Zambezia (1981), IX (ii).REMARKS ON SOME SYNTACTIC NOUN FEATURESIN SHONAN.C. DEMBETEMBEDepartment of African Languages, University of ZimbabweWE MAY REPRESENT the knowledge that we possess of the idiosyncratic propertiesof words as a kind of internalized dictionary, commonly referred to as a lexicon. It isa fact that, although a Shona speaker has never explicitly been told that certainverbs and adjectives may not have certain types of nouns as subject when theyoccur in deep structures of sentences,1 he nevertheless knows very well that, forexample, a noun like mombe (a cow) is not the type of noun that can occur as thesubject of the verb -verenga (read), that a noun like bold (porridge) cannot be thesubject of an adjective like -tatu (three), that the verb -nyord (write) must have assubject a noun with the property of humanness (e.g., mudzi'dzisi, teacher; mwand,a child; mukomand, a boy). Any grammar which is intended to describe ourknowledge of our language must be able to account for this type of informationamong other things. It must be able to describe why string (1) is an acceptablesentence, while (2) is not:(1) Mabhdrani dkanyord tsdmba. (The clerk wrote a letter.)(2) *Mombe ydkaverenga bhuku. (*A cow read a book.)The lexicon will contain in particular: (a) those aspects of phonological structurewhich it is not possible to predict by general rales; (b) those properties which arerelevant to the functioning of transformational rules, e.g., subject raising, objectdeletion; (c) those properties of a lexical formative which arc necessary forsemantic interpretation; and (d) lexical features which show the positions in whicha lexical formative can be inserted (by a lexical rule) in a preterminal string.Although a distinction is being made in this presentation between syntactic andsemantic features, it should be noted nonetheless that this distinction is not clear-cut. It is still a vexed question. For instance, some linguists, such as Chomsky,would regard features such as {+ count] and [± human] as syntactic, while others,such as Grinder and Elgin, would regard these same features as semantic.2 It is notyet possible to determine the exact boundary between syntactic and semantic11 am grateful for the comments of Professor G. Fortune on an earlier draft of this article.1 The view adopted in this article is that noun phrases which involve a noun and an adjectivequalifier start off as sentences in the underlying structure,2N. Chomsky, Aspects of the Theory of Syntax (Cambridge MA, M.I.T. Press, 1965);J.T. Grinder and S.H. Elgin, Guide to Transformational Grammar (New York, Holt, Rinehart andWinston, 1973).103104REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONAfeatures. Nor is it known yet what range of syntactic and semantic features isavailable in natural languages or in any particular language. For the purposes inhand, however, a feature will be considered to be a purely semantic lexical feature ifit is not referred to by any rule of either the phonological or the syntacticcomponent. A feature is syntactic if it is referred to by some transformational raleof the syntactic component. A syntactic feature will determine, for instance, whatpro-form should occur after some transformation has applied in a given type ofsentence, or whether or not a given noun can be used with a secondary prefix. Onthe other hand a semantic feature will determine what lexical item is privileged tooccur in a given frame as dictated by the other lexical items in that frame. Thedifference between phonological and syntactic features is obvious enough not torequire elaboration. Since in this study the features mentioned above in (a), (c) and(d) play an insignificant role, they will not be discussed any further.Syntactic features may influence in important ways the choice of wordsand/or their arrangement in surface structure. This will be illustrated later. Thesesyntactic features may be subdivided into various types, depending precisely on thetype of reference to a 'frame' that is implied. While some of these features forexample [gender] and [± feminine], may be regarded as inherent in the sense thatthey specify a unit without any reference to a frame, others refer to the position of aformative in the phrase structure, e.g., [± noun], [± verb]. These latter features arecalled contextual or categorical features, and they define the lexical category towhich a formative belongs. Still others specify the frame of constituents (ofcategories) in which a given formative can be inserted, e.g., + NP, Š NP,These are commonly known as subcategorization features.Syntactic features are concerned with subcategorization rather than with'branching' or 're-write' rules. It would also appear that the only categoriesinvolved in this respect are those which comprise lexical formatives as members. Ithas already been noted by McCawley that re-write rules are an inappropriatedevice to effect subcategorization of lexical categories because this subcategori-zation is not strictly hierarchic, but involves instead cross-classification.3 Syntacticfeatures will include rule specification, e.g., object raising. This rule will beapplicable only to those sentences with verbs which are positively specified withrespect to object raising.The rest of this study will be concerned with those syntactic features whichsubcategorize nouns only. These features include: [gender], [± common],[± count], [± human], [± locative], [± abstract], [± time]. I shall try to adduceevidence indicating to what extent in a grammar of Shona these features aresyntactically relevant.3 J.D, McCawiey, 'Concerning the base componeot of a transformational grammar', Foundationsof Language (1968), IV, 243-69.N. C. DEMBETEMBE1051. The feature [+N]Every noun in Shona when entered in the lexicon will be positively specified for thefeature [+ N], indicating that it belongs to the grammatical category 'noun*, asopposed to the categories 'verb', 'adjective', etc. By a lexical redundancy rale everynoun will be negatively specified for the categories 'verb', 'adjective', etc. It isobvious enough not to require discussion that lexical items specified for [+ N] willenter certain syntactic frames from which lexical items specified for {+ Verb] or[+ Adjective] will be excluded.2. The feature [gender]This is a property of every noun in Shona. It is this feature which enables nouns tocontrol concordial agreement within a noun phrase or a sentence. So much hasbeen written in Shona, and indeed in Bantu languages as a whole, aboutgrammatical concord which is dictated by nouns that it would be superfluous todwell on it again here. The question of number is, however, another matter. Giventhe system of genders in Shona, whether the singular/plural feature can be said tobe syntactically significant is in my view open to question. It would have beenpertinent to dwell on this aspect at some length, but it is not possible in this shortpaper.3. The feature [± common]In order to appreciate the relevance of this feature in Shona it may be pertinent todwell a little on primary and secondary prefixes. A prefix is considered to beprimary with respect to a given noun if it is the usual or normal form that occurs withthat noun when signifying a normal specimen of the phenomenon or object beingreferred to. For example, the stem -komand normally appears with either[gender 1 ] or [gender 2|. The nouns mukomana (aboy) and vakomand (boys) referto a normal boy and normal boys respectively. There is no connotation of size orother quality implied.However, it is quite common for a noun to possess a gender with which it is notnormally associated. This happens when it is intended to supply information aboutsome quality of the object referred to. That is, in addition to stating the type ofspecimen it is, something is also said about its size or other quality. For example,the stem -komand, in addition to occurring with {gender 1] and [gender 2], canalso occur with the following genders:(3) gender 5 Š gomand (a huge boy)7 Š chikomand (a short and stout boy)11 Š rukomand (a thin, emaciated boy)12 Š kakomand (a small boy)14 Š ukomand (boyhood)Thus the genders 5,7,11,12 and 14 are used in secondary association in respect of106 REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONAthe noun -komand.4 The assumption made here is that primary prefixes willappear in deep structure while their secondary function will be accounted for by atransformational process.Fortune has subdivided secondary prefixes into two groups, namely,commentary and supplementary prefixes. The former indicate special orabnormal specimens of the nouns referred to and they include genders 5, 6, 7,8,11,12 and 13. The latter do not specify an object as abnormal but introduce a setof new and supplementary references; among those genders are la, 2a, 7, 14,16, 17 and 18.The feature common/non-common appears to be justified in a grammar ofShona on the grounds that, although sometimes two nouns may share the samegender among their features, one of them may take on a gender in secondaryfunction because it is a common noun while the other may not, simply because it isnot a common noun. For instance, the nouns Sard (the name) and sekuru (anuncle) each have [gender la]. But whereas we can say kasekuru (12) (an under-sized uncle) we cannot normally say kaSard (12) (a small Sara).5 Consider thenoun Harare as another illustration. This noun may have as one of its features[gender 9]. As with Sard above we cannot say, for example, ruHardre (11),whereas with a noun like mombe (a cow), which also has [gender 9], we can happilysay rumombe (11) (a thin under-sized cow). The nouns sekuru and mombe caneach occur with a gender expressing a secondary idea, namely, genders 12 and 11respectively in this case, because they are common nouns, and the nouns Sard andHarare cannot, simply because they are not common nouns.Given the division of secondary prefixes into those that are commentary andthose that are supplementary, we observe that the feature [± common] is relevantonly in respect of secondary prefixes of the commentary type. Hereunder arefurther examples involving commentary prefixes:(4) a. With common nounsPrimary association Secondary associationbadzd (5) (a hoe) chipadzd (7) (a worn-out hoe)rupadzd (11) (a despised hoe)kapadza (12) (a little hoe)zibadzd (21) (a big hoe)chipunu (7) (a spoon) kapunu (12) (a small spoon)zipunu (21) (a big spoon)mhuru (9) (a calf) chimhuru(l) (a fat calf)rumhuru (11) (an undernourished calf)4 For primary and secondary prefixes see G. Fortune, 'The references of primary and secondarynoun prefixes in Zezuru', African Studies (1970), XXIX, 81-110.5 The figure which occurs immediately after a noun in this article indicates the gender or noun classto which that noun belongs.N, C. DEMBETEMBE107kamhuru (12) (a small calf)zimhuru (21) (a big calf)b. with non-common nounsPrimary association Secondary associationZiize (la) (a name) *chiZuze (7)*ruZuze{\\)*kaZuze (12)Domboshava (5)(name of a hill)Zambezi (la)(name of a river)Łz/Ziize (21)*chiDomboshava (7)*ruDomboshava (11)*kaDomboshava (12)*chiZambezi (7)*ruZambezi (11)*kaZambezi (12)*ziZambezi (21)However, supplementary prefixes can be used with both common and non-common nouns as illustrated below:(5) a. With common nounsPrimary associationdandd (5) (a log)murombo (1) (a poor man)b. With non-common nounsPrimary associationAH (la) (a name)Herd (9/17a)(name of a tribe)Secondary associationDandd (la) (a name)vdDanda (2a) (Mr Danda)paddndd (16) (at the log)vdMurombo (2a) (Mr Murombo)urombo (14) (poverty)mumurombo (18) (in a poor person)Secondary associationvaAU{2&) (Mr AH)chiAH (7) (in the manner of AM)muHerd (1) (member of Heratribe)uHerd (14) (Buhera district)chiHerd (7) (in the manner of theHera tribe)108REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONA4. The feature [± count]4.1. Now consider the nouns muti (a tree) and muto (soup), each of which hasamong its features [gender 3]. The noun muti can be accompanied by the numeralmumwe (one, single) as in muti mumwe chete (one tree). But this is not possiblewith muto. The string muto mumwe chete when taken to mean 'one or single' soupis ungrammatical. This string is, however, grammatical only if itmeans 'same soup'(i.e., the same as another). This latter meaning is also applicable to the string mutimumwe chete (the same tree). Take another example: the nouns mago (wasps) andmamera (sprouted grain) have each the feature [gender 6]. But whereas wecan say;(6) mago maviri (two wasps)mago matatu (three wasps)mago mana (four wasps)we cannot say:(7) *mamera maviri*mamera matatu*mamera mana, etc.Although in the case of muti and mago we can speak of the one as a singular nounand the other as a plural noun, we do not have a plural noun which corresponds tomuto nor do we have a singular noun which corresponds to mamera. The nounsmuti and mago are count nouns while muto and mamera are non-count nouns.This seems to confirm the syntactic relevance of the count/non-count feature inShona. This argument relies on the behaviour of nouns with numeral adjectives.4.2 It is noteworthy also that nouns like muto, mamera and rudo (love) maynot be used with singular secondary prefixes of the commentary type, e.g.:(8) muto *chimutomamera *chimamerarudo (love) *chirudomvitrd (water) *chimvurdchando (cold) *chichandoBut such nouns as musikand (a girl) and badzd (a hoe) may be used readily withsingular secondary prefixes of the commentary type, e.g.:*rumuto*rumamera*rurudo*rumvurd*ruchando*kamuto*kamamera*karudo*kamvurd*kachando*zimuto*zimamera*zirudo*zimvitrd*zichando(9) musikandbadzdchisikandchipadzdrusikandrupadzdkasikandkapadzdzisikandzibadzdThe point which is being made here is that nouns in the former group may not beused with singular secondary prefixes because they are non-count nouns, whereasthose in the latter group can be so used because they are count nouns.N. C. DEMBETEMBE1095. The feature [± human]5.1. In asking certain questions in Shona the interrogative formatives used willdepend on whether the noun phrase being questioned is human or non-human. If itis the former the interrogative formative employed is ani (who, whom), and chii(what) if the latter, e.g.:(10) a. Sekai dkatsvodd Pawakwenyewa. (Sekai kissed Pawakwenyewa.)b. Sekai dkatsvodd ani? (Sekai kissed whom = Whom did Sekaikiss?)(11) a. Vakomand vdkaurdyd nyoka. (The boys killed a snake.)b. Vakomand vdkaurdyd chil? (The boys killed what = What didthe boys kill?)(12) a. Kateya dkapd Shaya bhora. (Kateya passed to Shaya the ball =Kateya passed the ball to Shaya.)b. Kateya dkapd ani bhora? (Kateya passed to whom the ball = Towhom did Kateya pass the ball?)c. Kateya dkapd Shaya chil? (What did Kateya pass to Shaya?)Note, however, that in the case of those non-human nouns which refer to place ortimeŠin other words, adverbials of place or timeŠthey will not be substituted forby chii.6 These will be discussed later. The interrogative formative is chosen, notaccording to concord relationship with the object concerned, but in accordancewith the expectation of syntactic category in the answer. The determining factor ofsyntactic subcategorization in this case is the human/non-human distinction.If it is an object, direct or indirect, which is being questioned, that object issimply substituted for by ani or chii as the case may be. This is exemplified in(10)Š(12) above. If it is a subject noun phrase that is being questioned, the sameinterrogative pro-forms are used but the sentence is modified a little. In addition tosubstituting for the string under consideration, the interrogative pro-form isstabilized, or, in other words, turned into a predicate with the rest of the sentencebecoming a relative clause, as shown by the 'b' sentences of the examples whichfollow:(13) a. Chipo dkabdtwd nemapurisa neziiro. (Chipo was arrested by thepolice yesterday.)b. Ndimni akdbatwa nemapurisa nezuro? (lit., it is who-the onearrested-by the police-yesterday = Who was arrested by thepolice yesterday?)(14) a. Ndege inofdmbd nemudenga. (An aeroplane travels in the air.)b. Chii chinofdmbd nemudenga? (lit, it is what-which travels-inthe air = What travels in the air?)6 In Shona, adverbials of place and time are in effect noun phrases.110REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONA5.2. It is observed also that when two or more subject noun phrases or objectnoun phrases which refer to human beings are conjoined, the concord which theydictate and control together is usually that of [gender 2]. In the case of object nounphrases this concord is required sometimes as a substitute referent or as a co-referent. Examples (15 )-(20) refer to conjoined subject noun phrases while (21)-(23) pertain to co-ordinated object noun phrases. (In these examples the numberafter a noun phrase indicates the gender or class of the head noun.)(15) Mukomand (1) nemusikand (1) van' kutaurirana. (A boy and a girlare talking to each other.)(16) Sekuru (la) natete (la) vknoiiyd mangwdna. (Uncle and aunt willcome tomorrow.)(17) Narume (2) nevakddzi (2) vkchdenda kumusangano. (Men andwomen will go to the meeting.)(18) Sekuru (la) nemuzukuru (1) vknoddnd, (An uncle and his nephewlove each other.)(19) Murimi uyu (1) nemadhomeni (6) vknonzwdnd. (This farmer andthe agricultural demonstrators get on well.)(20) Shdmwariydngu (9) nehdrahwa iye (9) vkkdtosvorana. (My friendand that old man quarrelled.)(21) a. Ndaona hanzvddzi (9) nehanzvadzi (9). (I saw a brother and asister.)b. Ndavkona. (I saw them.)(22) a. Ticha dpa vakomand (2) nevasikand (2) zviwitsi. (The teachergave both boys and girls some sweets.)b. Ticha dvapd zviwitsi. (The teacher gave them some sweets.)(23) a. Takdsdngana naJo (la) naMatenzeni (la) kwdMachipisa. (Asfor Jo and Matenzeni we met them at Machipisa.)b. Jo naMatenzeni takdsdngana ndvo kwdMachipisa. (As forJo and Matenzeni we met them at Machipisa.)But when two or more subject noun phrases or object noun phrases which donot refer to human beings are conjoined the concord which they usually controltogether is [gender 8]. The examples in (24)-(26) pertain to conjoined subject nounphrases and those in (27)~(28) to conjoined object noun phrases.(24) Pasi (16) nedenga (5) zvkkdsikwa naMwari. (Earth and heavenwere created by God.)(25) Makudo (6) nembdda (10) zvkkavengdnd. (Baboons and leopardsare enemies.)(26) Bere (5) nedhongi (5) hazvidi kuonand. A hyena and a donkey donot want to see each other.)N. C. DEMBETEMBE111(27) a. Tinofudza mombe (10) nehwdi (10) pamwe chete. (We let cattleand sheep graze together.)b. Mombe nehwdi tinozvifudza pamwe chete. (As for cattle andsheep we let them graze together.)(28) a. Waona kambwdnand (12) nemhuru (9) kupi? (lit, you saw apuppy and a calf where = Where did you see a puppy and acalf?)b. Wazviona kupi ? (lit., you saw them where = Where did you seethem?)In the strings (15)Š(28) the choice of either concord va- or zvi-depends on whether the conjoined noun phrases refer to human beings or not. Tothat extent the human/non-human distinction is syntactically significant.Occasionally, however, conjoined noun phrases which refer to human beingsmay control together either [gender 2] or [gender 8] as illustrated below:(29) Murimi uyu (1) nemhdhomeni (6) zvinonzwdnd. (cf. string (19).)(30) Shdmwari ydngu (9) nehdmhwa iye (9) zv&kdtosvorana. (cf.string (20).)The following strings, however, are ungrammatical:(31) *Sekitru (la) natete (la) zvinouyd mangwdna. (cf. string (16).)(32) *V&rume (2) nevakddzi (2) zvichdenda kumusangano. (cf. string(17).)Further investigation is required here to determine when [gender 8] may or may notbe used optionally with conjoined human noun phrases. What is uncontroversial,though, is that when human noun phrases are co-ordinated they control [gender 2]concord. That is, no cases have been found in which conjoined human noun phraseswill control [gender 8] concord to the exclusion of [gender 2] concord.When two non-human noun phrases, both plural and belonging to the samegender and semantic class, are joined together, they may optionally control theconcord of their gender.(33) a. Ndakdtema mlsasd (4) nemlnhondo (4) ya/vd mumunda.b. Ndakdtema mlsasd neminhondo zva/vd mumunda. (I cut downthe musasa and munhondo trees which were in the field.)It is significant that non-human noun phrases, unless they are personified, willnever control the concord of [gender 2].What is intriguing, though, is a conjoined structure which involves a humanand a non-human noun phrase. If a choice of gender referring to the two nounphrases together has to be made, it will have to be that of [gender 8] rather than of[gender 2]:112REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONA(34) Muvhimi (1) nembwa ydke (9) zvakdtsdkatika. (The hunter and hisdog disappeared.)(35) *Muvhimi (1) nembwa ydke (9) v&kdtsdkatika.(36) Ndakdona mukomand (1) nemombe dzdke (10) nmchinetsdna.(lit., I saw a boy and his cattle troubling one another == I saw aboy being given a hard time by his cattle.)(37) *Ndakdona mukomand (1) nemombe dzdke (10) vachinetsdnd.Normally we tend to avoid conjoining such noun phrases. The human noun phraseis given the privilege of controlling concordial agreement while the non-humannoun phrase is made an adverbial of some sort, usually a prepositional phrase:(38) Muvhimii kkdtsdkatika pamwe chete nembwa ydke. (The hunterdisappeared together with his dog.)(39) Ndakdona mukomand achinetsdna nemombe dzdke. (I saw a boyhaving trouble with his cattle.)It is noteworthy nevertheless that when two or more noun phrases areconjoined the concord which they control together is usually either that of [gender2] or [gender 8], the former with human noun phrases and the latter with non-human noun phrases.6. The feature [± locative]6.1. There are certain verbs in Shona which require the presence of a locativenoun phrase as an indirect object. To this end, consider the following sets ofsentences:(40) a. Tinoisa shuka mubota. (We put sugar into porridge.)b. Tinoisa mubota shuka. (We put into porridge sugar.)c. Tinoisa bota shuka. (lit., we put porridge sugar == we put intoporridge sugar.)(41) a. Simbi dkapd bhora kuna Kuda. (Simbi passed on the ball toKuda.)b. Simbi dkapd kuna Kuda bhora. (Simbi passed on to Kuda theball.)c. Simbi dkapd Kuda bhora. (lit., Simbi passed on Kuda the ball =Simbi passed on to Kuda the ball.)(42) a. Ndakwereta man kushamwari yangu. (I borrowed some moneyfrom my friend.)b. Ndakwereta kushamwari yangu mari. (I borrowed from myfriend some money.)c. Ndakwereta shamwari yangu mari. (lit., I borrowed my friendsome money = I borrowed from my friend some money.)N. C. DEMBETEMBE113The meanings of the sentences in each set are the same. The parts printed in boldtype are locatives. In the 'a' sentences the locatives occur in a position after thedirect object, but in the 'b* sentences these same locatives are placed immediatelybefore the direct object.7 Although the 'b' sentences are not altogether rejected asungrammatical, nevertheless they have a jarring effect. In other words, they arelower than 'a' sentences on the acceptability scale. On the other hand, if the locativenoun phrases in the 'bs sentences have their locative features removed, thesesentences become quite acceptable, as in V. I shall refer to the transformationwhich has moved the locatives in the 'b' sentences as the Indirect Object Movementrule. The point being made here is that this rale is usually accompanied by thedeletion of the locative feature. Notice also that in, for instance, (40)a, if thelocative feature is deleted without moving the indirect object, the resulting string isan unacceptable sentence in the writer's dialect:(43) *Tinoisa shuka bold. (We put sugar porridge.)This seems to suggest that the locative/non-locative feature is syntacticallysignificant. In (41 )a and (42)a if the locative feature is shed, the strings which resultare, however, not totally unacceptable, but are lower on the acceptability scale.6.2. In Shona there is a class of verbs which requires locatives as direct objects.In the examples which follow the locatives are set in bold type. If the locativefeature is reduced, the strings become ill-formed, as indicated by the 'b' sentences:(44) a. Mhungii ydpinda muguru. (The black mamba slipped into a holein the ground.)b. *Mhungu ydpinda guru.(45) a. Vand vdnoenda kumiisha mangwdna. (The children will go hometomorrow.)b. *Vand vdnoenda mushd mangwdna,(46) a. Tinosvika pachlkomo icho zuvd richinyurd. (We shall arrive atthat hill at sunset.)b. *Tinosvika chikomo icho zuvd richinyurd.(47) a. Vanhu vese vdkanzi vduye kumiisangano. (All the people weretold to come to the meeting.)b. *Vanhu vese vdkanzi vduye musangano.As has been pointed out already above, this class of verbs requires as directobjective a locative noun phrase. If the locative complement is left out in (44)the result is the ungrammatical sentence in (48):(48) *Mhungu ydpinda. (*The black mamba entered.)7 It is assumed in this article, but not proved because it is not crucial to the discussion in hand, thatin the underlying structure the direct object comes before the indirect object.114REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONAThis shows that the verb -pinda must be followed by a complement. Notice thatsentences (45)-(46) would be grammatical even if the locative complement is leftout. This is illustrated in (49)-(51):(49) Vand vdnoenda mangwdna. (The children will go tomorrow.)(50) Tinosvika zuvd richinyurd. (We shall arrive at sunset.)(51) Vanhu vise vdkanzi vduye, (All the people were told to come.)This is so because verbs like -enda, -svika and -uyd also belong to anothersubclass of verbs which permit the optional deletion of the object.Note that noun phrases which function as adverbials of place will also bespecified for the feature [locative].The locative feature which is spelt out as ku, i.e. [gender 17], appears to beredundant with some locative nouns like kumhiri (on the bank of a river),kuzasi (down below), kushure (behind), kumberi (in front), and place-names,which are on the whole not named after people, e.g. kuDomboshava, kuHardre,kuMazowe. These locatives can also occur simply as mhiri, zasi, shure, mberi,Domboshava, Harare and Mazowe, respectively. Places which take their namesfrom people have in place of ku the possessive kwa, which is not deletable,e.g. kwaMrewa, kwdMtoko, kwdMatdmbo. It is interesting to note that someplace-names which were originally named after some people have through thepassage of time changed the kwa to ku, e.g. kuHardre, kuSinoia.8 One name whichis still in the process of undergoing that change is Mukdro (a place in GutuDistrict). The concord used in all these cases is that of [gender 17], e.g.:(52) kumberi uko cf. mberi uko (in front there)kuzasi kwdZambezi cf. zasi kwdZambezi (the lower reaches of theZambezi)kushure kwedu kuri kunaya cf. shure kwedu kuri kunaya (behind usit is raining)kuSinoia ndiko kune guva rdbabd cf. Sinoia ndiko kune guvardbabd (father's grave is at Sinoia)This optional deletion of the locative feature does not happen in the case of theother two locative features, namely, pa and mu.1, The feature [± abstract]Non-count nouns may be subdivided into two groups on the basis of their use withsome prefixes in secondary function. The relevant secondary prefixes here are zvi-of [gender 8] and tu- of [gender 13]. The prefix zvi- here has the meaning8 Harare was named, according to one version, after a man who lived on Salisbury Kopje. It is saidthat he kept a fire burning on all nights at his village, and so people nicknamed him Haarare (he does notsleep), which was later corrupted to Harare.The town of Sinoia got its name from Chief Chinoi, who used to live there a long time ago.N. C. DEMBETEMBE115'despised, contemptible', while tu- means ' a little amount, not much'. The firstgroup comprises those non-count nouns which can be used with either zvi-or tu-:(53) munyii (3) (salt) zvimunyii tumunyudovi (5) (groundnut butter) zvidovi tudovimaburo (6) (boiled mixture zvimaburo tumaburoof malt and water)shuka (9) (sugar) zvishuka tushukarupiza (11) (relish made zvirupiza turupizafrom ground, roastedcow-peas)wp/w (14) (mealie meal) zviupfu tuupfuThe nouns in the second group do not allow these two prefixes generally:(54) rudo (11) (love)ngonono (10) (snoring)tsviyo (10) (epilepsy)mahayo (6) (pain in side)nzara (9) (hunger)havi (9) (craving)*zvirudo*zvingonono*zvitsviyo*zvimabayo*zvi nzara*zvihavihdnganwd (10) (forgetfulness) *zvihdnganwd*turudo*tungonono*tu tsviyo*tumabayo*tunzara*tuhavi*tuhdnganwdIt is observed that the nouns in the latter group are all abstract in nature, while thosein the former group are non-abstract. They are concrete nouns. To this extent itwould appear that the abstract/non-abstract feature is significant in Shona.However, the latter group is not very tight. While, for instance, nouns like hope (10)(sleep) and hdshd (10) (anger) are abstract in nature, they nevertheless can occurwith the prefix tu in secondary association, as in tuhope (12) (little sleep) andtuhdshd (12) (little anger), but these nouns cannot occur with the prefix zvi-:*zvihope, *zvihdshd.Notice also that there are some nouns which, though incorporeal in nature, arenevertheless count nouns syntactically, e.g.:(55) pfungwd imwe chete (9) (one idea)pfungwd mbiri (10) (two ideas)mweyd mumwe chete (3) (one soul)mweyd mitatu (4) (three souls)ngiivd imwe chete (9) (one time)ngiivd zhinji (10) (many times)Such is the nature of the language.8. The feature [± time]It would appear that the head nouns of noun phrases which function as adverbials of116 REMARKS ON SOME SYNTACTIC NOUN FEATURES IN SHONAtime should be considered to have as one of their cluster of features the feature[time].That time adverbials as a grammatical category in Shona are noun phrases isshown immediately below. First, the head noun can be qualified in the usual way,e.g.:(56) mangwdnani dno (6) (this morning)nhdsi uno (la) (lit., today this = today)Secondly, they can control concordial agreement in clauses and sentences:(57) gore rakdpera (5) (lit, the year which ended = last year)manheru adkauya (6) (the night on which he came)When an adverbial of time in a sentence is questioned the interrogativeformative used is rinhi (when). The choice of rinhi rather than any otherinterrogative formative is determined by whether or not the NP being questionedhas the feature [time]. In the examples which follow, the time adverbials have beenset in bold type:(58) a. Vanhu vdnoenda kumusangano mangwana. (People go to themeeting tomorrow.)b. Vanhu vdnoenda kumusangano rinhi? (lit., people will go to themeeting when = When will people go to the meeting?)(59) a, Tinovhara chikoro muni Zvita. (We close school In December.)b. Tinovhara chikoro rinhi? (When do we close school?)(60) a. Tsitsi dkdberekwa gore rakipera. (Tsitsi was born last year.)b. Tsitsi dkdberekwa rinhi? (When was Titsi born?)When referring to a specific time within the day or the month or the year, rinhimay be replaced by ngitvai (at what time):(61) a. Uchdenda kudhorobha nhisl masikati. (You will go to towntoday In the afternoon.)b. Uchdenda kudhorobha nhasi nguvai ? (You will go to town todayat what time?)c. *Uchdenda kudhorobha nhasi rinhi?(62) a. Chitima chinosvika manhera. (The train will arrive in theevening.)b. Chitima chinosvika rinhi? (When will the train arrive?)c. Chitima chinosvika nguvai ? (The train will arrive at what time ?)(63) a. Chitima chinosvika nhasi. (The train will arrive today.)b. Chitima chinosvika rfnhi? (When will the train arrive?)c. *Chitima chinosvika nguvai?N, C. DEMBETEMBE117Notice that (63)c by itself would be grammatical, but considered as being relatedboth structurally and semantically to (63) a and (63)b it is ungrammatical.CONCLUSIONIn this study I have tried to show the noun features which are syntacticallysignificant in Shona. These are: gender, common/non-common, count/noe-count,human/eon-human, locative/non-locative, abstract/non-abstract, and temporal/non-temporal. These features will influence the syntax of sentences in Shoea in oneway or another. It may be concordial agreement, use with secondary prefixes, orwith a certain type of adjective, or the choice of formatives with certain types ofsentences. All this information should be accounted for in any grammar of Shonawhich purports to be descriptively adequate.