PDF hosted at the Radboud Repository of the Radboud University Nijmegen The following full text is a publisher's version. For additional information about this publication click this link. http://hdl.handle.net/2066/14630 Please be advised that this information was generated on 2017-06-14 and may be subject to change. C a n a d i a n J o u r n a l of L i n g u i s t i c s / R e v u e c a n a d i e n n e d e L i n g u i s t i q u e 3 3 ( 4 ) : 4 0 9 - 4 2 2 D e c . / d é c . 1988 Media Lengua and Linguistic Theory P I E T E R M U Y SK E N U niversiieit vaji A m s te r d a m In this p a p e r I address the question w h a t M edia L eng ua can tell us a b o u t the distinction between stem s a n d affixes, the n a t u r e of lexical en tries, and the relation between the lexicon, sy n tax , and phonology. It is p a r t of a m uch larger a t t e m p t , coming from a variety of subdisciplines to pro vide w h a t is so m etim es term ed external evidence for p a rtic u la r a s su m p tio n s a b o u t the n a t u r e of the h u m a n language faculty. M edia L e n g u a is a linguistic variety spoken b o th as a native and as a second lan g u ag e by a c c u ltu r a te d In d ia n peasants, craftsm en, and con struc tion workers in C e n tra l E c u a d o r (M uysken 1981b). It is essentially Q u e ch u a w ith th e vast m a j o r i t y of its stem s replaced by Spanish forms. T h is pro cess of rep lacem ent is c o m m o n ly called “relexification” . E x am p les of M edia L e n g u a u tte ra n c e s are given in (1) th ro u g h (3): (1) a. unu fabur- ta pidi- nga- bu bini- xu- ni (M edia Lengua) one favor AC ask NOM for come P R 1 ‘1 come to ask a favor.’ b. shuk fabur-da m ana-nga-bu sham u-xu-ni (Quechua) c. vengo p a ra pedir un favor (Spanish) It is clear t h a t ( l a ) has resulted from p u t ti n g the phonological shapes of the words in ( l c ) into the lexical entries in ( l b ) . Several things should be noted. First, we get an e m p h a t i c form of the indefinite article in M edia Lengua, u n u , r a t h e r t h a n S panish u n e m p h a tic un. Second, the Spanish irregular verb form vengo a p p e a rs in a regularized ste m form bini. T h ir d , the Q u e c h u a rule voicing accusative case -ta to -da after fa b u r has not applied in M ed ia Lengua; Q u e c h u a dialectological evidence suggests t h a t this is a very recent rule. F o u rth , w h a t is peculiar a b o u t M edia L engua is n o t so much t h a t it contains Spanish words (m a n y dialects of Q u e ch u a do as well), b u t t h a t all Q u e c h u a words, including all core vocabulary, have been replaced. F ifth, the Spanish forms have been a d a p te d phonologically 409 410 C J L / R C L 33(4), 1988 to Q uechua; mid vowels have been replaced by high vowels. Q uechua word order and m o rp h o lo g y have been retained. (2) a. kuyi-buk yirb a nuwabi -shka (Media Lengua) cavia for grass there is not SD ‘T h e re tu rn s out to be no grass for the cavias.’ b. kuyi-buk k ’ivva illa-shka (Quechua) c. No hay hierba p a ra los cuyes (Spanish) Note t h a t th e Q u e c h u a word kuyi a p p e a rs in the local Spanish as well. T h e M ed ia L en gu a verb m a in ta in s the Quechua-specific “sudden discovery tense” m a r k in g -shka. T h e Q u e c h u a negative existential verb stem ilia- has been relexified with a newly formed “frozen” stem nuwabi-, derived from S panish no a n d haber ‘h a v e ’. T h e Spanish verb ‘h a v e ’ has an impersonal form hay which also has existential m eaning. (3) a. yo- ga awa -bi kay -m u -ni (M edia Lengua) I T O w ater LO fall CIS 1 II come after falling into the w a te r .’ b. iiuka-ga yaku -bi urm a-m u -ni (Quechua) c. vengo despues de caer en el agua (Spanish) E x a m p les such as (3) show the e x te n t to which M edia L en gu a follows Q u e c h u a verb sem antics. Cislocative -m u- can be a tta c h e d to n o n -m o v e m e n t verb stem s in d ic atin g t h a t the s u b je c t comes after some action. T h is pos sibility exists in b o th M ed ia L en gu a and Quechua. W h a t ex am ples (1) to (3) show us is th a t: (a) M edia Lengua is essentially the p ro d u ct of replacing Q uechua phonological shapes with Spanish forms, m aintaining the rest, of the Q uechua s tru c tu re (b) the Spanish forms chosen have undergone regularization and a d a p ta tio n to Q uechua morphophonology (c) Media Lengua is conservative in som etim es reflecting earlier stages in Q uechua pronunciation. It is not m ade up on the spot every time it is spoken (d) T h e occurrence of Spanish strong alternants, frozen composites, etc. is an indication t h a t the process of vocabulary replacem ent is not a simple one. We now tu r n to th e im plicatio n s of the relexification process o p e r a n t in M edia L engua for our view of the relation between the lexicon an d the o th e r c o m p o n e n ts of the g r a m m a r . In the first section I look at the word fo rm a tio n c o m p o n e n t. In section 2 lexical sem antics and s y n ta x are dis cussed. Section 3 deals with the issue of w h a t external evidence can be in theoretical linguistics. T h e im plicatio n s of M edia L eng u a relexification for our view of the in te ractio n between the lexicon and the phonology lie beyond th e scope of this article. P I E T E R MUYSKEN 411 1. Roots versus Affixes, IA versus IP T h e first issue on which M ed ia L en gu a d a t a m ay be able to throw some light is the re p resen tatio n of affixes as p a r t of word form ation rules or as s e p a r a te lexical entries. T h is issue has a long history in s tr u c tu r a l linguistics (R o c k e tt 1955), where it was discussed in term s of I te m - a n d - A r r a n g e m e n t (IA) versus Ite m - a n d - P r o c e s s (IP). In the IA view, words are con stru cted from build ing blocks which all have equal s t a t u s in the theory of the lexicon, except t h a t some are specified as roots, others as affixes, etc. Consider a Q u e c h u a form such as (4): (4) ri - naku - nga - mi go R E C - P L 3FU AF ‘T h e y will (certainly) go.’ In th e IA view, th e Q u e c h u a lexicon contains, am o n g others, the entries in (5): (5) ri naku nga mi STEM A FFIX A FFIX C L IT I C go reciprocal or conjoint action third person future affirmative T h e se are th en com bined by “a r r a n g e m e n t ” rules to form com plete words as in (4). T h u s the a r r a n g e m e n t rules for Q u e c h u a will specify t h a t roots go to the left, clitics to the right, a n d affixes, in a certain order, in between. In th e IP view, words are c o n s titu te d m ostly o u t of o th e r words, th ro u g h word fo rm a tio n processes. In t h a t view, there will be one lexical entry, and two affixation rules, to derive (4): (6) ri ‘go ’ rule 1: rule 2: V —►V 4* naku V —►V + nga ‘conjoint or reciprocal a c tio n ’ ‘third person future form of a c tio n ’ (To the s t a t u s of -m i in the IP account we re tu rn below). W i t h i n generative m orphology, which initially inherited the preoccu p a tio n w ith R U L E S from th e Aspects (C h o m sk y 1965) and S o u n d P a ttern s ( C h o m s k y a n d Halle 1968) model, the first e la b o ra te t r e a t m e n t was w ithin the IP t r a d i t i o n (Aronoff 1976). O nly later do we find a shift tow ards the IA perspective, in L ie b e r ’s (1981) thesis and, m o st explicitly, in Selkirk (1982). In th e la tte r, word fo rm a tio n is s ta te d in term s of phrase s tr u c tu re rewrite rules. M c C a r t h y ’s (1981) analysis of Semitic vowel a lte rn a tio n s a n d M a r a n t z ’s (1982) t r e a t m e n t of reduplication, p n m a facie th e stron gest cases for an IP m odel, in term s of a b s t r a c t m o rp h e m e s on s e p a ra te p h o n o logical levels then provided the basis for a wholesale shift to the IA model. T h i s shift was m a d e possible by the m o d u la riz a tio n of the theory to the e x te n t t h a t the p u t t i n g to g e th e r of m o rp h e m e s (th e realm of m orphology) 412 C J L / R C L 33(4), 1988 was strictly s e p a ra te d from the way these m o rp h e m e s are then pronounced to g eth er (th e realm of phonology). W h a t evidence does Media Lengua have to bear on this issue? Consider first the way (4) would appear: (7) i - naku - nga - mi go R E C - P L 3FU AF ‘T hey will (certainly) go.’ Notice t h a t only the Q u e c h u a stem ri- has been replaced by i-. T h e Q u e c h u a affixes rem ain. A brief glance at the exam ples in (1) to (3) will give the sam e result: only roots are relexified, not affixes. In a theory in which roots and affixes are b o th lexical entries, differing only in t h a t affixes obligatorily have a s u b c a te g o riz a tio n feature, this result is not easily explainable. In a theory, however, in which affixes are introduced by linguistic rules, this is w h a t we would expect. C a n we claim, then, t h a t M edia Lengua data, provide external evidence favoring an IP account of word fo rm a tio n ? I do n o t th in k the evidence is so s tra ig h tfo rw a rd as t h a t . Several issues need to be considered. (1) Do we not find any cases of affix relexification? (2) Could it not be for in d ep en d en t, p e rh a p s sem antic, reasons t h a t affixes are not relexified? (3) W h a t a b o u t clitics? If they are n ot relexified, w h a t does this m ean for the IA versus IP controversy? In the following subsections I will take up these issues in tu rn . 1.1 Are no affixes relexified? I m en tio n ed above t h a t no affixes were relexified. T h is is not entirely true. We find the Spanish s u b o r d i n a t i n g gerundial affix -ndu (originally n d o ) in M edia Lengua, in places where in Q u ech ua th e “sam e s u b j e c t ” s u b o r d i n a t e -sha a p p e a rs or the “different s u b je c t ” s u b o r d i n a t o r - k p i . M edia L engua shows all three forms. In (8) and (9) I illu strate “s a m e s u b je c t ” s u b o r d in a tio n w ith -ndu and -sha, respectively: (8) alia -bi -ga entonces -ga artu terreno propio tini -ndu -ga there LO T O then T O much land own have SUB T O riku-ya -na, no? rich become NOM no ‘T h e re one could become rich then, having o n e ’s own land, no?’ (9) Isi- m unda- ga a sta kolera muri- sha bini- xu- ka- ni this from T O even anger die SUB come PR PA 1 ‘From this I was coming even dying from an g er.’ In (10) and (11) different s u b je c t s u b o r d in a tio n s with - ndu and -kpi are illustrated: P IETE R MUYSKEN (10) 413 ahi-da- ga abin, piru tarde-yandu-ga there T O there is but late become SUB T O gana- u- nga -y win P R 3FU E M P ‘It is there, b u t when it becomes late he will be w inning.’ (11) m a m a -m un abiza-k i -kpi -ga dizi- slika m o th e r to tell AG go SUB T O say SD ‘She said it when I went to tell my m o th e r .’ In M uysken (1981b) I showed t h a t cases such as (10) are quite rare, b u t sentences such as (8) frequent in the samples. It is not evident t h a t -ndu is a co u n te re x a m p le to the claim t h a t no af fixes ar relexified in M edia L en gu a because there is increasing evidence for an earlier A m e r in d ia n -S p a n is h c o n tact language, spoken th r o u g h o u t Ecuador, the m o st strik in g feature of which was the use of -ndo or -ndu as a general ized verb m a rk e r (M uysken 1984). It m ay well be t h a t M edia L engua was influenced by this c o n ta c t vernacular and took over the -ndu ending from it in certain adverbial contexts, r a t h e r t h a n relexifying directly from Q u e ch u a in these cases. T h e r e are several o th e r Spanish affixes t h a t occur in M edia Lengua as well: d im in u tiv e - i i u / - i l a and p ast participle -do. T h e first occurs in Q u e c h u a as well, a n d can be seen as a borrowing. T h e second occurs only in S panish adjectives t h a t were p ro b a b ly taken over as a whole. Therefore we can m a i n ta in the generalization t h a t no affixes were relexified. 1.2 Possible s e m a n tic reasons f o r not relexifying Even if no affixes are relexified, one m ig h t argue t h a t this is not due to s tr u c t u r a l p ro p erties of roots and affixes, b u t to th e fact t h a t it would have been too com plicated to find th e sem an tic equivalents of Q u e c h u a affixes in Spanish. T h u s , one m ig h t argue, the fact t h a t the Q u e c h u a causative suffix -chi- occurs in M edia L engua as well is due to the absence of a similar causative suffix in Spanish. T h is a r g u m e n t is very plausible at first sight, b u t it has several deficits. First, n o t all Q u e c h u a affixes are w ith o u t a Spanish equivalent. T h u s, Q u e c h u a second person singular - nki could have been replaced by S p a n ish -5, Q u e c h u a first person plural -nchi by Spanish -m os. In a dialect of Q u e c h u a spoken elsewhere the Q u e c h u a agentive suffix -k is often replaced by S panish-derived -d u r (from -dor), b u t not so in M edia Lengua. We could extend this list for o th e r suffixes as well. T h e point is t h a t sem a n tic consid e ra tio n s m ay have led to the relexification of some affixes, b u t not others. T h e y c a n n o t acco u n t for the fact t h a t no affixes were relexified. C J L / R C L 33(4), 1988 414 Second, relexification of lexical items has also involved very considerable se m a n tic a d j u s t m e n t s som etim es. Take the Q u e c h u a im personal verb y a r u a - n ‘it. hungers (s o m e o n e ) ’. In Spanish you say yo tengo hambre ‘1 have h u n g e r ’. Now it would a p p e a r to be impossible to relexify here. Still, there is a M edia L en gu a form: the im personal verb am bri-naya-n lI feel like h u n g e r ’, co n tain in g the Q u e c h u a suffix -naya- ‘having a physical incli n a t i o n ’. Sim ilar cases a b o u n d . T h e y d e m o n s t r a t e t h a t se m a n tic factors did n ot prevent relexification of lexical items, even in com plicated cases. We find some Q u e c h u a items in M edia Lengua, b u t they generally refer to spe cific c u ltu re items, an d are p e rh a p s best th o u g h t of as loans. An ex a m p le is Q uechua shutich iy ‘b a p t i s m ’. Here it is n o t sem a n tic com plexity t h a t has prevented relexification, since there is the perfectly equivalent Spanish b a u tis m o , the te rm the priest would use anyway. Sim ilar a r g u m e n ts will be discussed in th e next section, where we dis cuss the relexification of g r a m m a ti c a l items or function words. T h e r e is no general s e m a n tic correspondence between Q u e c h u a and Spanish g r a m m a t i cal items. Still they are all relexified. To be fair, there is one g r a m m a ti c a l ite m t h a t is n o t relexified, and this m ay well have been p a rtly for sem a n tic reasons: the Q u e c h u a verb ka- ‘b e ’. In Spanish there are two verbs, ser and c s t a r , w ith a se m a n tica lly complex d istrib u tio n . T h is m ay have been one of the reasons t h a t M edia L e n g u a has consistently m a in ta in e d ka-. T h e possible o th e r reason is m orphological: esiar is s o m e w h a t irregular, and ser suppletive, like English kb e ’. On the whole, we can safely conclude, se m a n tic reasons do n ot explain why Q u e c h u a affixes were not relexified. 1.3 Clitics in Media Lengua So far I have established t h a t no affixes have been relexifiied an d t h a t this c a n n o t be for sem an tic reasons. Does this m ean t h a t the relexification process o p e r a n t in M edia L en g u a c o n s titu te s evidence for an Ite m and P r o cess (IP) view? I th in k not, and the reason is t h a t clitics are not relexified either. Q u e c h u a has a class of phonologically d e p e n d e n t elements, which do n ot have affix pro p erties (M uysken 1981a; Lefebvre and M uysken 1988) such as being sensitive to the category of the word to which they are a tta c h e d (Aronoff 1976), being m orphological heads (W illia m s 1981a), and being a t ta c h a b le only to lexical categories (Aronoff 1976). T h is class includes th e elem ents in (12): (7) a. - mi affirmative b. -shi -cha -chu hearsay d u b ita tiv e negation, question P IE T E R MUYSKEN c. -pish -tik -ri -m a -ga too, indefinite e m p h a tic e m p h a tic e m p h a tic topic Even th o u g h these elem ents c a n n o t a p p e a r as in dep en den t words, they have all th e sy n ta c tic p rop erties of being separately introduced by the phrase s tr u c t u r e rules. In M e d ia L en g ua they are not relexified: (13) isti sabi- nga- mci-bish this know 3FU E M P too ‘he will certainly know ’ (14) kital- mi uyari- xu- n how A F hear RE 3 ‘how it s o u n d s’ (15) nustru- ga alia -wa -bi -mi sinta- nchi we T O there DIM LO AF live lpl ‘we live th e r e ’ None of th e m are. It is h ard to establish t h a t this is not for sem antic reasons, b u t the considerations presented in section 1.2 hold here as well. In a d d itio n , there is evidence from bilingual rural Spanish t h a t there is an equivalent for -pish ‘too, indefinite’. In rural S panish we so m etim es find enclitic -tan (from S panish ta m b ien ‘t o o ’) used in the sam e way as Q u e ch u a and M ed ia L e n g u a -pish: (16) onde -ta n ha ido where too has gone ‘where then has h e /sh e gone?’ (17) yo -ta n quiero I too want ‘I w ant it t o o ’ T h e r e is no reason why this form was not a d o p te d in M edia Lengua. T h e reason t h a t clitics were not relexified c a n n o t be the fact t h a t they are p a r t of a word fo rm a tio n rule, since they are not. R a th e r , this suggests t h a t the lexicon is divided into several parts: a stem lexicon and a clitic lexicon, only the former of which p a rtic ip a te s in relexification. In t h a t perspective, however, there is no reason not to assum e the existence of an affix lexicon as well (Halle 1973; Lieber 1981; Selkirk 1982). T h u s the evidence from M ed ia L e n g u a for the IP ap p ro a c h tu rn s o u t to be invalid. 416 C J L / R C L 33(4), 1988 T h e re is evidence, of course, for a principled separation between roots on the one h a n d and affixes and clitics on the other. We m ay wonder w h a t the im plications of this grouping are. T h e three classes of elements m ay be assumed, within the IA perspective, to have the following properties: (18) phonological dependence roots — affixes + clitics -f subcategorized for a base — -f — finite list — ? ? W hile it is tru e t h a t there is a finite, though extensive, list of affixes in Q u e ch u a (and the sam e holds for clitics), the list is in no way stru ctu red . E xcept for certain su b d o m a in s , such as person and case, it rath er ap p ears a loose collection of elements. It is hard to see this difference between roots and affixes a n d clitics as principled. T h is leaves us for the m o m e n t with phonological dependence as the one distinguishing feature t h a t determ ines w hether relexificat.ion has taken place or not. In the next section we will see t h a t lexical m eaning is not w h a t sets roots a p a r t from the rest. A n u m b er of relexified stem s does not have a lexical m eaning. 2. S e m a n tic and syntactic properties o f lexical entries In the lexicalist theory developed in the early seventies, increasing a t tention was paid to the internal s tr u c tu r e of lexical entries, and to the types of in fo rm a tio n t h a t need to be present in such entries. In JackendofT (1975) entries of th e following type were postulated: (19) /biliv/ weak conjugation +V [ - f ___ NP] phonological representation conjugation class syntactic features subcategorization features N P j BELIEVE N P -2 semantic interpretation [human] s e l ect i onal r e s t r i c t i o n s M uch later work, including W illiam s (1981b), has tried to derive the p ro p e r ties of lexical items in (19) as much as possible from in d ep en d en t principles. In a d d itio n , work of Stowell (1981) and others has been directed at deriv ing p rop erties of phrase s tr u c tu r e rules directly from the lexicon. In this section I will discuss M edia Lengua from the perspective of the s tr u c tu r e of the lexical entry. In 2.1 I consider lexical versus g ra m m a tic a l m eaning, in 2.2 the process of lexicalization and sem an tic specialization of derived forms, a n d in 2.3 the directionality of governm ent. Before going on, I should stress t h a t the very process of relexification confirms th e concept! i of he lexical entry as a loosely organized bundle P I E T E R MUYSKEN 417 of a t least three feature clusters: th e phonological rep resen tatio n , the m o r phological characteristics, and the sy n ta c tic and se m a n tic properties. T h is should n o t be in te rp re te d , however, as directly confirming th e notion of dual re p re se n ta tio n s for words t h a t we find in work of S p ro a t (1987) an d others. M ed ia L en g u a Spanish-derived roots are the base for Q u e c h u a suffixes, and th e suffixation process is governed by the m o r p h o s y n t a x of Quechua. ‘2.1 Lexical and g ra m m a tica l m eaning M edia L en g u a relexification takes place on the basis of m e a n in g corre spondences. A S p an ish ste m as close as possible in m e a n in g to the Q u e c h u a original is used to replace it. T h e question now is how closeness in m e a n ing is d e te rm in e d . It tu r n s o u t we have to distinguish here between lexical m e a n in g and g r a m m a t i c a l m eaning. T h e former can be d e term in e d by ref erence to some extralinguist.ic entity, th e former only by reference to the lan g u ag e sy stem s themselves. It tu rn s o u t t h a t relexification is feasible, with all th e difficulties m en tio n ed , for lexical items, b u t o p e ra te s in a very in co m p le te m a n n e r for g r a m m a t i c a l items. In Muysken (1988) I illustrated this for d e m o n s tra tiv e s . Here I will try to argue th e sam e po in t for question words. In (20) th e relevant forms are given for Q uechua, Spanish and M edia Lengua: Q uechua Spanish Media Lengua pi im a m ayxan m a sh n a quién quén cuál cuánto a como cuándo qué horas por qué kin ki/inki kwal kw antu a komo kwandu ki uras purki (purki-m unda) ‘kom o’ undi-bcase im a uras im a -m u n d a im a-shna m ay+ case cómo (d)ónde ‘w h o ’ ‘w h a t ’ ‘w hich’ ‘how m a n y ’ ‘for how m u c h ’ ‘w h e n ’ ‘at w h at t i m e ’ ‘w hy’ ‘how ’ ‘w here’ T h e se e x a m p les show t h a t M ed ia L en gu a forms, with the exception of the form for ‘w h e r e ’ are m odeled on S panish r a t h e r t h a n Q u e c h u a p a tte r n s . If it h a d been th e reverse, we would have found forms such as * (in )k i-m u n d a and * (in )k i-sh n a , which do n o t occur a t all. T h e form p u r k i-m u n d a , which follows b o th Q u e c h u a an d S panish, occurs only once. 2.2 The process o f lexicalization O n e of th e differences between Q u e c h u a a n d S panish is t h a t th e la tte r has m o re s e p a r a te verbal roots, while th e former has fewer roots and tends C J L / R C L 33(4), 1988 418 to form new verbs th ro u g h a complex system of verbal derivational suffixes. T h u s we have: (21) Q uechua Spanish apaapa-m u- llevar tra e r ‘ta k e ’ ‘b rin g ’ Here the cislocative suffix -m u - is added to the original root to create a cislocative verb whose m e a n in g in S panish (and English) is expressed with a s e p a r a te verb. S o m e tim e s the derived m e a n in g is simply com positionally d e te rm in e d , and often th e ro o t + affix c o m b in a tio n has un d erg o n e se m a n tic specializa tion or drift. We m ay hypothesize t h a t the lexicalized forms are replaced in M edia L en gu a by Spanish sim plex roots, an d the com p ositio nal forms by forms directly p a t t e r n e d on the complex Q u e c h u a original, or include b o th a S panish ro o t w ith the complex m e a n in g and the Q u e c h u a derivational suffix. In M uysken (1981) d a t a are presented which illu strate this p a t t e r n for the Q u e c h u a verb riku- ‘see’ and derived forms. Here a few ex am ples are given directly related to the forms in (21). Sentences (22) and (23) illu strate the use of the S panish-derived forms irayi- and lleba-, respectively: (22) intonsi lindu radiyu- da trayi- shka then nice radio AC bring SD ‘T h e n it turned o u t t h e y ’d brought a nice ra d io .’ (23) avva -da ahi -m u n d a lleba- nga zin A m b a tu - mun w ater AC there from take 3FU they say A m b a to to ‘From there they will take the water to A m bato, they say.’ In (24) an e x a m p le of a r e d u n d a n t form is given: (24) g ra b a d o ra -d a trayi- mu -ngi ta p e recorder AC bring CIS 2 ‘Bring the ta p e record er.’ T h e in teractio n between relexification and derivational m o rp h o lo g y needs to be stu d ied in m uch m ore detail, for different groups of verbs. T h e m ain p oin t here was t h a t relexification can be used to s tu d y the degree of lexicalization of a p a r tic u la r ro o t -f affix c o m b in a tio n . 2.3 The lexicon and s y n ta x : the directionality issue As was briefly m e n tio n e d above, research in the early eighties has a t t e m p t e d to drive pro p erties of ph rase s tr u c t u r e such as order specifications from p ro p erties of the lexicon (Stowell, 1981, and later work by various a u thors). T h is was done in te rm s of d irectionality of governm ent: verbs a n d P I E T E R MUYSKEN 419 p rep osition s govern rig h tw a rd in English, therefore English has [P NP] and [V NP] s tru c tu re s. We will now see w h a t evidence M ed ia L e n g u a adduces on this issue. It should n ot surprise us t h a t M e d ia L en gu a shows the sam e percentage of XV (80%) versus VX (20%) sentences as the Q u e c h u a spoken in th e sam e region. M ed ia L e n g u a verbs are Q u e c h u a vebs, for all practical purposes, and they are m a rk e d with Q u e ch u a affixes. Q u e c h u a verbs govern leftward. T h u s we ex pect M ed ia L en g u a verbs to govern leftward as well. W i t h elem ents of the category P the s itu a tio n is more com plicated. Q u e c h u a has a few p o stp o sitio n s, such as k ’ipa ‘a f t e r 1, and shina ‘like’. W h e n these are relexified, we find p o stp o sitio n s in M edia Lengua, as with the u n derlined c o n s titu e n ts in (25) a n d (26), respectively: (25) miza despwesitu kaza-mu i -naku -ndu-ga, ahi -bi b o d a Mass after home to go R E C - P L SUB T O there LO feast d a -naku -n, ahi -bi bayla -naku -n give R E C - P L 3 there LO dance R E C -P L 3 ‘Going home after Mass, there they give a feast and d a n c e .’ (26) asi -11a- di kumu bos, bos kwenta- 11a- di gringu thus DEL E M P like you you like DEL E M P gringo kunusidu -g u n a tini-n bastanti miu Rosalina acqu aintan ce PL have 3 plenty my Rosalina ‘My R osalina has plenty gringo acquaintances ju s t like you, like you pre cisely.’ As k u m u ‘like’ (derived from S panish como) (26) shows, however, we also find a few p rep o sitio n s in M ed ia Lengua. T h u s we have entre ‘a m o n g ’ and asta ‘u n t i l ’, in (27) a n d (28), respectively: (27) nustru- ga entre seys -mi ga-nchi we T O am ong six A F be lpl ‘T h e re are six of u s .’ (28) solo isti syera asi, kazi isti M achachi-m unda only this Highlands thus alm ost this Machachi from asta Chimborazo R iobam ba-gam a, b a ra to paga-naku -n until C h im b o razo R io b am ab a-u n til cheap pay R E C -P L 3 ‘Only here in the Highlands, alm ost from Machachi here to Chim borazo, to R io b a m b a do they pay low wages like th is .’ T h e Q u e c h u a equivalents of this words are affixes r a th e r t h a n s e p a r a te ele m ents. 420 C J L / R C L 33(4), 1988 I in te rp re t th e co n trast between (25)—(26), on th e one h a n d , and ( 2 7 ) (28), on th e o th er, to s u p p o r t the lexically-based view of directionality. T h e r e t u r n s o u t to be no general [NP P] or [P NP] order in M ed ia Lengua. T h e P elem ents t h a t are relexified from Q u e c h u a govern leftward, th e el em e n ts newly in tro d u c e d from S panish govern rightw ard. Sim ilar results, t h a t I will n o t discuss here, can be o b ta in e d from the in tro d u c tio n of Spanish com plem entizers. 3. Conclusion T h is concludes my discussion of some of the evidence t h a t can be g a t h ered from M e d ia L engua for our conception the o rg an izatio n of the g r a m m ar. In w h a t sense can it be viewed as extern al evidence? It is not external in th e sense t h a t it refers to the interaction with a non-linguistic (cogni tive) system , which I th in k would be the p ro p er definition of w h a t e x te rn a l evidence for the o rg a n iz a tio n of th e g r a m m a r is. It is only external t h a t for the in te r p r e ta tio n of the M edia L engua d a ta , two o th e r linguistic sys tems, Q u e c h u a and S panish, are taken into account. B u t inasm uch as all recent g r a m m a t i c a l research is inherently c o m p a r a tiv e in n a tu r e , this does n o t m ak e this ty p e of research special. I will conclude with raising the (open) question of w h e th e r we can th in k of M edia L e n g u a as th e result of a kind of word internal code mixing. S chem atically this possibility is presented in (29), where th e su p erscrip ts s an d q refer to the S panish, respectively Q u e c h u a language index of roots a n d affixes, a n d the su b sc rip t V to th e categorial identity of an element: (29) ROOTv s A F F IX v q T h e affix is the head (W illia m s 1981a) and determ in es th e lang uag e index of th e whole word. T h e a d v a n ta g e of this view is t h a t it relates th e possibility for relexification in Q u e c h u a to the highly a g g lu tin a tiv e c h a ra c te r of the language: since there will always be affixes to m a r k a word as belonging to th e language, the ro o t can be derived from a different language. T h e d isa d v a n ta g e of this view is t h a t it c a n n o t account for cases such as despwesitu ‘a f t e r ’ in (25). T h i s form behaves as a Q u e c h u a p o stp o sitio n , even th o u g h it is entirely S panish. We would have to p o s tu l a t e a 0 Q u e c h u a case affix on it, m a r k in g it as Q uechua. I will leave this issue open for fu rth e r research. P IE T E R MUYSKEN 421 Appendix: List of a b b re v ia tio n s used in glosses 1, etc. AC AF AG CIS DEL DIM EMP FU LO NOM PA first person etc. marker accusative affirmative agentive PL, pl PR RE R E C -P L SD SUB TO cislocative (m ovem ent towards speaker) marker delim itative dim inutive e m p h a tic m arker future locative nominalizer past plural progresive aspect reflexive conjoint action: reciprocal or plural sudden discovery tense adverbial su b o rd in a to r topic m arker REFERENCES Aronoff, M ark 1976 Word Form ation in Generative G ram m ar. C am bridge, Mass.: M IT Press. Chomsky, Noam 1965 Aspects o f a Theory o f S y n ta x . C am bridge, Mass.: M IT Press. Chomsky, Noam, and Morris Halle 1968 The S o u n d P a tte rn o f English. New York: H arper and Row. Lefebvre, Claire, and P ieter Muysken 1988 Mixed Categories. N o m in a liza tio n s in Quechua. Dordrecht: Reidel. Lieber, Rochelle 1981 On the Organization o f the Lexicon. Bloomington: In d ian a Univer sity Linguistics Club. McCarthy, John 1981 A Prosodic T h eo ry of N onconcatenative Morphology. Linguistic I n quiry 12:373-418. M arantz, Alec 1982 Re R eduplication. Linguistic In q u iry 13:435-482. C J L / R C L 33(4), 1988 422 Muysken, P ieter 1981a Q uechua Word S tru ctu re . Pp. 279-328 in B indin g and Filters. F. Heny, ed. C am bridge, Mass.: M IT Press. 1981b Halfway Between Q uechua and Spanish: T h e Case for Relexification. Pp. 52-78 in H istoricity and Variation in Creole Studies. A. Highfield and A. Valdman, eds. Ann Arbor: Karom a. 1984 Sources for th e Study of A m erindian C o n ta c t Vernaculars in Ecuador. Pp. 66-82 in A m s te r d a m Creole Studies IV. Pieter Muysken, ed. A m sterd am : University of A m sterd am . 1988 Lexical R estru c tu rin g in Creole Genesis. Pp. 193-210 in A k te n des IV . E sse n e r Kolloquiums. N. Boretzky et al, ed. Bochum: N. Brockmeyer. Selkirk, Elizabeth 1982 The S y n ta x o f Words. C am bridge, Mass.: M IT Press. Sproat, Richard 1985 Dual R epresentations in Morphology. Ph.D . dissertation, M IT. Stowell, T im o th y 1981 T h e Origins of P hrase S tru cture. Ph.D. dissertation, M IT. Williams, Edwin 1981a On the Notions “Lexically R elated ” and “Head of a W ord” . L i n guistic In q u iry 12:245-274. 1981b A rg u m e n t S tru c tu re and Morphology. The Linguistic Review 1:81— 114.
© Copyright 2026 Paperzz