease in valence. According to Mithun, Type I noun incorpo-ration involves the incorporation of an O argument into the verb stem whichresults in a decrease in valence. Such a type of NI historically must precedeall other types of NI, and should at least be found in relic form if such a pro-cess is no longer productive in the language (1984: 874, 891). In Gumuz, NIas a valence-decreasing operation only appears to be productive for reflexives.When the incorporated nominal -Vts ‘body’ is suffixed to certain verbs, the re-sulting complex verbal stem can be part of a transitive or intransitive construc-tion. When part of an intransitive construction, the action is directed towardthe S argument.10 However, reflexives can be expressed within either construc-tion. Such is the case for the verb Paf ‘wash’ in which one can either use thetransitive construction (53) or the intransitive construction given in (54). How-ever, when -Vts ‘body’ is incorporated into certain verbs (like ‘wash’) in atransitive construction, the complex verbal stem does not necessarily take on areflexive meaning unless the possessed (reflexive) form of á@g@ ‘person/body’serves as the O argument (see Section 3.3.1). In this regard, noun incorporationand hence predicate classification functions on an absolutive basis (Aikhenvald2000: 162, Keenan 1984).11

(53) Paf-Etswash-body

á@g-uábody-2S

‘Wash your body/yourself.’

(54) Paf-Etswash-body‘Wash up/Bathe.’

10. In addition, there is a tonal distinction between the transitive and intransitive constructionsregardless of whether or not the meaning is reflexive. For the transitive construction, subjectagreement for all persons save 3S is marked with H tone; 3S is marked with L tone. For theintransitive construction, all subject agreement is marked with HL tone while 3S is markedwith H tone.

11. While Aikhenvald alludes to only predicate classification functioning on an absolutive basis,there is evidence in Gumuz that noun incorporation in general functions this way: in intran-sitive constructions, whether functioning as classifiers or not, the incorporated noun tends toqualify the S argument (see examples (38)–(41)). Furthermore, the S arguments of complexverbal stems can take either nominative or accusative ( /0) case marking when following theverb while simple verb roots in intransitive constructions only take nominative case marking.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 15: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 173

Other valence-decreasing reflexive constructions arise from incorporating spe-cific body part nouns into the verb. The transitive construct in (55a) below canalso be expressed as an intransitive construct which then takes on a reflexivemeaning (55b). Such a valence-decreasing process can be seen as the result ofpossessor raising which is discussed in more detail in the following section.

(55) a. d-é-Paf-@[email protected]

lící-maface-1S.poss

‘I will wash my face.’b. d-é-Paf-@r-Ť@líc

AFF-FUT-wash-1S.intrans-face‘I will face-wash.’

When the incorporated nominal -Vts ‘body’ is suffixed to certain verbs, theresulting complex verbal stem can be part of a transitive or intransitive con-struction. When part of an intransitive construction, the acion is directed towardthe S argument.12 However, reflexives can be expressed within either construc-tion. Such is the case for the verb Paf ‘wash’ in which one can either use thetransitive construction (51) or the intransitive construction given in (52). How-ever, when -Vts ‘body’ is incorporated into certain verbs (like ‘wash’) in atransitive construction, the complex verbal stem does not necessarily take on areflexive meaning unless the possessed (reflexive) form of á@g@ ‘person/body’serves as the O argument (see Section 3.3.1). In this regard, noun incorporationand hence predicate classification functions on an absolutive basis (Aikhenvald2000: 162, Keenan 1984).13

(35) tsbody-2S

á@g-uá

‘your body/yourself.’

(36) -Ets‘Wash up/Bathe.’

12. In the first person, however, there is a tonal distinction between the transitive and instransitiveconstructions regardless of whether or not the meaning is reflexive. This is not true for thesecond and third persons for which the tone remains the same regardless of transitivity.

13. While Aikhenvald alludes to only predicate classification functioning on an absolutive basis,there is evidence in Gumuz that noun incorporation in general functions this way: in intran-sitive constructions, whether functioning as classifiers or not, the incorporated noun tends toqualifiy the S argument (see examples (24)–(28)). Furthermore, the S arguments of complexverbal stems can take either nominative or accusative ( /0) case marking when following theverb while other simple verb roots in intransitive constructions only take nominative casemarking.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 16: Noun incorporation and predicate classifiers in Gumuz

174 Colleen Ahland

Likewise, the reciprocal forms of verbs can take the incorporated nominal-Vts ‘body’ in addition to the reciprocal prefix a- resulting in either an intran-sitive construct (56) or the transitive construct (57).14

(56) a-Xok-@[email protected]‘Let’s make each other warm!’

(57) a-Xok-@[email protected]

á@gá-kwabody-1P.INCL

‘Let’s make each other warm!’

Type I noun incorporation involves the incorporation of an O argument into theverb stem which results in a decrease in valence. Such a type of NI historicallymust precede all other types of NI, and should at least be found in relic formif such a process is no longer productive in the language (1984: 874, 891). InGumuz, there are only a few lexicalized forms of Mithun’s Type I noun incor-poration as intransitive verb ‘be quiet’. Other than reflexives and reciprocals,there are only a few lexicalized forms of Mithun’s Type I NI. For example, in(23b) above, ‘ear’ is incorporated into the transitive verb ‘do’ in order to formthe intransitive verb ‘be quiet’.

2.3.3. Possessor raising. One common function of noun incorporation isthat of incorporating a body part noun which results in promotion of the pos-sessor of the body part to a primary case role. This is what Mithun has labeledType II incorporation (1984: 856). In Gumuz, such a process is transparentas the incorporated body part noun often maintains the possessive marking.15

In (58), ‘mouth’ in its possessed form is incorporated into the verb ‘blow’.The possessor of the mouth, ‘the child’, therefore functions as the O argumentwhereas before noun incorporation, ‘child’s mouth’ functions as the O argu-ment (59). The verb is still part of a transitive construction but the possessorhas been raised to core argument status when the possessed form of the O ar-gument is incorporated into the verb.

(58) fwítS-íl-isblow-POSS-mouth

duwachild

d-@-k’@N-1l-1sAFF-3S-bite-POSS-mouth

á[email protected]‘Blow on the child’s mouth; the hot pepper is burning him.’

14. It has not yet been determined why the velar fricative [X] is uttered in this construction andnot elsewhere for the root ‘heat’.

15. It was found when using the verb ‘hit’ that ‘tooth’, ‘stomach’, and ‘rear end’ were incorpo-rated without the POSS prefix for sentences meaning ‘I hit the man in the__’.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 17: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 175

(59) fwítSblow

lí-sa-duwaPOSS-mouth-child

‘Blow on the child’s mouth.’

3. The system of predicate classification

3.1. Typologies of predicate classification

Both Aikhenvald (2000) and Grinevald (2000) have placed predicate classi-fiers within an overall typology of noun classification. According to Aikhen-vald, verbal (predicate) classifiers come in three forms: (1) classificatory nounincorporation, (2) verbal classifiers as affixes, and (3) suppletive classificatoryverbs (2000: 149). Her first type of verbal classifier is akin to that described byMithun (1984, 1986) whereby “a noun is incorporated into a verb to catego-rize an extra-predicate argument . . . usually in S or O function”. With this typeof verbal classifier, there is frequently a generic-specific relationship betweenthe incorporated NP and the external NP which accompanies it (Aikhenvald2000: 149–150). Aikhenvald’s second type of verbal classifiers are classifierswhich are affixes and not nominal roots. These affixes might have arisen fromnouns or verbs and thus, such classifiers could have developed (historically)from classificatory noun incorporation (2000: 152). With her third type of ver-bal classifier, suppletive classificatory verbs, the choice of verb is conditionedby the inherent properties of the S/O argument or their orientation/stance inspace. As the name suggests, these verbs are suppletive and therefore not an-alyzable synchronically (2000: 153). One of the major differences betweenAikhenvald’s types 1 and 2 is that in classificatory noun incorporation, the clas-sifier cannot be used in other classifier environments. That is, in some “multipleclassifier languages”, the same classifying morpheme that is used in complexverbal stems may also be used in nominal compounds.16 If such is the case,Aikhenvald suggests this is characteristic of verbal classifiers as affixes ratherthan classificatory noun incorporation. Furthermore, with this latter type ofverbal classifier, verbs retain their same argument structure and the incorpo-rated noun categorizes an overtly expressed argument (2000:160).

Grinevald’s typology of verbal (predicate) classifiers is quite similar to thatof Aikhenvald’s except Grinevald only recognizes two subtypes: (1) incorpo-rated classifier construction and (2) verbal classifying affixes. Grinevald dis-misses Aikhenvald’s third type as it is “a covert lexical means of nominalclassification” which “can be found in any language” (2000: 68). Similar to

16. Aikhenvald (2000) considers what are labelled ‘class terms’ in this paper to be classifierswhile Grinevald (2000: 59) does not.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 18: Noun incorporation and predicate classifiers in Gumuz

176 Colleen Ahland

Aikhenvald’s classificatory noun incorporation, she describes the classifiermorpheme in an incorporated classifier construction as still recognizable asa generic noun. On the other hand, her verbal classifying affixes are phono-logically very eroded much like Aikhenvald’s verbal classifiers as affixes. Herfirst type is akin to noun classifiers while her second type is semantically akinto numeral classifier types (2000: 67). Grinevald describes numeral classifiers(those typically found in quantification expressions) as classifying physical cat-egories: round, long, flat/flexible, etc. Noun classifiers, on the other hand, tendto classify according to material essence: animal, rock, man, etc. (2000: 72).

Gumuz appears to have characteristics of both Aikhenvald’s and Grinevald’stypes 1 and 2. Aside from the semantic differences in these classifier types de-scribed by Grinevald, both authors acknowledge that the difference betweenthese types can be merely degrees of grammaticalization with incorporatednouns towards the lexical end of the continuum and the classifying affixes to-ward the grammatical end. For this reason, Aikhenvald considers incorporatedclass terms (classifying morphemes in nominal compounds) to be more gram-maticalized and therefore more akin to classifying affixes.

3.2. Predicate classifiers as incorporated nouns

According to Mithun, predicate classifiers arise from incorporation of genericnouns, usually body parts (1986: 383).17 The body part nouns ‘head’, ‘eye’,‘tooth’, and ‘ear’ as mentioned previously, are often incorporated into Gumuzverbs to form new verbal compounds. In some instances, these incorporatednouns function as classifiers. In the genesis of predicate classifier systems, in-corporated nouns move from merely qualifying external noun phrases to repre-senting a basic level category in which the incorporated noun and the externalnoun phrase are functioning within a hypernym-hyponym relationship (Mithun1986:387). For example, one might be tempted to consider (38) through (41)above as examples of predicate classification. However, while the incorporatednouns in these instances may narrow the scope of the verb and therefore limitthe type of S argument, the incorporated nouns do not function as hypernymsfor the S arguments.18 On the other hand, in the examples that follow in this

17. Aikehenvald has identified cases where certain predicate classifiers have arisen from verbs(2000: 152).

18. Examples (38)–(41) likely arose from Mithun’s Type II noun incorporation (1984: 856) inwhich for example (40) ‘child’ was promoted to S argument and ‘body’ was incorporated.Thus, example (40) could be historically derived from a clause like ‘The body of my child issmall’. While Type II NI can lead to a system of predicate classification, it does not appear tohave done so in this case.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 19: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 177

Table 1. Incorporated nouns/classifiers with corresponding free forms

Gloss Simpleincorporated noun

Incorporated nounin the possessive

Free noun

‘head’ -Vk’w -ilúk’w lúk’wá‘eye’ -Vc -1líc lícá‘tooth’ -1k’wós -il1k’wós k’wósa‘ear’ -ts’ê -il1ts’ê ts’éa

section, the incorporated nouns do indeed function as hypernyms for the S/Oand, at times, for the peripheral arguments of the clause.

In Gumuz, not only are the above-mentioned body part nouns incorporatedas simple roots but also as possessed nouns (marked with the possessive markeril-).19 Table 1 provides the incorporated noun forms/classifiers in the secondand third columns and their corresponding free nominal forms in the fourthcolumn.

In addition, -Vts ‘body’ functions as a classifier in Gumuz. Unlike the bodypart nouns mentioned above, there is no known free nominal form that stillexists in Gumuz and there exists no incorporated possessed form.

3.3. Evidence for a classifier system

The incorporated nouns from Section 3.2 above classify the O (and occasion-ally the S) arguments of certain verbs in Gumuz. This system of predicate clas-sification is presented in Table 2 below. Nouns incorporated as simple rootsand nouns incorporated as possessed items form distinct classes when affixedto certain verbs. However, for other verbs, these classes have merged and thereappears to be no distinction in class membership of the O arguments they clas-sify. As can be seen in Table 2, a simple generalization for members of eachclass is difficult as the system of classification is somewhat opaque (at least fornon-possessed noun roots).

Predicate classifiers co-occur with at least 19 verbal roots in Gumuz. Such aconfined set is not atypical as “the use of [predicate] classifiers is often limitedto certain semantic groups of verbs” (Aikhenvald 2000: 149).20 However, not

19. Such classifiers likely evolved from the process of possessor raising mentioned in Section2.3.3.

20. Aikhenvald states that predicate classification is often realized with telic verbs. For example,‘see’ being atelic is less likely to use classifiers than say ‘look’ (2000: 165). Curiously, in Gu-muz, the verbal root ‘see’ changes in meaning to the more telic verb ‘check’ when classifiersare incorporated.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 20: Noun incorporation and predicate classifiers in Gumuz

178 Colleen Ahland

Table 2. System of classification for S/O arguments

Predicate classifier Semantics of classified nouns Examples of nouns

-Vk’w ‘head’ entities that are head-like inshape and/or function orclosely associated with suchobjects

– fingers, toes– water, sauce, beer, lotion,

soap (in a container)– ears of corn– pots, pans, cans

-Vc ‘eye/seed’ entities associated withwounds, fire, small seed-likeobjects, and outer coverings

– knives, needles, spears– clothes, tree bark, orange

peel– wounds, blood, water,

coffee, tea, sauce– fire, light– beans, seeds

-1k’wós ‘tooth’ entities tooth-like in sizeand/or shape and/or whichform groups of identicalobjects

– beads, pins, metal spoons,firewood

-Vts ‘body’ the human body, entitiesassociated with the humanbody or that are functionallyor physically body-like(objects longer than they arewide)a

– people, hands, feet, face,head

– walls, shoes, yams, cars,wooden spoons

-ilúk’w

‘head.POSS’fruits, vegetables and otherobjects that are head-like inshape and function

– pumpkins, oranges, rocks

-1líc‘eye/seed.POSS’

seeds, beans as well asobjects which bear the wordfor ‘seed/eye’ within its name

– sorghum, millet, beans

-il1k’wós‘tooth.POSS’

small tooth-like entities thatcome in quantities or groupsof identical objects and arepart of another object

– buttons, beads, seeds

-ts’ê / -il1ts’ê‘ear’/‘ear.POSS’

flat, thin, flexible objects– mushrooms, paper, leaves

a. The material of the object seems to play some role in the classification as metal spoons andwooden spoons are classified separately but this could be due to the larger size of the woodenspoon.

all classifiers are equally employed with each verbal root. Thus, the result isa limited use of predicate classifiers in Gumuz. Table 3 summarizes what is

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 21: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 179

currently known regarding co-occurrence of verbal roots and predicate classi-fiers in Gumuz. While many verb-noun combinations are possible in Gumuz,not all incorporated nouns function as classifiers when combined with a par-ticular verbal root even if other incorporated nouns do function as classifiers.For example, wiâ-ok’w see-head means ‘visit’ but other incorporated nounsin combination with wiâ ‘see’ function as classifiers, changing the meaningto ‘check’ (see examples (127)–(130)).21 Other verb-noun combinations aresimply not possible, e.g., *tSEP-ok’w dry-head.

Of these complex verbal stems, nearly all are transitive and, for the mostpart, classify the O argument of the verb. In some instances, the S argument isclassified and thus, as previously mentioned, predicate classification operateson an absolutive basis. In at least one other instance, a peripheral argument, theinstrument, is classified.

3.3.1. Classification of S/O arguments. The verb with possibly the mostelaborate system of predicate classifiers in northern Gumuz is Paf- ‘wash’. Thisverbal root is bound in the variety of Gumuz spoken in Gilgel Beles (northernGumuz), requiring a predicate classifier. However, in another northern Gumuzvariety (that of Dangur), such a verb can be uttered on its own as in (60) below.This bound verbal root (in Gilgel Beles) is not surprising, though, consider-ing that in the development of a classificatory system, “certain verbs of widescope occur less and less frequently by themselves, and more frequently withparticular incorporated nouns” (Mithun 1986: 384).

(60) ára1S

d-@r@-PafAFF-1S-wash

lí-c@-maPOSS-face/eye-1S

‘I washed my face.’

For this verb, the classifier -Vc ‘eye/seed’ is used for a seemingly arbitraryclass of clothes, sharp metal objects, wounds, and blood (61, 62). But the pos-sessive counterpart -1lic classifies all seeds and grains, fairly consistently. Inaddition, ‘arrow’ is part of this latter class as it is a possessed ‘eye/seed’ as perthe compound name for ‘arrow’, ‘eye-bow’ (64) (see Section 3.6 for furtherdiscussion).

(61) áXwaclothes

Paf-@cwash-eye

‘Wash the clothes!’

21. One might assume that ‘head’ is classifying people in this instance (for ‘visit’). However, withthe meaning ‘check’, people are classified with the morpheme -Vts as in (131). Ultimately, itis difficult to determine for certain whether or not -Vk’w is classifying the O argument, as theverb with the sense of ‘visit’ only takes people as O arguments.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 22: Noun incorporation and predicate classifiers in Gumuz

180 Colleen Ahland

Table 3. Co-occurrence of predicate classifiers and verbal roots

Verbal Roots Non-possessed/Possessed predicate classifiers

-Vk’

w‘h

ead’

-ilú

k’w

-Vc

‘eye

’-1

líc

-1k’

wos

‘too

th’

-il1

k’w

ós

-ts’

ê‘e

ar’

-il1

ts’ê

-Vts

‘bod

y’

Paf- ‘wash’ × × × × × × × × ×k’woS ‘pierce/penetrate’nS ‘soak’

× × × × × ×

wok ‘heat’ × × × × × ×Sil ‘pick/pick up’ × × × ×fuN ‘smell’ × × × × × ×w1â ‘see/check’ × × × × × × × ×fwic’ ‘blow’ × × × × × ×tSEP ‘dry’ × × × × ×kor1k’ ‘peel’ × × × ×t@k’ ‘spit’ampoX ‘jump over’

× × ×

ca ‘give’ × × ×Sá ‘die/extinguish’ × × ×nt’ ‘twist’ × × × × ×g1s ‘fry/grill/roast’ × × × ×koâ ‘skin/strip’ × × × ×Nar ‘take’ × × ×c1m ‘sew’ × ×

(62) Paf-@cwash-eye

m@Xablood

‘Wash off the blood!’

(63) S1mbiráchickpea

Paf-1l-ícwash-POSS-eye

‘Wash the chickpeas!’

(64) d-@r@-Paf-1l-ícAFF-1S-wash-POSS-eye

c@-dogwaeye-bow

‘I washed the arrow.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 23: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 181

The incorporated simple noun root, -Vk’w ‘head’ classifies extremities of thehuman body (or body parts that function like a head). Thus, it is incorporatedinto the verb when the object is one’s face, toes, fingers, etc. as exemplifiedin (65) and (66). The possessive form of the noun, however, classifies thoseobjects that are head-like in shape and size, mainly fruits and vegetables asin (67) and (68).22 Therefore, it should be noted that one’s own head cannotbe classified with this incorporated noun as shown in (66) even though thefree form of this noun is inherently possessed and literally refers to one’s head(66).23

(65) Paf-ók’w

wash-headl1-c-úáPOSS-eye-2S

‘Wash your face!’

(66) Paf-ók’w

wash-headk’wa-tS@g-uá24

head-foot-2S‘Wash your toes!’

(67) Paf-il-úk’w

wash-POSS-headb@rtukánaorange

‘Wash the oranges!’

(68) patú@

pumpkind-a-Paf-1l-úk’w

AFF-3S-wash-POSS-head‘S/he washed the pumpkins.’

(69) *li-k’ú-maPOSS-head-1S

d-@r-Paf -il-úk’w

AFF-1S-wash-POSS-head‘I washed my head.’

Likewise, ‘tooth’ incorporated as a simple noun root (70) classifies a distinctcategory of objects from that of its possessive counterpart, (71) and (72). In theformer, however, only one object, ‘spoon’, was found to fit the class.

(70) Paf-1k’wóswash-tooth

mankiametal.spoon

‘Wash the spoon(s)!’

22. It may be that these fruits and vegetables are classified as ‘head’ because they function asmetaphorical heads of the plant.

23. The utterance in (69) is only grammatical as an intransitive with an intransitive tonal melody:d@rafílúk’w ‘I washed my head’.

24. This is much like Mithun’s Type II noun incorporation (1986: 383) which is not considered aclassifier system but leads to such a system historically. However, it is unlike Type II incor-poration in that the body part, ‘head’, is marked both on the verb and the possessor ‘foot.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 24: Noun incorporation and predicate classifiers in Gumuz

182 Colleen Ahland

(71) Paf-il1-k’wóswash-POSS-tooth

minzabead

‘Wash the beads!’

(72) Paf-1lí-k’wóswash-POSS-tooth

k’wal1fbutton

‘Wash the buttons!’

In addition to the above mentioned classifiers, one other noun was foundto be incorporated into the verb for ‘wash’ and function as a classifier: ts’éa‘ear’. Unlike the other incorporated nouns mentioned in this section, this noundoes not appear to be incorporated into as many verbs in Gumuz. Nonetheless,it serves as a classifier when incorporated into four of the verbal roots fromTable 3. The incorporated noun -Vts’ê ‘ear’ and its possessed counterpart -il1ts’ê both classify flexible, ear-like objects such as mushrooms (73, 74) aswell as other thin, flat, flexible objects such as paper or leaves.

(73) Paf-@ts’éwash-ear

waagáwhite.mushroom

‘Wash the white mushrooms!’

(74) Paf-ílí-ts’êwash-POSS-ear

@m1dadêblack.mushroom

‘Wash the black mushrooms!’

Lastly, -Vts ‘body’, only exists in its non-possessed form. This is not sur-prising as this morpheme no longer exists as a free noun in the language. Asa classifier, -Vts refers to the human body (including body parts) and entitiesthat are body-like in shape (longer than they are wide) and function. In (75),the verb is intransitive and therefore classifies the S argument (as discussed inSection 2.3.2.2).

(75) ára1S

d-@r-Paf-EtsAFF-1S-wash-body

‘I bathed.’

(76) ára1S

d-@r-Páf-EtsAFF-1S-wash-body

ejâ-mhand-1S

‘I washed my hands.’

(77) Paf-Etswash-body

tSampáshoe

‘Wash the shoes!’

(78) Paf-Etswash-body

á@ga-m1ts’abody-house

‘Wash the walls!’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 25: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 183

This system of classification of S/O arguments is further demonstrated inother verb roots. The verb, wok- ‘heat’, is also a bound verbal root as the incor-porated noun/derivational affix functions as both a verbalizer and a classifier;the root wok- comes from the noun, ooka ‘sun’. For the verb ‘wash’ in ex-amples (65)–(69) above, the incorporated simple noun root, ‘head’, classifiesextremities of the human body, whereas this same classifier classifies a quitedistinct set of objects for ‘heat’. In (79) through (81) below, -Vk’w classifiesliquids that are typically contained. That is, the classifier ‘head’ classifies thecontainer that these liquids are placed in (as such a container is necessary forthe verb ‘heat’).

(79) wok-ók’w

heat-headajawater

‘Heat the water.’

(80) wok-ók’w

heat-headkúXwasauce

‘Heat the sauce.’

(81) wok-ók’w

heat-headkealocal.beer

‘Heat the beer.’

The possessive counterpart -iluk’w ‘head (POSS)’, however, closely mirrorsthe categorization found with ‘wash’ above. This classifier classifies anythinghead-like in shape and size.25

(82) wok-il-úk’w

heat-POSS-headk’wa-gíSáhead-rock

éé-Paf-@ra-k@-tsFUT-wash-1S-APPL-body

‘Heat the rock and put it in the water for me to bathe with.’(The rock will heat the water.)

With the simple noun root -Vc ‘eye’, there is some overlap in classification.This morpheme is used to classify liquids as does ‘head’ above. However,‘head’ cannot be used to classify ‘coffee’ or ‘tea’ while the classifier -Vc ‘eye’can. It is assumed that this classifier is classifying the liquids regardless of theircontainers while -Vk’w ‘head’ classifies liquids in terms of their container (orliquids that are contained). In any case, such overlap is not unusual for pred-icate classifiers as Aikhenvald observes, “every noun in a language does notnecessarily require a verbal [predicate] classifier; some nouns may be associ-ated with more than one classifier” (2000: 149).

25. It appears as though there is an old system of class terms that still exists to some extent inGumuz. See Section 3.6 for further discussion.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 26: Noun incorporation and predicate classifiers in Gumuz

184 Colleen Ahland

(83) wok-@cheat-eye

kúXwasauce

‘Heat the sauce.’

(84) wok-@cheat-eye

Sajatea

‘Heat the tea.’

Again, the possessive form of the noun, -il1c, classifies the object of ‘heat’ inthe same manner as ‘wash’ above. This classifier is used for any type of seedor grain.

(85) wok-1l-ícheat-POSS-eye

cá-ófókwaéáseed-corn

‘Heat the corn kernels.’

(86) wok-1l-ícheat-POSS-eye

tSópabutter.bean

‘Heat the butter beans.’

Lastly, the classifier -Vts ‘body’ again classifies objects that are associated withthe human body or are body-like/flesh-like as in (87) and (88) below.

(87) wok-@tsheat-body

b1tS’ameat

‘Heat the meat.’

(88) wok-@tsheat-body

unayam

‘Heat the yams.’

As mentioned previously, this classifier can also be used to classify the S argu-ment of a reflexive or a reciprocal construction as in (89) below.

(89) a-Xok-@[email protected]‘Let’s make each other warm!’

3.3.2. Classification of instruments. While it is most common for predicateclassifiers in Gumuz to classify the S/O argument, it is also possible for these toclassify instruments. Classifying peripheral arguments is not so unusual amonglanguages with predicate classifiers (Aikhenvald 2000: 162). With the verb c1m‘sew’, for example, if one sews by machine, the classifier ‘body’ is required(90). On the other hand, if one sews by hand the classifier ‘eye’ is necessaryas this classifies sharp metal objects such as needles (91). This latter classifierhappens to also classify clothes (which is the O here) but in this example itclassifies the instrument.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 27: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 185

(90) áXwaclothes

c1m-@tssew-body

‘Sew the clothes by machine.’

(91) áXwaclothes

c1m-@csew-eye

‘Sew the clothes by needle (by hand).’

3.4. Merging of categories/classes in certain verbs and not others

The incorporated nouns mentioned in Section 3.3 show up as classifiers onother verb roots as well (as demonstrated above in Table 3), but there doesnot always seem to be a strong distinction in category between the possessedas opposed to the simple root body part nouns; for many verbs these two areinterchangeable. For the verb ‘burn/roast’ below, -1l1k’wos and -k’wos are in-terchangeable, (92) and (93).26

(92) g1s-ílí-k’wosburn-POSS-tooth

ófkwaéacorn

‘Roast the corn.’

(93) g1s-k’wosburn-tooth

ófkwaéacorn

‘Roast the corn.’

Other verbs as well have been found to maintain some distinction in mean-ing between the possessed and simple root incorporated nouns. For example,‘soak’ or ‘immerse in water’ can take the following incorporated noun forms:-Vk’w, -Vc, -1l1c, -1l1k’wos, and –Vts. The distinction between -Vc and -1l1c issimilar to that of the verb ‘wash’ above: -Vc is used for clothing (94) and -1l1cis used for seeds and grains (95). As for the remaining classifiers, -Vk’w is usedfor head-like objects (both in terms of shape and function) such as corn (96)and pieces of wood (97), -1l1k’wos overlaps with -Vc, classifying seeds andgrains (98) and lastly, -Vts again classifies body-like objects (99). The otherbody part nouns (possessed or simple root) are not incorporated in this verb.

(94) áXwaclothes

nS-@csoak-eye

‘Soak the clothes.’

26. ‘Corn’ usually does not fall into the class marked by either the possessive or simple root formof the incorporated noun ‘tooth.’ However, because corn (on the cob) is normally roasted inrows, it falls under the description ‘groups of identical objects.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 28: Noun incorporation and predicate classifiers in Gumuz

186 Colleen Ahland

(95) nS-1l-1csoak-POSS-eye

c@-taNkaseed-millet

‘Soak the millet seeds.’

(96) nS-ok’w

soak-headófókwaéácorn

‘Soak the ears of corn.’

(97) nS-ok’w

soak-headéawood

‘Soak the pieces of wood.’

(98) nS-1l-k’wossoak-POSS-tooth

taNkamillet

‘Soak the millet.’

(99) nS-Etssoak-body

éawood

‘Soak the long pieces of wood.’

The verb, k’woR1k’ means ‘remove an outer layer from something’ or‘peel’.27 The possessed and non-possessed forms of ‘head’ are both incorpo-rated into this verb (100)–(105) and are often used interchangeably for the sameobjects as in (101) through (105). Other possessed nouns are not incorporatedinto this verb but the simple root forms, -Vts and -Vc are as shown in (106)through (109).

(100) k’woR1k’-ók’w

peel-headsóXwametal.object

‘Open the canned food.’

(101) k’woR1k’-ók’w

peel-headb@rtukánaorange

-1l-úk’w

-POSS‘Peel the orange.’

(102) k’woR1k’-ók’w

peel-headg1S@

rockek’weNkúXwaradio

‘Remove the batteries from the package.’

(103) k’woR1k’-ók’w

peel-headíS1mEtáegg

‘Peel the hard-boiled egg.’

27. At least one speaker of the northern variety of Gumuz preferred to use a different verb rootfor ‘peel’, ts’eR1k’. His choice of classifiers, however, was the same as those given in (100)through (105).

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 29: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 187

(104) k’woR1k’-1l-úk’w

peel-POSS-headg1S@

rockek’weNkúXwaradio

‘Remove the batteries from the package.’

(105) k’woR1k’-1l-úk’w

peel-POSS-headíS1mEtáegg

‘Peel the hard-boiled egg.’

(106) k’woR1k’-@tspeel-body

mozábanana

‘Peel the banana.’

(107) k’woR1k’-@tspeel-body

éámpasugarcane

‘Peel the sugarcane.’

(108) k’woR1k’-@cpeel-eye

batSolapeanut

‘Shell the peanut.’

(109) k’woR1k’-@cpeel-eye

opabean

‘Remove the bean from the pod.’

3.5. Further examples of classification by classifier

3.5.1. The classifier -Vk’w. As mentioned previously, the classifier -Vk’w

‘head’ classifies entities that are head-like in shape and/or function. The oneseeming exception is the inclusion of liquids in this class. However, if oneconsiders that liquids are commonly contained and such containers fit into thisclass, the class as a whole remains cohesive. Further evidence that liquids areclassified by their container is given below. In (110), if one jumps over a stream(in which the water is not contained) the classifer -Vc ‘eye’ is used. However,when water is in a container, the classifier -Vk’w ‘head’ is used (111).

(110) ájawater

d-@r-ampó[email protected]

‘I jumped over the stream.’

(111) ájawater

l@-bElíLOC-bucket

d-@r-ampóX-ók’w

AFF-1S-jump.over-head‘I jumped over a bucket of water (lit. ‘water in a bucket’).’

Another example of ‘head’ classifying contained liquids occurs with the verb‘give’. If one gives water to another, the classifier -Vk’w ‘head’ must be usedpresumably because if liquid is to be given, it must be given via a container(112). In contrast, when a body part noun (metaphorically) is the O argument

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 30: Noun incorporation and predicate classifiers in Gumuz

188 Colleen Ahland

of the verb ‘give’, as in the idiom for ‘request permission’ (lit. ‘give hand’),the -Vts ‘body’ morpheme must be used (113). Hence, at least two classifierscan be used with ‘give’ demonstrating that cók’w is not a lexicalized formfor ‘give’ and that the morpheme -Vk’w ‘head’ is indeed classifying containedliquids.

(112) duwachild

d-a-có-k’w

AFF-3S-give-headajawater

kó-ób1-máto-father-3S

‘The child gave water to his father.’

(113) c@-tsgive-body

e-wáhand-2S

‘Request permission!’

3.5.2. The classifier -1luk’w. Besides classifying fruits and vegetableswhich function as metaphorical heads of the plant, the classifier -1luk’w hasbeen shown to classify head-like entities such as rocks as in (82), but also thisclassifier can classify a ‘head’ of an animal as in (114).

(114) k’woâ-1l-úk’w

skin-POSS-headlúk’@-Xosahead.POSS-cow

‘Skin the cow’s head.’

3.5.3. The classifier -Vc. The classifier -Vc ‘eye’ has the most varied classmembership. Previous sections gave examples of this classifier used withclothes ((61), (93)), blood (62), seeds/nuts ((108), (109)), liquids ((29), (83),(84), (110)) and sharp objects like needles (91). Similar to ‘clothes’ which isan outer covering, the outer skin/peel of fruits, vegetables, and plants is alsoincluded in this class. In examples (100) through (109) above, the verb ‘peel’classified the object being peeled. However, if the classifier -Vc ‘eye’ is markedon the verb, this can either refer to small seed-like objects being peeled ((108),(109)) or to the outer covering itself (115). Therefore, because éámpa ‘sug-arcane’ belongs to a different class than its outer covering, the verb takes adistinct classifier depending on whether the object is the outer skin of the sug-arcane (116) or the sugarcane itself (107).

(115) k’woR1k’-@cpeel-eye

béé-batSolaskin-peanut

‘Shell the peanut shell.’

(116) k’woR1k’-@cpeel-eye

bee-éámpaskin-sugarcane

‘Remove the sugarcane skin.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 31: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 189

Other examples of -Vc classifying sharp objects are provided in (117) and(118).

(117) muXwaspear

d-@r-Paf-@cAFF-1S-wash-eye

‘I washed the spear.’

(118) antílaneedle

d-@r-Paf-@cAFF-1S-wash-eye

‘I washed the needle.’

Also falling under this classification and perhaps related to sharp objects orperhaps blood are wounds. In (119) below, ‘leg wound’ requires the -Vc ‘eye’classifier.

(119) m@X@b@-tS@gú-mawound-leg-1S

d-@r-Paf-@cAFF-1S-wash-eye

‘I washed my leg wound.’

Beyond the items mentioned above, the morpheme, -Vc ‘eye’, is also used forclassifiying ‘fire/light’. Verbs such as ‘blow’ must take this classifier when theO argument is ‘fire’ (120). The same is true for the verb ‘jump over’ (121).

(120) d-@r-fítS’-@cAFF-1S-blow-eye

manéafire

éé-tS1N@âá-nFUT-cook-on

kóXwasauce

‘I blew on the fire so that the sauce would cook on it.’

(121) d-@r-ampó[email protected]

m@-s1gúNMLZ-burn

manéafire

‘I jumped over the burning fire.’

Furthermore, when the classifier -Vc is marked on the intransitive verb for ‘die’Sá the valence of the verb is increased and the causative meaning ‘extinguish’is derived. This newly derived verb can only be used with fire or objects asso-ciated with light as in (122) and (123).

(122) S@-cdie-eye

b@t@riflashlight

‘Turn off the flashlight.’

(123) S@-cdie-eye

manéafire

‘Put out the fire.’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 32: Noun incorporation and predicate classifiers in Gumuz

190 Colleen Ahland

3.5.4. The classifier -1l1c. The possessive form of the classifier ‘eye’, -1l1c,fairly consistently classifies seeds, grains and seed-like objects. On verbs thatcan pragmatically take both liquids and seeds as O arguments, the possessiveform of ‘eye’ classifies seeds while the simple root form -Vc ‘eye’ classifiesliquids. This is made clear with the verb ‘smell’ in which different forms ofcoffee can be distinguished depending on which classifier is employed. In (124)below, the liquid form of coffee is the O argument. Therefore, the classifier -Vc is used whereas in (125), the bean form of coffee is the O argument whichis differentiated by the classifier -1l1c. This is not so surprising for classifiersconsidering that “agreement is with real-world referents or precepts, not withnouns per se” (Beckwith 2007: 105).

(124) fúN-1csmell-eye

bunacoffee

‘Smell the (liquid) coffee.’

(125) fuN-íl-ícsmell-POSS-eye

bunacoffee

‘Smell the coffee beans.’

3.5.5. The classifier -1k’wós. In addition to items that typically form groupsof identical objects (seeds, beads, spoons, etc.), the classifier -1k’wós ‘tooth’can also be used for animal bites as in (126) below. However, inclusion of‘bite’ may be due to the nominal compound of the O argument (see Section 3.6for further discussion).

(126) d-@r-Paf -1k’wósAFF-1S-wash-tooth

k’wós-áínéátooth-scorpion

gwá-r1-k’@N@

place-POSS-bite‘I washed the scorpion bite/sting.’

One other seemingly anomalous example is given in (127) below. It appearsas though -1k’wós ‘tooth’ is classifying ‘sauce.’ However, if one considers that‘spoon’ is a member of this class, one could analyze this sentence as classifyingthe instrument of the predicate. That is, one would likely ‘check sauce’ using aspoon.

(127) w1â-1k’wóssee-tooth

k@Xwasauce

d-@-kaX-@Z-ilAFF-3S-enter-PERF-in

Pneadirt

‘Check the sauce; dirt has gotten in it.’

3.5.6. The classifier -il1k’wós. The possessive form of ‘tooth’ classifiesitems that form groups of identical objects. Because of this semantic group-ing, many items overlap with other classes, e.g., seeds. However, this class is

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 33: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 191

broader than the class marked by -1l1c ‘seed.POSS’ in that beads and firewoodin addition to seeds are also included.28

(128) w1â-1l-k’wóssee-POSS-tooth

taNk’amillet

‘Check the millet.’

(129) w1â-1l-k’wóssee-POSS-tooth

minzabeads

‘Check the beads.’

(130) w1â-1l-k’wóssee-POSS-tooth

k’wós-1éatooth-wood/tree

‘Check the firewood.’

3.5.7. The classifier -ts’ê/-il1ts’ê. There appears to be no difference in termsof classification regarding the possessed and non-possessed forms of the incor-porated noun ts’éa ‘ear’. Both forms of the noun classify flat, flexible objects.The inclusion of mushrooms in this category came about via metonymy – asthe top of the mushroom fits this category. This is made apparent by comparingthe incorporation of this noun in the verbal root c’@S ‘cut’ (131) with its incor-poration in the verbal root Paf- ‘wash’ ((73) and (74)). In (131), the action ofthe verb only refers to the top (flat, ear-like) part of the mushroom while theverb in (73) and (74) refers to washing the whole mushroom, not just the topportion.

(131) c’ES-íl-íts’êcut-POSS-ear

waagáwhite.mushroom

‘Cut the top off of the mushroom.’

The classification of flat, flexible objects is further demonstrated with the verbnt’ ‘twist’ in (132) and (133).

(132) nt’-íl-íts’êtwist-POSS-ear

norígapaper

‘Twist the paper.’

(133) nt’-íl-íts’êtwist-POSS-ear

ts’ínéaleaf

‘Twist the leaf.’

28. Again, regarding example (130) refer to Section 3.6 on the discussion of noun-noun com-pounds.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 34: Noun incorporation and predicate classifiers in Gumuz

192 Colleen Ahland

3.5.8. The classifier -Vts. Lastly, the morpheme -Vts ‘body’ is marked onthe verb to classify humans, body parts and body-like/flesh-like entities. Onefurther example of this morpheme classifying humans is given below with theverb ‘check’ (134). However, this classifier also classifies other body-like ob-jects as demonstrated with the same verb ‘check’ in (135) and (136).

(134) w1â-1tssee-body

á@gaperson

k-á-k-ot-ESCOND-3S-APPL-exist-down

‘Check to see if there are any people.’(135) w1â-1ts

see-bodyé1-norígatree-paper

k-á-á@l@k’ák’ósaCOND-3S-write

‘Check the pen to see if it writes.’(136) w1â-1ts

see-bodysanéaplate

[email protected]

‘Check the plate to see if it’s clean.’

3.6. Predicate classifiers and nominal compounds

In Gumuz, noun-noun (N-N) compounds are quite common. When one of thesecompounds serves as an O argument, the first noun root of the compound seemsto have special influence in the choice of predicate classifier being used, espe-cially if the first noun is of the same nominal origin as that of one of the classi-fiers. This appears to be the case regardless of the type of N-N compound.

In (137) below, the object of the verb, ‘smell’ is a N-N compound meaning‘fruit’ in which the first noun root ‘seed’ is compounded with the noun root‘thing.’ Because the first nominal element of the compound is ‘seed/eye’, thepredicate classifier must be the possessive form of ‘eye’, -ílic. On the otherhand, when a specific type of fruit is mentioned, especially one that is some-what large and round, the classifier -ilúk’w ‘head.POSS’ is used (138). Both ofthese commands can be used in reference to an orange.

(137) fúN-íl-icsmell-POSS-eye/seed

c1-daseed-thing

‘Smell the fruit.’(138) fuN-il-úk’w

smell-POSS-headb@rtukánaorange

‘Smell the orange.’

Likewise, when one asks a question such as the one given in example (139)in which the initial noun root of the O argument contains a nominal that alsofunctions as a classifier, that same classifier must be marked on the verb.29

29. Example (139) may be an example of a class term compound (to be discussed in this section).

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 35: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 193

(139) abéwhich

k’wós-ídatooth-thing

l@-Paf-1k’wosPROG-wash-tooth

‘What (tooth thing) are you washing?’

Noun-noun compounds are often formed with the nominal root cá ‘eye/seed’in Gumuz. In (140), ‘seed/eye’ forms a compound with ‘millet’ and thereforerequires the classifier -Vc ‘eye’ on the verb.30 The same is true in the word for‘arrow’ which is expressed as the compound ‘seed/eye-bow’ in Gumuz (64).However, in that instance, the possessive form of the ‘seed/eye’ classifier isrequired.

(140) w1â-ícsee-eye

c@-t@Nkaseed-millet

k@-Pia-cáCOND-be-eye

éé-g@X@S-1kwáFUT-grind-1P.INCL

‘Check the millet seeds; if there are many, we will grind them.’

The noun root ‘tooth’ is also often found in N-N compounds as in (141).When the verb classifies such a compound that serves as an O argument, theclassifier -k’wós ‘tooth’ must be used.

(141) Nar-k’wóstake-tooth

k’wós-iéatooth-tree/wood

‘Take the firewood.’

One distinct category of N-N compounds in Gumuz which may hold the keyto the origins of predicate classifiers in Gumuz is that of class term compounds.According to Grinevald, “class terms are classifying morphemes which partic-ipate in the lexicogenesis of a language” (2000: 59). These are often foundwithin nominal compounds, such as those found in English, e.g., man: mail-man, policeman etc. More specifically, class terms “are morphemes which oc-cur as the head of . . . noun compounds which are exemplars of the categorylabeled by the class term” (DeLancey 1986: 438).31 In Gumuz, class term mor-phemes are often optional as the system of class terms appears to be an old sys-tem in decay. Nonetheless class terms appear to affect the choice of predicateclassifier on the verb.32 In example (82) of Section 3.3.1, the N-N compound

30. With regard to the verb ‘be many’ (140), as mentioned previously, -cá ‘eye’ can be used toindicate (large) amounts. This is perhaps related to the grammaticalization pathway eye >seed > things that come in large numbers.

31. One might be tempted to consider c@-t@Nka ‘millet seed’ (140) a class term compound usingthis definition. However, t@Nka ‘millet’ refers to the plant itself, not the seed. Thus, cá ‘eye’is not a category for millet plants; rather, it is referring to a part of the millet plant.

32. The co-existence of these two systems of classification, in addition to the agreement betweenthese, is likely to have been significant in the historic development of predicate classifiers inGumuz. In Tai, it has been documented that classifiers originated as class terms (DeLancey1986) which could very well be the case for Gumuz.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 36: Noun incorporation and predicate classifiers in Gumuz

194 Colleen Ahland

with the class term, k’wa ‘head’ combined with the noun root ‘rock’ appearedas the O argument of the verb ‘heat’. The choice of predicate classifier on theverb, therefore, was -ilúk’w ‘head.POSS’ in accordance with this class term.Many of these class terms are optional as is the case with ‘rock’. Likewise,‘bone’ can be uttered either as a class term compound or as the simple rootZ1kwa ‘bone’ without the class term k’wa ‘head’.33 With the class term com-pound, the predicate classifier -ilúk’w is required (142).

(142) g1s-1l-úk’w

burn-POSS-headk’wa-Z1kwahead-bone

‘Roast the bones.’

Other examples of class term compounds which affect the choice of predi-cate classifier also involve the class term ‘head’. In the case of ‘finger’ below,what is suspected to be a class term, k’wa ‘head’ (or possibly an old compoundin a genitival relationship), is not optional as the second noun root does notexist as a free nominal form. The word for ‘finger’ is nonetheless suspected ofbeing a N-N compound as the predicate classifier required on the verb is -Vk’w

‘head’.

(143) Paf-ók’w

wash-headk’wá-ts1tsuáhead-finger

‘Wash your fingers.’

One final suspected class term compound is found with the term for ‘cowrieshells.’ One can either say p@X1k’a ‘cowrie shell’ or k’wós-p@X1k’a ‘tooth-cowrie shell’. Again, this class term affects the choice of predicate classifier onthe verb (144).

(144) k’wós-p@X1k’atooth-cowrie.shell

d-@r-af-1k’wosAFF-1S-wash-tooth

‘I’m washing cowrie shells.’

While it is clear from the examples above that N-N compounds affect thechoice of predicate classifier on the verb, it still remains unclear as to howto predict which classifier – possessed or simple root – to use when the Oargument is a class term compound.

33. k’wa-Z1k’wa ‘head-bone’ is a class term compound and does not refer to ‘skull’ as one mightsuspect. The word for ‘skull’ requires what could be called an associative construction with‘bone’ as the head of the compound and ‘head’ following the possessive marker: Z1k’wa-lu-k’wa ‘bone-of-head’.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 37: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 195

3.7. Inconsistencies in classification

While it is possible to establish an overall system of classification for the setof predicate classifiers found in Gumuz, oftentimes classification seems to de-pend upon the host verb. That is, the set of objects that are classified by anygiven predicate classifier may be broader or narrower depending on the verb.Moreover, as presented in Section 3.3, certain verbs simply do not take certainclassifiers. Therefore, the remaining set of predicate classifiers that are allowedon these verbs are thus semantically much broader in scope in terms of theset of objects that they classify. This is best exemplified via coordination ofverbs sharing the same O argument within a clause. Each of the coordinatedverbs must take the proper predicate classifier in terms of the O argument. Itwould be expected then that if the O argument is the same, so also should bethe predicate classifiers on each host verb. However, such is not always thecase. For example, ‘clothes’, are not always classified with the morpheme -Vc,even when a series of coordinated actions all share ‘clothes’ as the O argument(145).34 Furthermore, -Vc as a predicate classifier is allowed for the verb ‘dry’(147) but the set of nouns it classifies when part of this verb is identical to thepossessive classifier -1líc which classifies seeds and grains only (146) but not‘clothes’. Other predicate classifiers are allowed for this verb as given in (148)and (149). The O argument ‘clothes’, therefore, becomes part of the class ofobjects delimited by the predicate classifier -Vts ‘body’.

(145) áXwaclothes

éé-n"S-á-c

FUT-soak-2S-eyeéé-Paf-á-cFUT-wash-2S-eye

éé-tSEP-á-tsFUT-dry-2S-body

‘After soaking the clothes, wash and dry them.’

(146) tSEP-1l-ícdry-POSS-eye

cá-ofokwacáseed-corn

‘Dry the corn kernels.’

(147) tSEP-Ecdry-eye

cá-ofokwacáseed-corn

‘Dry the corn kernels.’

(148) tSEP-íl-úk’w

dry-POSS-headb@X@g@

gourd‘Dry the gourd.’

34. One speaker of the Gumuz lect spoken in Gilgel Beles felt that the verb for ‘dry’ used inexample (145) is incorrect and should be intransitive; the proper verb, nt@X ‘dry’, should beused in which case the classifier would be consistently -Vc ‘eye’ throughout: aXwa éenSacééPafác éent@Xac.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 38: Noun incorporation and predicate classifiers in Gumuz

196 Colleen Ahland

(149) tSEP-íl-1k’wosdry-POSS-tooth

kwantSasorghum

‘Dry the sorghum.’

Likewise when the noun aPúná ‘yam’ is the O argument, the classifier useddepends on the host verb. For ‘dig’, -Vk’w must be used35, but for ‘wash’ and‘boil’ -Vc must be incorporated into the verb (150).

(150) éé-k@baréaF-K@bare

d-a-k’wóS-ók’w

AFF-3S-dig-headaPúnáyam

d-a-Paf-@cAFF-3S-wash-eye

d-a-PatS’@N-@cAFF-3S-boil-eye‘K@bare having dug the yams, washed and boiled them.’

Thus, the system of classification is not consistent across the board in Gu-muz. According to Grinevald, semantic heterogeneity of classes exists in someof the most prototypical of classifier systems (2000: 80). Furthermore, she sug-gests that the issue of prototypes is very difficult, as classifier systems tend tobe very language and dialect specific and, for verbal classifiers, the issue ofprototypes is even more difficult. Nevertheless, Grinevald places her first sub-type of verbal classifiers “at the margin of classifier systems, emerging as suchfrom a general syntactic process of noun incorporation” (2000: 83). Indeed,Gumuz does exhibit at least one characteristic of this subtype in that manyof the predicate classifiers are still recognizable as free nouns. Thus, it is notsurprising if such a “system” of classification appears to be marginal in termsof classification systems in general. Moreover, Aikhenvald suggests that it isnot uncommon in predicate classifier systems for some nouns to be associatedwith more than one classifier (2000: 149). Certainly, the semantics of the Gu-muz classification system naturally leads to overlapping categories: a class ofseed-like elements (i.e., small, numerous, etc.) and a class of elements whichare found in rows may exhibit a large number of members in common. As aresult of this overlap, such classifiers can readily absorb other members whencertain classifiers do not co-occur with certain verbal roots.

35. In the Yaso dialect of southern Gumuz, -c@ ‘seed/eye’ is used for all three of these verbs when‘yam’ is the O argument. It could be that yams were originally viewed as ‘seeds’ because theyare tubers but over time, certain Gumuz speakers saw these as being too large to be classifiedwith ‘seeds’ and thus the classifier may have become more lexicalized as ‘seed’ in some verbsbut not in others.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 39: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 197

4. Syntactic operations and predicate classifiers

There are syntactic operations in Gumuz which provide additional support forthe existence of a predicate classifier system. Promotion of arguments to thedirect object position as well as the coordination of O arguments for a singleverb lends credence to the existence of such a system in Gumuz.

4.1. Promotion of arguments

For certain verbs, one can promote peripheral arguments to the direct objectposition by marking the verb with the applicative marker -k(@). When the ap-plicative is used, the verb must take on the appropriate predicate classifier forthe newly promoted object. With the verb ‘spit’, for example, it is possible tohave the object being spit, e.g., ‘saliva’, as the O argument (151) or, one canpromote the peripheral argument with the semantic role of GOAL or LOCA-TION to the O argument position by using an applicative suffix on the verb.In such a case, it is obligatory for the verb to take the appropriate classifierassociated with the newly promoted O argument. Therefore, when ‘child’ ispromoted to the position of O argument, the classifier -Vts is used (152). Like-wise, when ‘rock’ is promoted to the O argument, the classifier -Vk’w is used(153).

(151) t@k’-@cspit-eye

gant1kóásaliva

‘Spit saliva!’

(152) duwachild

t@k’@-k-@tsspit-APPL-body

k@-gantikóáwith-saliva

‘Spit on the child (as a blessing)!’

(153) t@k’á-k-@k’w

spit–APPL-headgiSarock

k@-gantikóáwith-saliva

‘Spit (saliva) on the rock.’

4.2. Coordination of O arguments

When coordinating more than one O argument of a single verb, various strate-gies are employed with regard to predicate classifiers. If the O arguments aremembers of the same class, the O arguments can be coordinated as the object ofone verb-classifier combination (154). However, if the O arguments belong todifferent classes, one can coordinate two sentences, employing the appropriateverb-classifier combination for each O argument (155), or one can coordinatethe O arguments using the classifier of the O argument closest (in utterance) to

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 40: Noun incorporation and predicate classifiers in Gumuz

198 Colleen Ahland

the verb. For example, because ‘pumpkin’ is uttered directly before the verb in(156), the classifier -íluk’w is used. However, when ‘peas’ is uttered directlybefore the verb, -ílíc is used (157).

(154) [á@g@-ts1-m1ts’aperson-body-house

m@kina=kwé]car=CONJ

Paf-@tswash-body

‘Wash the walls of the house and the car.’

(155) [minzabead

Paf-1k’wós]wash-tooth

[aXwa=kwéclothes=CONJ

Paf-@c]wash-eye

‘Wash the beads and the clothes.’

(156) [ánt@rápeas

patúa=kwé]pumpkin=CONJ

Páf-íl-uk’w

wash-POSS-head‘Wash the peas and the pumpkin.’

(157) [patúapumpkin

ánt@rá=kwé]peas=CONJ

Páf-íl-ícwash-POSS-eye

‘Wash the pumpkin and the peas.’

5. Alternative analysis to a predicate classifier system

According to Mithun (1986: 384), it can be difficult at times to differentiatebetween Type II NI and classificatory NI. As previously stated, with Type IINI, a transitive verb incorporates its direct object, then an instrument, location,or possessor (non-core arguments) may assume the vacated object role (Mithun1984: 856). Such incorporation is similar to classificatory NI. Mithun (1986:384) gives an example in Ngandi in which an interpretation is ambiguous (158).

(158) baru-walNa-karfi-i-pulathey body chased him

nfii-wolothat

nfiiyul- /0-yuN

Aboriginal‘They pursued that Aboriginal’s body’ (or ‘They pursued that Abo-riginal’?)

Similar interpretations could be made for Gumuz. Consider in particular thepossessed incorporated nouns. Example (159) could either mean ‘Heat the ker-nels of corn’ or ‘Heat the corn (in kernel form)’. However, if one actually says‘kernels of corn’ as in (160), the classifier -il1c ‘seed.POSS’ is still used. If suchan alternation is possible, this must not be an example of Mithun’s Type II NI.At the very most, (160) is an example of predicate classification and (159) isan example of Mithun’s Type II NI. Indeed, Type II NI is a necessary stage inthe development of such a system of predicate classifiers (Mithun 1986).

(159) wok-ilícheat-seed.POSS

ofokwácacorn

‘Heat the kernels of corn’ or ‘Heat the corn in kernel form’

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 41: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 199

(160) wok-ilícheat-seed.POSS

cá-ofokwácaseed-corn

‘Heat the corn kernels.’

Likewise, the incorporated noun in example (114) above classifies the ‘cow’shead’ and does not promote ‘cow’ to O argument status as ‘head’ still formsa compound with ‘cow’ as the O argument. As Type II NI has been shown toexist in Gumuz (Section 2.3.3), it should at least be recognized that the twotypes of NI – Type II and Type IV (classificatory NI) (Mithun 1984) – coexistand in some cases, as in examples (159) and (160), there may be no difference.

6. The Gumuz system of predicate classifiers in typological perspective

As some incorporated nouns are more grammaticalized than others in Gumuz,the system of predicate classifiers has characteristics of both classificatory NIas well as verbal classifiers as affixes. The erstwhile nouns incorporated as sim-ple roots are similar to verbal classifiers as affixes while the possessed nominalforms appear to have more in common with classificatory NI. According toAikhenvald, both of these types of classifiers can coexist, reflecting differentstages of grammaticalization (2000: 149). All classifiers in Gumuz, however,have at least one characteristic of Aikhenvald’s description of classificatory NIas nouns are incorporated into verbs and the resulting complex verbal stemretains the same argument structure (2000: 160). This is generally true for allclassifiers in Gumuz save -Vts ‘body’ which can structurally decrease the va-lence of the verb when functioning as a reflexive.36 In this way, -Vts ‘body’ ismore akin to verbal classifiers as affixes. Indeed it is an affix as this classifierno longer exists as a generic noun in the language.

The classifiers in Gumuz do not fully resemble Aikhenvald’s classificatoryNI in that such classificatory morphemes should not be used in other classi-fier environments (2000: 160). That is, if one considers the relic system ofclass term compounding in Gumuz, the simple noun roots k’wá ‘head’, k’wósa‘tooth’, and cá ‘eye’, are employed in two classifier environments which ismore characteristic of Aikhenvald’s verbal classifiers as affixes.

With regard to the semantics of classifiers, Grinevald describes incorporatedclassifiers as being akin to noun classifiers, which classify according to materialessence such as ‘animal’, ‘rock’, or ‘man’ (2000: 72). The incorporated nounsin Gumuz do not appear to classify according to these categories. Rather, theyseem to classify according to shape or physical categories similar to numeral

36. As discussed previously, some classifiers, such as -Vk’w ‘head’ and -Vc ‘eye’ occassionallyincrease the valence of an intransitive verb.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 42: Noun incorporation and predicate classifiers in Gumuz

200 Colleen Ahland

classifiers which, according to Grinevald, are akin to verbal classifying affixes.Classifiers in such a system may have categories such as ‘round’, ‘long’, or‘flat’/‘flexible’ (2000: 67). It could be the case that Gumuz has an old systemof classification and thus, what were originally categories of material essence,over time, began to include other physically similar items via metaphorical ex-tension. Indeed, semantic opacity of classifiers is one characteristic of an oldersystem (Grinevald 2000: 84) which is especially true for the simple noun rootsthat are incorporated in Gumuz. Furthermore, Mithun states that classificationby quality (e.g., ‘eye’ for small round objects, etc.) rather than kind (‘water’representing all potable liquids vs. anything in a liquid state) represents thefinal extension of classificatory incorporation (1986: 390).

Within Aikhenvald’s typology, however, both classificatory NI and verbalclassifiers as affixes can classify according to shape. The difference is that clas-sificatory NI tends to characterize the referent in terms of animacy, shape, andconsistency (2000: 150) while verbal classifiers as affixes characterize the ref-erent in terms of inherent properties such as shape and size (2000: 152). What-ever the case, the system of predicate classifiers in Gumuz does not fit neatlyinto either subcategory – verbal classifiers as affixes nor classificatory NI – asdiscussed in Grinevald (2000) and Aikhenvald (2000) and perhaps representssome in-between stage of development.

7. Conclusion

Noun incorporation is a regular feature in Gumuz verbal morphology, a mor-phosyntactic reality that has, up to now, remained undocumented in previouslinguistic descriptions of the language. A small subset of these incorporatednouns function as predicate classifiers, a system of classification that was notknown to exist among languages on the African continent. Documentation ofsuch a system within Africa will likely be of great interest to language typolo-gists concerned with noun categorization and classification systems.

The Gumuz system of predicate classification involves the incorporation ofbody part nouns, at least historically and to a certain degree synchronically. Ineither case, the system appears to be a closed set of body part nouns whichrepresent partially overlapping classes. Part of the reason for the overlap islikely due to the co-existence of two types of incorporated nouns: simple rootsand their corresponding possessed forms. Both appear to have arisen fromMithun’s Type II NI, the latter having developed more specifically from posses-sor raising and perhaps more recently considering the semantic transparency ofclasses. The classes represented by the incorporated simple noun roots, on theother hand, are much more semantically opaque and tend to classify by qual-ity rather than kind, suggesting a relatively older system of classification. For

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 43: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 201

some verbs, these two types of incorporated nouns form distinct classes whilein others, these classes overlap in membership or are identical in membership.

In the absence of historical data on the Gumuz language or thorough gram-matical descriptions of the languages thought to be most closely related to Gu-muz, it is difficult to ascertain whether or not Gumuz represents an older systemperhaps in decay. If such is the case, one would expect to find evidence of sucha system in closely related Nilo-Saharan languages. If the system is young andin the beginning stages of development, one would not expect to find such asystem in closely related languages. Indeed, these predicate classifiers exist inat least four Gumuz lects, two of which are not mutually intelligible (Ahlandet al. 2002: 23). Thus, such a system must not be very young. More researchon Nilo-Saharan languages and perhaps other languages of western Ethiopiashould shed light on this question. As Nilo-Saharan languages are some of theleast described languages in Africa, description and documentation of theselanguages is even more crucial.

Abbreviations

1S first person, singular1P.INCL first person, plural inclusive2S second person, singular2P second person, plural3S third person, singularA most agent-like argument of a multi-argument clauseAFF affirmativeAPPL applicativeCENT centric (directional denoting movement toward speaker) – this can

also be used to refer to an event occurring far from the speakerCOND conditional moodCONJ conjuction

converb markerDEM demonstrativeEMPH emphasisF feminineFUT future tenseLOC locativeM masculineNEG negativeNFUT non-future tenseNMLZR nominalizerNOM nominative

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 44: Noun incorporation and predicate classifiers in Gumuz

202 Colleen Ahland

O most patient-like argument of a multi-argument clausePERF perfect aspectPOSS possessive marker (possessum)PROG progressiveRECP reciprocalREL relativizerS only argument of a single argument clauseTR transitiveV verb

University of [email protected]

References

Ahland, Colleen, Michael Ahland, and Linda Jordan. 2002. Gumuz survey report. Unpublishedreport, SIL-Ethiopia.

Aikhenvald, Alexandra Y. 2000. Classifiers: A typology of noun categorization devices. Oxford:Oxford University Press.

Beckwith, Christopher. 2007. Phoronyms: Classifiers, class nouns, and the pseudopartitive con-struction. New York: Peter Lang Publishing.

Bender, M. Lionel. 1979. Gumuz: A sketch of grammar and lexicon. Afrika und Übersee 62. 38–69.

Botne, Robert. 1993. Noun incorporation into verbs: the curious case of ‘ground’ in Bantu. Journalof African Languages and Linguistics 14(2). 187–201.

Creissels, Denis, Gerrit J. Dimmendaal, Zygmunt Frajzyngier, and Christa König. 2008. Africa asa morphosyntactic area. In Bernd Heine and Derek Nurse (eds.), A linguistic geography ofAfrica, (eds.), 86–150. New York: Cambridge University Press.

DeLancey, Scott. 1986. Toward a history of Tai classifier systems. In Colette Craig (ed.), Nounclasses and categorization, 437–452. Philadelphia: John Benjamins.

Grinevald, Colette. 2000. A morphosyntactic typology of classifiers. In Gunter Senft (ed.), Systemsof nominal classification, 50–92. New York: Cambridge University Press.

Hutchison, John. 2003. Argument structure in Bamanakan compounding. In John Mugane (ed.),The linguistic typology and representation of African languages, 189–202. Trenton, N.J.:Africa World Press.

Irwin, Lee. 1966. Some Saysay sentence patterns. Unpublished mimeograph.Keenan, E. L. 1984. Semantic correlates of the ergative/absolutive distinction. Linguistics 22. 197–

223.König, Christa. 2006. Marked nominative in Africa. Studies in Language 30(4). 655–732.Kiessling, Roland. 2007. Alagwa functional sentence perspective and “incorporation”. In Azeb

Amha, Maarten Mous and Graziano Savá (eds.), Omotic and Cushitic language studies. Pa-pers from the Fourth Cushitic Omotic Conference, Leiden, 10–12 April 2003, 187–198.

Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60. 847–894.Mithun, Marianne.1986. The convergence of noun classification systems. In Colette Craig (ed.),

Noun classes and categorization, 379–397. Philadelphia: John Benjamins.Mous, Maarten. 1993. A Grammar of Iraqw. Hamburg: Helmut Buske.Neudorf, Susanne. 2008. The use of body part lexemes in Berta. Unpublished manuscript.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM

Page 45: Noun incorporation and predicate classifiers in Gumuz

Noun incorporation and predicate classifiers in Gumuz 203

Sasse, Hans-Jürgen. 1984. The pragmatics of noun incorporation in Eastern Cushitic. In FransPlank (ed.), Objects: Towards a theory of grammatical relations, 243–268. New York: Aca-demic Press.

Tosco, Mauro. 2004. Between zero and nothing: transitivity and noun incorporation in Somali.Studies in Language 28(1). 83–104.

Uzar, Henning. 1989. Studies in Gumuz: Sese phonology and TMA system. In M. Lionel Bender(ed.), Topics in Nilo-Saharan linguistics, 347–383. Hamburg: Helmut Buske Verlag.

Uzar, Henning. 1997. Aag@m@k@š am s@ Gumuzaaç?: Introductory course in the Gumuz language.Unpublished manuscript.

Brought to you by | University of VirginiaAuthenticated | 128.143.23.241

Download Date | 9/22/12 8:23 PM


Recommended