You are on page 1of 42

Nasality and Nasal Prosody in Chimila Review by: Terry Malone International Journal of American Linguistics, Vol.

76, No. 1 (January 2010), pp. 1-41 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/10.1086/652753 . Accessed: 24/09/2012 23:23
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of American Linguistics.

http://www.jstor.org

NASALITY AND NASAL PROSODY IN CHIMILA1 Terry Malone


Latin American Mission/FUSBC Medelln In Chimila, nasal consonants contrast with voiced stops at the same point of articulation and no surface phonetic contrast occurs between nasalized and oral vowels in core lexemes. Nevertheless, several lines of evidence indicate that the Chimila lexicon includes nasal morphemes in contrast with oral morphemes. Evidence for this contrast includes morphophonemic alternations of stem-forming verb sufxes, appearance of a glottal glide following intransitive verb roots that have been transitivized, allomorphs of the intransitive imperative, and a restricted alternation between word-initial voiced stops and nasal consonants in roots and some sufxes. A constraint requiring lexical nasality to attach to consonants in core lexemes and the interaction of this constraint with the prosodic system, lexical tone, syllable structure, and the morphology account for the varied manifestations of lexical nasality. [Keywords: Chimila, subsegments, nasality, nasal prosody, nasal vowels]

1. Introduction. Chimila is a Chibchan language spoken in northern Colombia within a community of 1,600 persons. Most published works on Chimila phonology and grammar (Trillos 1997 and Malone 2000; 2001; 2004) do not mention the existence of nasal prosody or phonetically nasalized vowels in interjections and onomatopoeia; another paper (Malone 2006:24, 26) barely mentions the existence of nasal prosody and nasal lexical roots in two footnotes. Nevertheless, evidence exists indicating that nasality is present, and that nasal prosody operates within lexical stems. In this paper, I review the evidence and argue that nasality is present in the Chimila lexicon, although no phonemic surface contrast exists between oral and nasalized vowels, except that of some interjections and onomatopoeia. I rst present necessary information concerning Chimila phonology (2) and then describe manifestations of phonetic nasality in core lexemes, interjections, and onomatopoeia (3). 4 presents evidence for lexical nasality, including morphophonemic alternations (4.1), insertion of syllable-nal /h/ (4.2),
1 I would like to express special appreciation to Margrit Hotz, who has provided much logistical support and thus has made it possible to conduct this analysis and write this paper. The Chimila data in this paper come entirely from my eld notes. I am grateful to many Chimila friends who have generously shared their language; unfortunately, names cannot be mentioned here due to political conditions in their homeland. I am grateful to the editors and three referees for comments on an earlier draft which have considerably improved this paper. Of course, any defects are mine.

[IJAL, vol. 76, no. 1, January 2010, pp. 141] 2010 by The University of Chicago. All rights reserved. 00207071/2010/76010001$10.0

international journal of american linguistics

and insertion of consonants /t/ and /k/ in the intransitive imperative (4.3). 5 presents a unied analysis accounting for these phenomena. In 5.1 I argue that lexical nasality must attach to consonants in core lexemes and that the interaction of this constraint with lexical tone, prosody, syllable structure, and the morphology accounts for the phenomena of 4, whereas in onomatopoeia and interjections lexical nasality must attach to vowels (5.2). In 5.3 I account for the variation in realization between voiced stops and nasals, by proposing optional nasalization of the root-initial consonant; this allows one to determine underlying forms of sufxes that show this variation. Section 6 presents conclusions based on the proposed analysis. 2. An overview of Chimila phonology. This overview is based on the analysis presented in Malone (2001; 2006); more detailed information is presented concerning segments and clusters crucial for understanding how lexical nasality functions. The inventory of Chimila consonant phonemes is as follows: (1) Voiceless Stops Voiced stops Nasals Fricatives Lateral Flap Glides Bilabial p b m Coronal t d n s l \ y Velar k g N h Glottal ?

Voiceless stops are nonaspirated. All voiced stops are prenasalized. All coronal segments are dental, except for the ap and the glide. The labial glide is preceded by a light nonprenasalized velar voiced stop; the alveopalatal glide is preceded by a light nonprenasalized alveopalatal stop. Consonants occur with three phonetic lengths: short, medium long, and long. Consonant length is not phonemic: some cases are analyzed as heteromorphemic geminates, some occur in order to fulll a minimal bimoraic trochaic foot requirement, and most occur as the result of the interaction of lexical tone with the metrical system (syllables bearing lexical tone must be bimoraic, i.e., heavy). (See Malone 2006 for more details on lexical tone, parsing constraints, and their relationship to consonant length and syllable structure.) The glottal fricative /h/ occurs in word-initial position in Chimila; it does not occur word-medially except in interjections (see 6 in 3.2), compound words ([kahhk\ab\] tree trunk (kah tree, hak\a? head, -b\i chunk), and loanwords ([huh] needle, from Spanish aguja). In all such words it occurs in the syllable onset. Preceding the vowels /o, a/ the fricative exhibits

nasal prosody in chimila

the allophone [h]; preceding the vowels /e, i, u/ the stricture of the glottal fricative narrows, so that it tends to vary freely with a velar fricative [x]. A glottal off-glide occurs word-nally in verb roots of the syllable structure /(C)(C)V/. Its stricture in this position is highly variable from word to word but constant for each word; in syllable-nal position the stricture cannot be predicted from the height of the preceding vowel. For instance, the stricture of the off-glide in [wih ~ wi] know, be able to is considerably less than that of [sih] black or [ih] be hard. It is often inaudible and, unlike the off-glide of [sih] and [ih], disappears in compound words or when the root is sufxed. Further, light stricture, as in [oh ~ o] do, contrasts with heavy stricture, as in [oh] roast. The distribution of this off-glide is restricted in syllable-nal position, much more so than any other consonant occurring in the syllable coda (see 4.2.2). There is some question about whether it should be analyzed as a syllable-nal occurrence of the phoneme /h/; in fact, Trillos (1997) analyzes it as an articulatory modication of the preceding vowel (1997:66), i.e., as breathy vowels (/Vh/ in contrast with /V/). Because the glottal off-glide in syllable-nal position is restricted in distribution, does not behave the same as syllable-initial /h/ with respect to stricture, and is associated with lexical nasality (4.2), the term glottal off-glide ([h]) is used here, in order to distinguish it from the syllable-initial glottal fricative. Its distribution and phonemic status is further considered in 4.2 and 5.1, where I conclude that it is best analyzed as a syllable-nal occurrence of the phoneme /h/. The following monosyllabic consonant clusters occur in syllable-onset position: /p\/, /b\/, /k\/, /g\/, /kw/, /gw/, /kw\/, /gw\/, /w\/, /ty/, /dy/, and /ny/. In syllable-initial position, the last three clusters are manifested respectively as an alveopalatal affricate preceded by a light voiceless dental stop [tc], a prenasalized alveopalatal glide in which a light alveopalatal stop intervenes between the dental prenasalization and the glide /ndy], and an alveopalatal nasal [].2 Articulatory, acoustic, and morphophonemic evidence support this analysis of these clusters. The palatalization associated with the phonetic realization of the clusters /ty/, /dy/, /ny/ is longer than one would expect for palatalization as an articulatory modicationin fact, three times as long (Malone 2000:7677). When a CV syllable bearing lexical tone occurs preceding the clusters /ty/ and /dy/ in word-medial position, the clusters split into a dental syllable coda and an alveopalatal syllable onset: (2a) [ndt.cata] (2b) [kn.a]
2

go in! eld

I do not reproduce predictable phonetic details in the data of this paper unless they are germane to the discussion at hand.

international journal of american linguistics

In (2b) the dental nasal is especially audible and shows clearly in acoustic diagrams (Malone 2000:7375). /ty/ occurs in contexts where one must conclude that two segments have combined. For example, the intransitive imperative morpheme [-ya] has the allomorph [-ca] in contexts where a glottal stop is part of the preceding monosyllabic verb root (3c) (note that in 3b the glottal stop occurs at the end of the stem). (3a) [saa\a.dya] (3b) [kita?.dya] (3c) [yut.ca] sleep! sing! go! (/saa\a/ sleep) (/k-ta?/ sing) (/yu?/ go)

Shortened forms in fast speech also suggest that [c] represents two combined segments, for instance, [hc] house instead of [htak] and [Ng\u ckw] instead of [Ng\takkw] elders, older person, where two syllables fuse to form the syllable [ca]. Lexical syllable structure is summarized by the formulas (X)V (C2), (X)V(C2), and (X)VV(C3), where X can be /p/, /t/, /k/, /b/, /d/, /g/, /m/, /n/, /N/, /h/, /s/, /l/, /\/, /w/, /y/, or the clusters /p\/, /k\/, /b\/, /g\/, /w\/, /ty/, /dy/, /kw/, /gw/, /kw\/, /gw\/; in word-medial position, /?/ can occur as X. C2 is lled by /m/, /N/, /?/, and in word-medial position /n/ also occurs in this set. Only /?/ occurs as C3. Chimila vowel phonemes are given below (from Malone 2006): (4) i e a Vowels occur with three contrastive phonetic lengths: short, medium long, and long. Medium long vowels are analyzed as canonically long vowels (monosyllabic, bearing one single phonetic tone), and long vowels as geminate vowels (disyllabic, bearing rising or falling phonetic tone) (see Malone 2001 for details). High tone contrasts with its absence on monosyllabic long vowels, as in /k?/ to buy vs. /ki?/ to receive. On disyllabic vowels a four-way contrast occurs, as in /too?/ maraca, /to?/ heart, /t?/ gourd, /tii\i/ thin, /ti\o?/ hawk. Canonically long and geminate vowels contrast, as in /to?/ to observe, look vs. /too?/ maraca or /k?/ to buy vs. /ki?/ sap. On CV syllables high tone contrasts with its absence, as in /\ku?/ species of owl vs. /\uku \uku/ rub, or /ska/ ([saka]) someone ground vs. /sAka/ ([saka]) someone tamped down earth; in these cases a requirement that syllables bearing lexical tone be heavy results in the lengthening of the consonant immediately following the vowel bearing high tone.3
3 According to the analysis presented in this paper, the vowels of the verb roots grind and tamp down earth are lexically nasal. Phonetic nasality is not present in their phonetic forms, so I use them here to illustrate the tone contrast. Finding contrasts that do not involve at least

u o

nasal prosody in chimila

Word-level phonetic stress is predictable: it always occurs on the leftmost syllable of the word and therefore is not marked in this paper. One default phonetic high tone occurs per word. The syllable locus for the default high tone depends on the number of syllables in the word, and whether or not the syllable bears lexical tone. The leftmost syllable is the unmarked position for word-level high tone; in this paper I do not mark default word-level high tone. Words parse from left to right into bimoraic trochaic feet, in accordance with proposals in Hayes (1995); see Malone (2006) for further details regarding foot structure and parsing. 3. Phonetic nasality: an overview. 3.1. Conditioned nasal vowels. In core lexical vocabulary, phonetic vowel nasalization is conditioned by the presence of contiguous nasal consonants. Vowels are nasalized immediately preceding nasal consonants and immediately following word-medial nasal consonants. Nasalization varies considerably preceding nasal consonants; it is most audible on back vowels and less audible or inaudible on front vowels. As might be expected, it is heavier in syllables with nasals in both the onset and the coda, and is usually heavy following nasal consonants in word-nal syllables of the form CV(?). Nasal consonants following a glottal off-glide are often partially and sometimes completely devoiced, with concomitant heavy nasalization of the immediately following vowel; the preceding vowel is nasal, but not as heavy. If the nasal consonant devoices completely, as in (5), the preceding vowel is nasalized:4 (5) [?hN6takE?Ente] /ahNa-ta-ke?e-de/ bring-toward-fut.d.o.-1ev Ill bring it back.
one lexically nasal vowel (such as /u$ ka/ [uka] to drink and /kwa/ [ukwa] to cut) is difcult. A possibility is /taka?/ with force and /tka?/ turn sour. Speakers perceive the velar stop of turn sour as longer than that of with force; acoustic measurements (two tokens of each) have been inconclusive. The acute accent indicates lexical tone; I mark phonetic high ( ! ), mid (-), and low ( ~) tone on phonetic data where it is pertinent to the discussion. Phonetic mid and low tone are inserted by default after the high tone representing word-level tone has been assigned and the phonological word has been footed. 4 Because phonetic devoicing in this environment and nasalization of vowels contiguous to nasal consonants are predictable, I do not indicate them elsewhere in this paper, unless germane to the discussion. The following abbreviations are used in the glosses of this paper: 1 = rst person; 2 = second person; 3 = third person; sg = singular; pl = plural; d.o. = direct object; adj adjective; ben benefactive; caus causative; cond conditional; conj conjunction; cont continuative; ev evidential; fut future; gen genitive; imp imperative; int interrogative; inv inverse; itr intransitive; mid middle voice; neg negative; nomz nominalizer; obj direct or indirect object; pl plural; ser serial verb linker; stem transitive stem-forming sufx; pas past; pat semantic patient; poss possessive; sg singular; stat stative; top topic.

international journal of american linguistics

Vowel nasalization preceding voiceless nasal consonants in this position is not attributable to the off-glide: no nasalization can be heard in vowels preceding the off-glide in words such as [kah] tree, stick, [kah mb\nta] piece of wood. [ah] hair/to go, or [sihtu?] its black. 3.2. Oralnasal vowel contrast in interjections and onomatopoeia. Nonallophonic nasalized vowels are found in some interjections and onomatopoeia. (6) lists interjections containing nasalized vowels: (6) [?], [?h] [h ~ hoh] [h?h? ~ h?] [hh] [?E$hE$?E$] [h] [ ? e hn I ] [ayo] [mISu] yes! here it is (take it)! right here! (pointing and touching place) no! ok! huh? ok, agreed! expression of pity kitty, kitty!

Note that the nasalization is not likely due to rhinoglottophilia (spontaneous nasalization occurring after the phoneme /h/) as described for Iapari (Arawakan) in Parker (1999). For one thing, two interjections occur containing two oral vowels and one nasal vowel: [?oho?oh] hi!; [?Ehe$e$I] eek! (reaction at being startled). If nasalization were due to rhinoglottophilia, one would expect all the vowels to be nasalized in these interjections. Second, other interjections occur in which oral vowels occur contiguous to /h, ?/ and which contrast with nasal vowels in the interjections above, for instance, [?ee] well, well!,5 [?ookya] well, ok then!, [ha??] ok!, [hoo\i] of course!, [ha?\i yu\i] you bet!. In addition, numerous words beginning with /h/ occur in the core lexicon whose vowels are not phonetically nasalized and which contrast with the nasalized vowels of the interjections in (6) above: (7) /oo?/ /h?o/ /hk\u?/ /hoowa/ [?oo?] [ho?o] [hok\u?] [hoowa] ax scrape it! housey to bathe /e/ [?e] /h?/ [he?] /haa?/ [haa?] to be at to toast shelf

Nasalized vowels also occur in onomatopoeia: (8a) [isa 3.conj


Ngo$ki Ngki

heehaw

heehaw

ni ser

huu\a shout

mbuu\u

donkey

sawi\i] male

The jack brays heehaw heehaw .


5

Older speakers pronounce this as [?e$e$e$], with rising intonation.

nasal prosody in chimila

(8b) [isa 3.conj

noNwa hear \h \h grunt

we$e$? ribbet niti? itr

we$e$? ribbet tu\uku] hog

ni ser

noNwa]6 hear-mid

One hears them (frogs), one hears ribbet ribbet . (8c) [isa 3.conj

The hog grunts. In these and other examples, nasalized vowels can occur following a stop, a semivowel, the ap, the liquid, and the sibilant. They contrast with oral vowels in other onomatopoeia: (9a) [isa 3.conj
Ngoi Ngoi

clonk

clonk

niya] say tu\uku] pig suh] angry

It clonks lightly. (dog gnawing a bone) (9b) [isa 3.conj kwe? kwe? squeal \u\u? grrr \u\u? grrr niti itr ni its tutikwi little (pl)

Piglets squeal. (9c) [isa 3.conj niya say pek\u dog isa 3.conj

The dog growls when it is mad. As in the case of interjections, the nasalized vowels contrast with oral vowels in core lexemes, for example, /wee?/ be hot, /gota/ to get dressed, and /\oo\o/ sp. anteater. In lexemes that are not onomatopoeia or interjections, nasalized vowels only occur contiguous to nasal consonants.7 In spite of this and other phonological differences, onomatopoeia in (8) (and elsewhere in the language) behave as normal intransitive verbs: the serial verb marker /ni/ follows the onomatopoeia of (8a) and (8b), as if the animal sounds were the rst verb in a serial verb construction; and in (8c) the intransitive verb sufx /ni?ti?/ is used to indicate sounds or involuntary actions, for example, /gim gim ni?ti?/ to blink on and off or /ansi ansi ni?ti?/ to sneeze. In other words, the onomatopoeia are inected as if they
This speaker alternatively pronounces ribbet ribbet as [Nwe$e$? Nwe$e$?], i.e., she optionally nasalizes the glide. 7 The phonology of onomatopoeia differs in two other respects from that of core lexemes: (1) voiced and voiceless stops can appear in syllable-nal position, in contrast with nasal consonants and the glottal stop: [p\? p\?] noise of soldiers marching, [p\k p\k] noise of car tires clicking, [k\? k\?] or [k\? k\?] noise of joints cracking, [sk sk] noise of clock ticking; [gwp gwp] swish of tail wagging, [k\g ndy] to make a snapping or creaking noise (tree branches); and (2) word-nal glottal stop does not delete in phrase-internal position.
6

international journal of american linguistics

were regular verb stems in the language. Furthermore, onomatopoeia and other verbs exhibit derivational relationships, for example, [tontn] sound of dripping water [tnt nt:?] to drip and [tntoNka] put drops in or [tU tU tU] sound of knocking and [tuntntuhNt:i?] to knock on door (each set of words is from speakers of a distinct dialect).8 Other evidence suggests that onomatopoeia function as part of the Chimila lexicon, even if they are phonetically and semantically marked in the sense of Klamer (2002). In Chimila onomatopoeia are often a source of lexical referents (or vice versa), as we saw with sound of dripping water and to knock on door. Other examples abound; a sampling includes [k\m], a species of owl that says [k\u\N k\u\N], [gwp gwp] swish of tail wagging, and [gwp gwp] to wag tail; the woodpecker makes the sound [ta? ta? ta? ta?] and [ta?tahnitaka] means to thunder; a species of crane [k\eu?] calls [k\eu? k\eu? k\eu?], [loN loN] sound of banging on hollow object and [lhlond? ~ hohllond?] hollow, [k\k k\k ni y?] to squeak or make a cracking sound (pack saddle) and [k\? k\?] snapping or popping sound ( joints). To some degree interjections also appear to migrate back and forth between the core and the margins of the lexicon, for instance, [hohkwa] take it! ([h ~ hoh] take it! [-kwa] pat) or [ookya] ah, yes! ([oo] ah [-kiya] then, consequently). These observations suggest that, just as Klamer (2002) assumes for Dutch and Kambera, onomatopoeia and interjections are part of the Chimila lexicon, and even though their phonology (and semantics) is complex, they must be taken into consideration if one is to adequately describe Chimila phonology. The most obvious hypothesis to account for this data would be to propose that a contrast between nasal and nonnasal vowels existed at one time in core lexemes and was manifested by a phonetically audible contrast between oral and nasal vowels; synchronically these contrasts have only been preserved in interjections and in onomatopoeia. 3.3. Variation in realization between voiced stops and nasals. As indicated in the consonant chart in 2 above, nasal consonants and voiced stops contrast at the same point of articulation. Examples of these contrasts are: (10) /bne?/ /mne?e?/ /do/ /n/
8

[mbn?] [mne??] [nd] [n]

lots (mass) to want it is bad to hear

The last pair of words suggests that the phonological markedness of onomatopoeia cannot extend to regularly inected verbs, and that at some point the phonological restrictions typical of core lexemes begin to operate on lexical roots derived from onomatopoeia. In fact, speakers of another dialect represent the sound of knocking as [tuN tuN].

nasal prosody in chimila

/dyO?/ /nyoo/ /g?/ /No?/

[ndy?] [] [Ng?] [N?]

hot pepper cheek (buttock) to grind again to make

In spite of these contrasts, free variation occurs in some words between voiced stops and nasals at the same point of articulation in both word-initial and word-medial positions (11). For each word I have listed the preferred or more frequent variant rst. (11a) /b/ ~ /m/ [mbee ~ mee] [masu? ~ mbasu?] [mbenta tu? ~ menta tu?] [me ?a\i? ~ mbe?a\i?] [munti? ~ mbunti?] [miniki ~ mbiniki] [\aamb\a? ~ \aam\a?] (11b) /d/ ~ /n/ [ndah ~ nah] [ndap\i ~ nap\i] [nek\u? ~ ndek\u?] [tulunda? ~ tuluna?] (11c) /g/ ~ /N/ [Naa\i ~ Ngaa\i] [Nata? ~ Ngata?] [Ngatika? ~ Natika?] [Ngakwta ~ Nakwta] [No? ~ Ngo?] where? nearby to reject to give birth sp. rodent how? mountain lion (archaic)

nose coffee (archaic) dog (archaic) cut off (adj) long pulp to desert answer! to make

The alternation occurs between the consonant cluster /dy/ and the alveopalatal nasal //, which in some word-initial occurrences represents the underlying cluster /ny/; the semiconsonant /w/ and the word-medial cluster /Nw/ also vary freely in some words: (12) [ndyaNNg\a? ~ aNNg\a?] [ooni? ~ ndyooni?] [tii\igwa? ~ tii\iNwa?] join of torso with thighs (in front) twisted fruit of plane tree

Speakers who admit this free variation are rarely able to distinguish between the oral segment and its nasal counterpart; they are usually not even conscious that this variation exists in their speech. At the same time many words

10

international journal of american linguistics

occur in which word-initial voiced stops or nasals are not allowed to vary freely with their counterparts:9 (13) [mata?] [mbute?] [no] [ndoo] *~ [mbata?] *~ [mute?] *~ [ndo] *~ [noo] far to think to hear it is bad

No consistent surface phonetic conditioning factor occurs that satisfactorily accounts for the lack of alternation for words such as those in (13). It instead suggests that there is some relationship between voiced stops and nasal consonants. The free variation between voiced stops and nasal consonants can be explained quite neatly if one assumes that these segments are lexical voiced stops in the onset of roots with nasalized vowels, and the consonant is optionally nasalized, as in /bYniki/ how. If no free variation occurs for a word beginning with a voiced stop, then obviously the voiced stop is lexical but fully capable of heading a root containing nasalized vowels, as in /dIta/ to cut a piece off from a mass (compare with /ditake/ water). If no free variation occurs for a stem beginning with a nasal consonant, the nasal consonant can be assumed to be lexical, as is the case for the stems /ma\i?/ sweet and /mAta/ to pass by. Nevertheless, the data presented so far provide little support for postulating the existence of lexical nasalized vowels in Chimila, at least according to the normal canons of phonemic analysis, and they can be explained away by various means. For example, the alternation between voiced stops and nasal consonants at the same point of articulation could be a synchronic remnant from a time when lexical nasalized vowels did exist, as with onomatopoeia and interjections. 4. Evidence for lexical nasality. 4.1. Morphophonemic alternations. Morphophonemic variations occur, however, which are best explained by assuming that some verb roots or their vowels are marked for nasality in the lexicon. The vowels of these lexemes are not phonetically nasalized contiguous to nonnasal consonants; nevertheless, they require stem sufxes beginning with nasal consonants, and these are in alternation with sufxes beginning with stops in other environments. In all cases a lexical root, which may be unspecied for transitivity (/sa?-/ move, long and thin entity), intransitive (/u-/ sit) or transitive (/mo-/
9 The absolute rejection of the version with nasal consonant of the word for hot pepper is perhaps due to the existence of /oo?/ cheek of buttock: older speakers alternate /dyoo?/ with /oo?/ in phrase-internal position in texts. Speakers do not accept the alternation /No? ~ go?/ of (11c), although it occurs in texts; this is perhaps due to the existence of /go?/ to regrind corn dough.

nasal prosody in chimila

11

carry), are sufxed by a limited set of sufxes to form lexical stems. Most of these sufxes indicate direction of movement or location. Their function in stem formation is illustrated in (14) for the following: /-Na/ stem, /-ka/ away, out, /-k\i/ elevated, /-mta/ inside, /-ta/ up, toward, and /-t?/ down, away: (14a) [sa?ndi?] [sa?Nati?] [sa?Ng\i] [sa?ndka] (14b) [uuk\i] [uuti?] [uumata] (14c) [moota] [moomata] [mootak\i] /sa?-t?/ /sa?-Na-t?/ /sa?-k\i/ /sa?-ta-ka/ /u-k\i/ /u-t?/ /u-mta/ /mo-ta/ /mo-mta/ /mo-ta-k\i/ get down (off donkey) took (child) down (off donkey) impale (with something long) slip out (something long) sit on chair sit down on ground sit inside something carry upslope carry below carry from below to an elevated position

Examples of an alternation which must be triggered by lexical nasality appear in (15) (intransitive verb) and (16) (transitive verb), where [-ta] up, toward occurs in alternation with [-na] up, toward. These allomorphs occur when the sufx [-ta] is afxed to a verbal root in order to form a verb stem. The lexical form for [-na] appears when the stem is inected to form the imperative in (15c), because the imperative morpheme /-ya/ (here with the allomorph [ca]; see 4.3) interposes itself between the verb root [hi] and the sufx [-ta]. It also appears in (16b), where the stem-forming morpheme [-ko] interposes itself between the verb root and the sufx [-ta]: (15a) [h] /hY/ to creep along

(15b) [hnak\] /hY-ta-k\i/ creep-up-elevated to go up a slope (15c) [htyatk\] /hY-tya-ta-k\i/ creep-impsg-up-elevated go up!

12

international journal of american linguistics

(16a) [tn]

/t-ta/

shuck corn

(16b) [tkot] /t-ko-ta/ shuck-stem-toward shuck corn! This alternation cannot be attributed to lexical tone, because the syllable preceding [-ta] in (16a) does not bear lexical tone (see also 20 and 21 below). In the case of a handful of morphemes, tone does condition alternations between voiceless stops and nasal consonants. For example, [-nta] negative (conditional) occurs immediately following high tone (17a) and [-na] occurs elsewhere (17b and 17c). Note that the consonant variant immediately following high tone is always a cluster composed of a nasal consonant and a voiceless stop, in obedience to the requirement that CV syllables bearing lexical tone must be heavy (i.e., bimoraic): (17a) [mka\t hti\ htaNnt] /ma-uka-\a-ti hti\a hta-Ne-t/ if.fut-2-pl-cont this.way do-cond-neg If you (pl) dont do it this way . . . (17b) [mkat be nyan kumya\ yNen] /maka-ti be neya-ni since.3fut-cont where still-3sg.gen yA-Ne-t/ be-cond-neg Since there wont be food anywhere . . . (17c) [s-t NapNa\ kken?win ki\?] /isa-ti NaapNa-\i k-ke-t-?wi-na if.3-cont wildcat-top eat-cond-neg-inv-1sggen If the wildcat hadnt eaten my chickens . . . Examples (18)(21) illustrate the variation between [-na] and [-ta] up, toward for verb roots beginning with voiceless stops (18), voiced stops (19), vowel-initial roots (20), and semiconsonants (21). (18) [c?-mi-k \esa ku-n] /tya?-mi-ka \eesa kVU-ta/ why-not-2int cow milk-toward why dont you (sg) do the milking? ki\o?/ chicken(s) kumiya-\i food-top

nasal prosody in chimila

13

[c?-mi-k \esa ku-kwi-t] /tya?-mi-ka \esa kVU-wi-ta/ why-not-2int cow milk-inv-toward why dont you (sg) milk the cow? (19) [ndn] /dO-ta/ enter-toward to enter [ndcat] /dO-ya-ta/ enter-imp-toward enter! (20) [n] /U-ta/ drink-toward to drink (through a straw) [kat] /U-ka-ta/ drink-stem-toward sip (through a straw)! /wI-ta/ pour out liquid (21) [wn] [wkothnin] /wI-ko-ta-ni-nu/ pour-stem-out-ben-1sg pour it out for me! This class of verb roots contrasts with other roots for which the nasal variant [-na] never occurs and which have identical phonetic tone patterns, as illustrated in (22a) and (22b): (22a) [ku-ta-k-nt] /k-ta-ka-de/ go.up-up-stem-1ev Im getting up

14

international journal of american linguistics

[ku-ya-t] /k-ya-ta/ go.up-2impsg-up get up! (22b) [ku-na-k-nt] /kVU-ta-ka-de/ Im milking

One can most easily account for the morphophonemic alternation in (22) by specifying the root to milk as nasal in the lexicon and the root to move up/away from source as oral. Under this analysis /-ta/ and other sufxes with nasal allomorphs are unspecied for nasality, whereas sufxes for which nasal allomorphs do not occur, such as /-k\i/ elevated, /-ka/ stem, /t?/ down, away, are specied as [-nasal]. Examples above suggest that the stem is the limiting domain for nasal prosody; however, forms such as [hri\ h-Ne-t-k\] lets go upstairs!, where [-ta] is preceded by the nasalized plural sufx [-Ne] (see 23a below), demonstrate that nasal prosody is limited to the sufx immediately adjacent to the lexical root. Vowels adjacent to the nasal consonant are phonetically nasalized by spreading of [+nasal] from the consonant in both directions. The most obvious alternative solution is to set up verb classes, one containing verb stems ending in [-na] and the other in [-ta], with the stipulation that [-na] has the variant [-ta] when other inectional morphemes intervene between it and the verb root. Nevertheless, this solution is awkward: it forces the arbitrary marking of verb classes in the lexicon for morphophonemic variants that obviously share some sort of phonological relationship, and results in multiple lexical sufxes with exactly the same meaning. In addition, generalizations are lost, given that verbs requiring the [-na] morphophonemic variation also require [-Ne] or [-ne] root plurals (23a) instead of the [-ke] or [-\e] root plurals required by verbs sufxed with the [-ta] allomorph (23b), and the nasal allomorph [-Nwa] nominalizer/middle voice instead of [-wa] as required by verbs sufxed with [-ta] (24a and 24b):10 (23a) [h-na-k\] to go up a slope [h\i-\ h-Ne-t-k\] /hu\i\a hY-ke-ta-k\i/ 1.imppl creep.along-pl-up-elevated lets go upstairs! /hYtak\i/

10 The initial consonant of the plural allomorph is controlled by the initial consonant of the verb root: [-ke] or [-Ne] occurs when roots begin with /V/ ([?V]), /k/, /N/, or /h/; otherwise [-\e] or [-ne] occurs.

nasal prosody in chimila

15

(23b) [u-t] to sit up [u-ke-w\a?y-t] /u-ke-wa\a?ya-ta/ sit-pl-imppl-up sit up! (pl) (24a) [uku] drink! [una] /U-ta/ drink-toward to drink with a straw [u-ku-ta] /U-ko-ta/ drink-stem-up drink with (the) straw! [kape uNwa-?a] /kpe U-wa-a/ coffee drinking-is its time to drink coffee

/uta/

/U-ko/

(24b) [ndina] remove hair from goat carcass [ndi-ko-ta] /dY-ko-ta/ cut-stem-up take the hair off! [ni ndi-Nwa-?a-k\i] /ni dY-wa-a-k\i/ gen cutting-nom-is-elevated what was trimmed off

/dY-ta/

Compare the nominalized forms in (24a) and (24b) with the corresponding forms of (25a) and (25b). (25a) [kii?\o] [kii?\-wa] (25b) [kake?] [kake-wa?]
11

to work work (noun) to be named name11

/ki?\/ /ki?\wa/ /kke?/ /kke?wa/

Word-nal glottal is always present for some words. For other words there is considerable variation between speakers as to whether the word-nal glottal is present or not.

16

international journal of american linguistics

Further, verbs requiring the [-na] allomorph of the directional morpheme up, toward also require the [-Na] allomorph of the directional morpheme [-ka]: (26a) [aNa] go away [acaka] /A-ya-ka/ go-2impsg-away go away! (26b) [ndiNa] cut off a piece [ndi-ko-ka] /dY-ko-ka/ cut-stem-away cut off a piece! Other verbs show analogous variants for the following stem-forming sufxes: /-ta/ ([-na]) up, toward; /-t?/ ([-ne?]) distributed; /-ka/ ([-Na]) away, out; /-\e/ ([-ne]) pl (subject); /-ke/ ([-Ne]) pl subject); /-wa/ ([-Nwa]) mid/nomz; and possibly /-\i/ ([-ni]) adjective. Verbs with roots that trigger nasal variants of one or more of these sufxes are listed in Appendix A; out of approximately 590 verbs appearing so far in the data, 77 (13%) have occurred in morphophonemic contexts indicating that the lexical root must be nasal. 4.2. The glottal off-glide. 4.2.1. The intransitive causative. The same roots that require [NV] allomorphs of certain sufxes also require the insertion of a glottal off-glide ([h]) immediately preceding the transitivizing or transitive causative sufxes /Na/ and /No?/. When transitivized by the sufxes /-No?/ or /-Na/, verb roots requiring the [-na] form of the directional morpheme change their phonetic form from [(C)V(V)] to [(C)V(V)h]in other words, the root is immediately followed by a glottal off-glide. This alternation is illustrated below for the verb stems /A-ta/ to return (27a), /dO-ta/ to enter (27b), and /mA-ta/ to go by (27c). (27a) [ah-Na-ta] go-stem-toward bring (it) back! (27b) [ndoh-No?-ta] enter-stem-toward put in, cause to enter /dY-ka/ /A-ka/

nasal prosody in chimila

17

(27c) [maah-Na-ta pass-stem-toward keep passing it on!

ni ser

yu?u] go.2impsg

The behavior of these verbs contrasts with the behavior of verbs for which the morpheme /-ta/ appears sufxed to the root, such as /kta/ to go out: when these latter verbs occur with /-No?/ or /-Na/ transitive stem the glottal off-glide does not appear. In (28) the morpheme /-ta/ occurs following the verb root /k/, the same allomorph of the verb root occurs preceding the stem-forming sufx /-Na/, and the same allomorph of the directional morpheme /-ta/ follows: (28) /k-ta/ open-toward go out (through a door) [ka-Na-ta] open-stem-toward open the gate! Thus the glottal off-glide alternation before verbs analogous to /A-ta/ return appears to be triggered by the same factor triggering the variation between [-ta] and [-na]the verb root is marked for nasality in the lexicon. Tone cannot be triggering the glottal off-glide: note that the phonetic and lexical tone patterns of the lexical root and the whole word cross a street (29a) and pass on (29b) are the same: (29a) [manak] /mA-ta-ka/ pass-toward-out cross a street (29b) [mahNat] /mA-Na-ta/ pass-stem-toward pass (something) on Equally, forms such as (29b) preclude one from arguing that the off-glide is inserted to provide a mora in obedience to the requirement that a syllable containing a vowel with lexical tone be heavy. In fact, the interaction of lexical tone with the metrical system for CV syllables, in which the onset of the following syllable lengthens in order to provide an extra mora, is blocked precisely in this causative (see 27a and 27b). This indicates that the insertion

18

international journal of american linguistics

of the off-glide is triggered by the morphology; otherwise this blocking should not occur.12 Contrasts such as (30a) and (30b) conrm the morphological function of the off-glide: in these examples it refers to an unspecied causative agent. (30a) [nAg\i] /nA-k\i/ be vertical-elevated come from (higher elevation) [nAhg\i] /nA-h-k\i/ be vertical-caus-elevated be set upright, set upright (30b) [kug\i] /kVV-k\i/ be.burned-elevated be burned and stuck on [kuhg\i] /kVV-h-k\i/ be.burned-caus-elevated be deliberately burned, and stuck on, deliberately burn and stick on The only viable alternative way to account for the insertion of the glottal offglide, other than specifying nasality in the lexicon, is to establish verb classesand exactly the same arguments apply as in the case of the allomorph [-na] of /-ta/. 4.2.2. Isolated verb roots and compound words. The connection of the off-glide with nasality is not as exotic as rst appears, when acoustic facts are taken into consideration: Blevins and Garrett (1992:14) note work by previous investigators indicating that phonation effects associated with [h] or breathy voice and vowel nasalization affect vowel spectra identically. This indicates that a natural acoustic and articulatory relationship exists between [h] or breathy voice and vowel nasalization. Beddor (1993:184) further notes that breathy voicing mimics certain of the effects of nasalization on the low-frequency spectrum. The connection can be seen in the Chimila data, not only in the case of the intransitive causative but also when nasal verb roots of the form CV occur alone (31a) or in aspectual or serial verb constructions (31b), in that the isolated verb root always has the form
In CV roots the morphologically triggered glottal off-glide does provide the second mora, fullling the requirement that syllables bearing lexical tone must be heavy (27a and 27b).
12

nasal prosody in chimila

19

[(C)Vh]. In these cases the stricture of the off-glide is heavy. (31c) shows that the off-glide is not inserted due to the serial verb linker /ni/: (31a) [isa sah] /isa s/ 3.conj grind its ground/it was ground (31b) [isa bih ni tekwa] /isa bY 3.conj roll.up.in.ball

ni ser

te-kwa/ position (round object)-extended surface it rolls up in a ball (centipede) (31c) [isa ana ni yu?u] /isa A-ta conj go-toward yu?-a/ go-cont

ni ser

its returning (toward speaker) The connection between the off-glide and lexical nasality can also account for the puzzling but consistent difference in stricture noted in 2 above. Upon sufxation isolated roots with heavier stricture turn out to be marked for nasality, and those with little or no stricture are oral. The stricture functions as a mora inserted to form a bimoraic foot, as stipulated by the minimal word constraint; light stricture disappears when the root is incorporated into a compound word ([witu?] be acquainted), whereas heavier stricture does not ([sihtu?] be black). In be acquainted the rst syllable forms a defective foot, but the bimoraic foot of the second member fullls the minimal word constraint, so that one must conclude that the off-glide of be black has not been inserted in order to fulll the minimal word constraint. The distribution of the glottal off-glide in nouns and adjectives also reveals a close association with nasality: in the vast majority of cases it occurs in stem-internal position preceding nasal consonants, the lateral sonorant /l/, or voiced stops (always prenasalized). In syllable-nal position its stricture does not vary according to the preceding vowel, in contrast to word-initial occurrences. In syllable-nal position it tends to blend with immediately following nasals, so that the nasal devoices to varying degrees;13 following
13 For some speakers the syllable /NV/, when immediately preceded by the glottal off-glide, takes the form [Hv$ ]; in other words a nasalized glottal fricative replaces the nasalized consonant and the following vowel is heavily nasalized. See (5) in 2 above.

20

international journal of american linguistics

bimoraic vowels, it tends to blend with immediately following nasal consonants. In words of tone pattern HML it occurs exclusively preceding nasal consonants or the lateral /l/: (32) [khmet] [thnidy?] [shnak] thin, delicate skunk night [\hNak?] [mbhlady?] [lhlond?] brush bow (weapon) hollow (adj)

In words of tone pattern HL the glottal off-glide occurs preceding nasal consonants or voiced stops: (33) [khm?] a little bit [mbhn?] two [ndhN?] dough [ndhb\] lazy

In words of phonetic tone pattern MHL the glottal off-glide occurs in the coda of the leftmost syllable preceding stops, nasals, or the liquid /l/: (34) [ehmt] [hohllond?] [kahmb\nt] go on an outing hollow (adj.) piece of wood [kahns] root [gwahNt] clear! (a eld) [ndihtnt?] shine

Words in which the off-glide immediately precedes oral stops are always nouns and adjectives formed by compounding a nominal root with an adjective (piece of wood), or they are compound verbs (lazy), or reduplicated forms (shine). Only in words of three or more syllables with initial tone pattern MH do voiceless stops appear following the glottal off-glide, and all the examples in the data involve reduplication.14 In contrast to verbs, there is less morphological evidence associating the glottal off-glide and nasality in nouns and adjectives. First, only seven monosyllabic noun roots of the form [(C)Vh] occur in the data: (35) [ah] [boh] [dah] [dih] hair manure nose tooth [gah] [kah] [mah] wing tree sweet potato

Of these, three occur with classier sufxes: (36) [NNgu\] /A-gw\a/ hair-branch body hair
See Malone (2004) for details on noun morphology, where I argue that all multisyllabic nouns are the result of word compounding or the addition of classier sufxes.
14

nasal prosody in chimila

21

(37a) [NgNNgu\] /gA-gw\a/ feather-branch arm (37b) [NgaNNk\?] /gA-kik\a?/ feather-bone wing (38) [kNNg\?] /kA-g\a?/ wood-border.of arrow Nasal prosody has occurred in (37b), where the rst syllable of the classier sufx /-kik\a?/ bone has nasalized.15 According to the analysis of Malone (2006) one must further conclude that feather bears lexical tone, because the onset of the following syllable lengthens, in response to the requirement that CV syllables in which the vowel bears tone must be heavy. In addition, the leftmost syllables of (36), (37a), and (38) all sport nasal consonant codas; according to the analysis of Malone (2006) the vowels of these syllables bear lexical tone, and the consonant has been inserted in order to provide the required second mora. Thus the nasal consonant represents the inserted mora in syllables when the lexical root containing the tone-bearing vowel is lexically nasal. Note that the behaviors described here exclude an analysis of the off-glide as a breathy or voiceless vowel. This would result in an unnecessary proliferation of phonemic vowel contrasts,16 when an entity already demonstrated to exist in the lexicon (nasality) can account for the same data. Also, it ignores the limited distribution of the glottal off-glide, its close relationship with nasal consonants and roots independently demonstrated to be nasal, and the morphophonemic data of (36)(38). 4.2.3. Other occurrences of the glottal off-glide. The following verbal sufxes can trigger a glottal off-glide when occurring on the right of the word: [-No?] transitive stem; [-Noye?] transitive causative; [-mata] inside
The sufx originates from the lexical root /kik\a?/ bone. This suggests that some lexical roots might be unmarked for nasality. At present this is the only potentially unmarked root that has appeared in the data. 16 Cho (1993:64) argues that the so-called voiceless vowels found in some languages have to be represented phonologically as aspirates; the Chimila data would appear to lend support to her proposal.
15

22

international journal of american linguistics

container; [-me] want; be able; [-mesa] purpose; [-na] 1.sg.poss; [-na\a] 1.pl.poss; [-ne] pl; [-ne?] pas; [-ni] 3.sg.poss; [-ni] ben; [-nine] 3.pl.poss; [-nu?] 1sg.obj; [-nu\a?] 1.pl.obj. The glottal off-glide is inserted immediately preceding the initial consonant of the sufx. All of these sufxes begin with a nasal consonant, and the vowel following the nasal consonant is heavily nasalized, whether or not the off-glide precedes itin contrast to the lighter nasalization on the vowels of syllables headed by nasal consonants in words such as [kene?] eat or [ndanu] be afraid. At rst glance one might suppose that the off-glide is inserted in order to prevent the occurrence of unfooted syllables (39), and this is often the case. However, it occurs in environments where if it were not inserted, no unfooted syllables would occur (40). Its insertion can result in unfooted syllables not only in the stem (always footed separately before sufxing) but also when it immediately precedes a disyllabic sufx (41). (39) [i sa ta a \ah ne] | | | | || | (mm) (mm) (mm)< m> /isa conj taa\a-ne$/ talk-pl

they are talking (40) [i h ne anawah ne] | | | || | || | (mm)<m>(mm)m(mm)<m> /inE? conj.pas A-ta-wa-ne$/ go-toward-pat-pl

they returned it (41) [maka wesa ooniwahnine ogwe?] | | | | | | | | || | | | | | (m m) (mm)m(mm)m (mm)m<m>(mm)<m> /maka conj.fut wesa really oni-wa-nIne$ cry-pat-their owe/ child

they will really mourn for their son/daughter In 4.1 we saw that sufxes exist which must be marked [-nasal] in the lexicon, and others which must be unmarked for nasality; one is tempted to conclude that the set of sufxes behaving analogous to [-ne] in (39)(41) is [+nasal]. This occurrence of the off-glide also raises the question of whether nasal prosody could be operating from right to left. Some evidence exists,

nasal prosody in chimila

23

for example (42a), where the voiced stop of [mbhn?] becomes a nasal consonant at the same point of articulation when prexed; (42b), with the same classier prexed to three, is included for comparison. (42a) [mbhn?] (42b) [mahn?] two (people) three (people) [ti-mhn?] two (birds) [ti-mahn?] three (birds)

Another possible example is (43), where the dental stop of the sufx /-ta/ becomes a nasal consonant at the same point of articulation when sufxed by /-mta/: (43) [ihne ha\u p\antamata haatakanamata] /inE? ha\u p\-ta-mta conj.pas all fall (drops)-toward-inside The (rain) got into everything in the house. The evidence is, however, ambiguous. Regarding (42), numerical classiers are the only prexes occurring in the data, and whether or not nasal prosody has occurred, voiced stops do not occur stem-internally in core lexemes in intervocalic position. In (43), the sufx /-namta/ could also originate from /nA/ be vertical and /-mta/ inside. If right-to-left nasal prosody can be unambiguously shown to occur, there are restrictions, depending on whether it originates with a lexically nasal root (42a) or a sufx (39, 40, 41, and 43): nasal prosody originating from sufxes cannot nasalize consonants, whereas nasal prosody originating from roots can. 4.3. Allomorphs of the intransitive imperative. If roots analogous to /A/ to go are lexically nasal, the insertion of the voiceless stops [t] and [k] in the imperative forms for roots containing bimoraic vowels such as /A-ta/ to return (44), /dO-ta/ to enter (45), and /mA-ta/ to go by (46) must be explained: one does not expect voiceless stops to be inserted after nasal morphemes in a language allowing nasal prosody. These stops occur in the coda of the syllable containing CV verb roots (44 and 45), whereas they occur in the onset of the imperative morpheme for CV and CVV verb roots (46): (44) [at-ya-ta] go-2impsg-toward return! (sg) [ak-wa?\e-ta] go-2imppl-toward return! (pl)

haataka-ta-mta/ house-toward-inside

24

international journal of american linguistics

(45) [ndot-ya-ta] enter-2impsg-toward enter! (sg) [ndok-wa?\e-ta] enter-2imppl-toward enter! (pl) (46) [ma-tya-t] pass-2impsg-toward pass by, go on! (sg) [maa-kwa?\e-ta] pass-2imppl-toward pass by, go on! (pl) The root /mA-/ already constitutes a minimal foot (thus fullling the minimal word requirement), so that inserted consonants (especially in the following syllable onset) would seem superuous.17 Nevertheless, all intransitive verbs which can independently be shown to have nasal roots insert [t] in the singular and [k] in the plural imperative, no matter what their syllable structure is.18 In contrast, most intransitive imperatives inect as in (47) below. Consonants other than the glottal stop do not appear between the verb root and the imperative morphemes in the singular and plural imperatives; note that the occurrence of the glottal stop is not predictable for disyllabic verbs. FurtherOnly four CVV verb roots with /aa/ occur so far in my data: /gw/ be positioned (something at, wide and oppy), /mA-/ pass by, /nA-/ be vertical, and /NA/ lie in hammock. /mA-/, /gw/, and /NA-/ condition nasal allophones. /nA-/ requires the plural morpheme [-\e] (instead of [-ne]) but otherwise behaves identically to /mA/ and /NA/. The occurrence of [na\] instead of *[nn] is the only exception I have been able to nd with respect to the conditioning of nasal allomorphs reported in this paper. 18 Consonants are also inserted immediately following nasal roots for the following sufxes: /-wi/ [-wi ~ -kwi] inv (compare 17 to 48c); /ya/ [-ya ~ -t ca] cond; /-wa/ [-wa ~ -kwa] pat; and the transitive plural imperative /-waro/ [-waro ~ -kwaro]. Malone (2006) represents the underlying form for the allomorph [-kwi] as /-kawi/. This representation fails to account for the fact that /-kawi/ would only be sufxed to verbs with nasal roots (i ), whereas when [-wi] does not appear [-ka] is sufxed to both nasal and nonnasal roots (ii ). (i ) [owi] s/he did it (ii ) [oka] someone did it /o-wi/ /o-ka/ [ukwi] s/he drank it [uka] someone drank it /U-wi/ /U-ka/
17

The representation /-kawi/ also obscures the analogy of the [k] insertion for /-wi/ with other morphemes where the presence of /-ka/ cannot be justied, as well as the analogy with [t] insertion: a generalization is lost.

nasal prosody in chimila

25

more, in the phonetic forms for the singular and plural imperatives, the light stops preceding the glides /y/ and /w/ are voiced instead of voiceless. No verbs independently shown to have nasal roots inect the imperative in this manner. (47a) /kta/ /k-ya-ta/ /k-wa?\e-ta/ (47b) /h-ta/ /h-ta-ya/ /hta-?wa\a-ya/ (47c) /k-ta?/ /k-ta-?ya/ /kta-wa\a-?ya/ [kata] [kadyata] [kagwa?\eta] [hata] [hatadya] [hata?gwa\adya] [kita?] [kita?dya] [kitagwa\a?dya] to go out get out! (sg) get out! (pl) to come come! (sg) come! (pl) to sing sing! (sg) sing! (pl)

At the same time, roots showing no evidence elsewhere in the language for lexical nasality require the insertion of the voiceless stops in the imperative, as in (48) with a disyllabic root ending in a syllable bearing lexical high tone. In all these cases a CV syllable with lexical tone immediately precedes the imperative morpheme, and the CV syllable is always the rightmost syllable in a multisyllabic stem: (48) /ki?\/ [kii?\tya] [kii?\kwa\a?ya] [kii?\gwa] to work work! (sg) work! (pl) work (noun) /ki?\-ya/ /ki?\-wa\a-?ya/ /ki?\-wa/

CV syllables bearing high tone must form a bimoraic foot when sufxed; the resulting metrical structure is illustrated in (49) for the singular imperative of to work. (49) [kii?\ t.y a] || || | (mm)(mm)<m> work! (sg) One cannot attribute the inserted consonants to tone in lexically nasal roots, because not all lexically nasal roots bear lexical tone (/gw-ya/ [gwaaca] lay an egg!). One is tempted to assume that syllables with lexically nasal vowels must also be heavy. Although there is evidence to support this assumption,19 and it accounts for the behavior of monosyllabic nasal roots, it
Stems such as /t-ta/ shuck corn, /u$ -ka/ drink, and /w$ ta/ pour out water vary with respect to the length of the sufx-initial consonant: sometimes it is single and sometimes it is bimoraic.
19

26

international journal of american linguistics

does not account for bimoraic roots such as /jY/ creep, /gw-/ be positioned (at, oppy object), and /mA-/ pass by. Instead, the inserted voiceless stop for the imperatives of roots analogous to /A/ and /mA-/ appears to block nasal prosody (15c, 19, 26a, 44, 45, and 46); the voiced transition stops do not, as is clear from (24a) and (24b) above. The constraints for apparent blocking appear to be related to the boundary between derivation and inection: nominalization and sufxation of directional/location morphemes create stems which in turn are a modication of inherent meanings of roots, whereas imperatives do not affect the inherent meaning of the root (or, for that matter, the stem). When imperatives are formed from stems, the stem-forming sufxes must separate themselves from the root, the inectional morpheme is sufxed to the root, and the stemforming morpheme is sufxed to the inectional morpheme. When bare roots function as stems (/gw-ya/ [gwaaca] lay an egg!), the relationship between inection and derivation is clear, whereas it is not as clear for derived stems. Thus nasal prosody operates within a stem, but not outside the stem, and the voiceless stop occurs at the root boundary; in the case of nonnasal verb stems analogous to /ki?\/ the voiceless stop occurs at the stem boundary. Transitive imperatives such as [kwakata] pour! and [ukuta] drink with (the) straw! demonstrate what is to be expected if nasal prosody is indeed in operation: the voiceless stop in the onset of the transitive stemforming morphemes /-ka/ and /-ko/ blocks nasal prosody, as is the case for other languages in which nasal prosody is known to operate (see Barnes 1996 for an example). Here one can assume that the transitive stem-forming morphemes /-ko/ and /-ka/ and the singular imperative /-ya/ and the plural /-wa\a/ are specied as [-nasal], whereas the directional morpheme /-ka/ is unmarked for nasality. Thus the inserted stops signal the root boundary; they also indicate that the preceding root is nasal and the immediately following sufx is oral, i.e., they signal a transition from nasality to orality. One might ask if the transition from orality to nasality is also signaled. Under certain conditions this seems to be the case, for morphemes of the form /-NV(CV)(?)/ at the right edge of the phonological word, as described in 4.2.3. However, a comparison of (50a)(50c) indicates that this is not the primary function of the inserted off-glide: in (50b) an alveopalatal affricate [c] (/ty/) appears instead of the expected glide [dy] (compare to 50c).
Speakers tend to lengthen stem-internal nasals; they also tend to lengthen the sufx-initial consonant following CV rootsas if the minimal word constraint were in the process of being reinterpreted as a minimal root constraint. Comparison with sufx-initial consonants in words such as /-kwa/ (cut-extended surface), whose roots are independently known to bear lexical tone, suggests that stems whose sufx-initial consonants vary in length are unmarked for lexical tone. Nasal vowels are not mentioned as instantiating syllable weight in the comprehensive survey of Gordon (2002).

nasal prosody in chimila

27

(50a) [maka /maka conj.fut

ooniwahnu] oni-wa-nU/ cry-pat-1objsg

s/he will mourn for me (50b) [maka /maka conj.fut ooniwatcu] oni-wa-yu/ cry-pat-2objsg

s/he will mourn for you (50c) [maka /maka conj.fut wiwidyu] wi-wi-yu/ send-inv-2objsg

s/he will send you /-wa/ pat is lexically oral, so the inserted stop in (50b) must instead signal the stem boundary when the stem is sufxed by an oral morpheme, much as the inserted stops of the intransitive imperatives indicate the root boundary of nasal roots followed by oral inectional morphemes. 5. Analysis. In this paper I have described phonetic and morphological behaviors pointing to the existence of lexical nasality or associated with nasality in Chimila. In this section I tie together loose ends of the analysis and demonstrate that a relationship exists between these seemingly disparate phenomena, using the proposal of Zoll (1996), who argues that subsegments such as nasality can express themselves as latent segments (inserted segments), attach to an already present segment as a feature ([+nasal]), or oat (remain unexpressed phonetically). 5.1. Lexical nasality in core lexemes. The nasal allomorphs documented in 4.1 leave little doubt that nasality is operating in the Chimila lexicon. There we saw that there must be verb roots marked [+nasal] in the lexicon; there must also be sufxes marked [-nasal] and [0nasal]. A curious gap in the inventory of sufxes (none were marked [+nasal]) was resolved, once it was demonstrated in 4.2 that the glottal off-glide is intimately related to nasality. If, as proposed in 4.2.3 and 4.3, its insertion and the insertion of the stops [t] and [k] in the intransitive imperative signal a transition between roots and sufxes differing with respect to their lexical specications for nasality ([h] signals a transition to a [+nasal] sufx and the stops signal a transition to an oral sufx), then [+nasal] sufxes must also exist. Thus the insertion of these transitional segments is controlled by lexical specications

28

international journal of american linguistics

for nasality, in interaction with morpheme class (root or sufx), position in the phonological word, and lexical tone. The assumption that morphemes in the core lexicon are marked for nasality avoids the theoretically awkward, ad hoc specication of some sufxinitial stops (for instance, the velar stops of /-ka/ stem or /-ke?/ aside) as blockers of nasal prosody, and others as unmarked for nasality (for instance, the velar stops of /-ka/ out, away or /-ke/ pl). Under this analysis the feature [+/-nasal] spreads onto root nodes dominated by the morpheme in question, in interaction with language-specic constraints on feature specication. This analysis is different from that of Walker (1998) (inter alia), who attributes the phenomena of blocking and transparent consonants to opacity effects. Her analysis calls on a phonological representation that may never surface because it cannot be physically implemented (Walker 1998:100), i.e., nasalized voiceless stops cannot be phonetically implemented. Nevertheless, in Chimila voiceless nasalized stops are phonetically implementedas nasal consonants at the same point of articulation. This (along with other data presented in this paper) indicates that lexical nasality in Chimila is not expressed phonetically on vowels in core lexemes unless they are contiguous to a nasal consonanta unique trait, judging by the comprehensive surveys of languages with nasal prosody in Piggott (1992) and Walker (1998). Lexical nasality is also unique in that it expresses itself as syllable-nal nasal consonants in non-word-nal syllables bearing high tone, as the glottal off-glide in roots functioning as isolated words, or word-nally in isolated verb and noun roots or in the rst member of compound words, and it indirectly manifests its presence through consonantal transitions between roots and sufxes ([h], [t], and [k]). The data force the conclusion that lexical nasality must be expressed as a nonvocalic subsegmental feature: it can only dock on consonants or, alternatively, it can only dock on syllable margins. It either expresses itself as a latent segment (inserted nasal consonants or the glottal off-glide, depending on its position in the word), attaches to an already present segment as the feature [+nasal], or oats (remains unexpressed phonetically), as is the case with /U-ka/ [uka] drink (compare /o-ka/ [oka] do).20 Regarding the representation of lexical nasality, mora insertion has been independently documented for Chimila in Malone (2006), in response to the requirement that syllables bearing lexical tone be heavy; this involves the insertion of a timing unit in CV syllables bearing lexical tone, whether or not a vowel is lexically nasal. In these cases, place features are copied from the following syllable onset; place features are also copied when nasal
Terminology is from Zoll (1996). I am grateful to an anonymous referee for referring me to Zolls work.
20

nasal prosody in chimila

29

Morpheme tier

M |\

Moraic tier

| \ | \ \

m /| / | [+cons] | | | | m
Fig. 1

Root nodes [-cons] Laryngeal node /

/ | [+nasal] / | | |

[tone] | | [pr o

pr o n i ?]

consonants are inserted. This is suggested by a variation which occurs between nasal consonants with place features and a nasal consonant [N] in syllable-nal position, as in [k\iNk\initi? ~ k\iNk\initi?] stay (pl), [kantawa? ~ kaNtawa?] jungle, or [p\omp\oni? ~ p\oNp\oni?] be face down (pl), where [N] is articulated in the boundary area at the rear of the alveopalatal region and the front edge of the velar region; it is further back than the alveolaralveopalatal glide [dy] but not as far back as the velar nasal [N].21 If a CV verb root is marked [+nasal], nasality attaches to the inserted (righthand) node, as in gure 1, for the leftmost syllable of /p\Op\OnI?/ [p\omp\oni?] (only mora-bearing root nodes are shown in gure 1). In fact, the requirement that syllables bearing lexical tone be heavy may have motivated the requirement that lexical nasality can only be expressed as a consonantal feature. One sees this in the case of morphemes which block the insertion of the glottal off-glide when they immediately precede sufxes which otherwise would trigger it (51a, where the sufx /-wi/ immediately
21 By implication, [N] cannot have place features, whereas other nasals (including the coronal nasal) in Chimila must have place features. This nasal, preceded by a phonetically nasal vowel, also appears as a cross-generational equivalent for nasal vowels in lexemes originating from onomatopoeia: compare [s?sa s?sa s?] stinging sensation, as pronounced by an older speaker, and the pronunciation of the same word by her daughter (also a uent, competent speaker): [sNksa sNksa sNk]. (The velar stop is fronted and unreleased; it appears to be transitional between the nasal consonant and the voiceless fricative.)

30

international journal of american linguistics

precedes the sufx /-ne$/), and in the case of verbs which bear lexical tone on the second syllable of the stem (51b, where the stem /wiwakwa/ immediately precedes the sufx /-ne$/). In (51b) the deletion of a morpheme triggers a tone shift (Malone 2006).22 This in turn triggers lengthening of the following consonant, in contrast to (51c), where the glottal off-glide occurs: (51a) [gwah Nowin e] || | | | | (mm)m(m m)<m> /gw-NO?-w-ne$/ kill-stem-inv-pl they killed it (51b) [wikw an e] | | || | (m m)(mm)<m> /wi-wa-kwa-ne$/ command-mid-extended-pl they sold (him/her/it) (51c) [wikw a h ne] | | || | (m m)(mm)<m> /wi-kwa-ne$/ command-extended-pl they sent (him/her/it) The following morphemes block the insertion of the glottal off-glide in this environment: [-No? ~ -ko] transitive stem; [-ne?] pas; [-ti] cont; [-ti?] down, away; [-wi] inv; [-ya?] fut; [-e? ~ -ye?] affected pat; [-yo?] transitive stem. All these sufxes bear lexical tone and therefore trigger doubling of the immediately following syllable onset (including nasal consonants), in obedience to the requirement that syllables bearing lexical tone must be heavy. The proposed representation suggests that the mysterious glottal off-glide [h] in root-nal position is best considered to be an allophone of the glottal fricative /h/, in spite of its phonetic differences and restricted distribution; in word-nal position (in isolated and in compound words) it functions as do nasal consonants inserted in word-medial position. The inventory of segments in word-nal position is more severely restricted than that of stem/-wa/ mid is included in the lexical form of (51b), on the basis of the meaning. The exact identity of the omitted syllable is uncertain: /-ya/ dative obj is another possibility.
22

nasal prosody in chimila

31

medial, syllable-nal segments: it is limited to [?], [h], [m], and [N]. The word-nal nasal consonants occur only in lexemes originating from loans or onomatopoeia that are in the process of migrating into the core lexicon, so that one can safely conclude that segments with place features are generally not allowed in word-nal position in core lexemes, i.e., only laryngeal consonants can occur in syllable codas in word-nal position. The observed relationship between the glottal off-glide and nasality suggests that the off-glide must be a sonorant; Hall (2000) proposes this representation for /h/ and /?/ in general, and Um (1996) argues that the representation of laryngeals with respect to [sonorant] depends on their behavior in the language in question. Obviously, the off-glide must bear the feature [+nasal], even though one hears no phonetic nasalization when a nasal consonant is not present. The representation in gure 1 accounts for the inserted glottal off-glide in words such as /kA/ [kah] tree, where the insertion of a nasal consonant is blocked in word-nal position because there is no way for it to acquire place features. Figure 1 also accounts for the insertion of [t] and [k] in the codas of CV verb roots for the intransitive imperative and the glottal off-glide; the constraint [-nasal] [-sonorant] accounts for the voicing specication of the inserted stops. In the case of CV and CVV verb roots, a representation such as that proposed in Steriade (1994) for Mazatec consonant clusters accounts for the phonetic association with the onset of the immediately following syllable. (52a) below gives representations for the stop /t/ and the glide /y/ [dy], where A0 represents the closure phase of the segment and Amax represents the release phase (Steriade 1994); the two phases are dominated by a single node.23 (52b) represents the cluster /ty/ [c]: The A0 and Amax phases of the onset stop have merged with the respective A0 and Amax phases of the glide, and the result is a single segment (dominated by one node). Place assimilation (both segments are coronal) triggers the merger. (52a) t | A0 Amax (52b) y | | A0 Amax
t

y | | A0 Amax

The inserted stop of the singular imperative /mA-ya-ta/ [maacata] pass by! can be represented as in gure 2; the representation for the inserted stop in the plural imperative /mA-wa?\e-ta/ [ma-kwa?\-t] is analogous (see
23 Steriade (1994:207) represents glides as only having Amax. The light stops preceding the Chimila labial and coronal glides suggest that they also have an A0 phase. For /y/, [+coronal], and for /w/, [+velar] associate to the A0 phase.

32

international journal of american linguistics

M /| [+nasal] | | t | A

M | \ | [-nasal] | y | Amax
Fig. 2

M /| [+nasal] | | k | A0

M | \ | [-nasal] | w | Amax
Fig. 3

gure 3). Again, the respective A0 and Amax phases of the stop and glide merge. The merger occurs in the onset of an oral morpheme, so the association with the nasal morpheme is deleted (M represents the morpheme tier). The representation in gure 4 accounts for cases such as /Uka/ [uka] to drink; note that [+nasal] oats, because there is nowhere it can dock within the verb stem (root nodes not bearing moras are omitted). 5.2. Lexical nasality in interjections and onomatopoeia. A referee has proposed that there could be a contrast between oral and nasal /h/ in interjections. However, in core lexemes and onomatopoeia, nasal morphemes can be shown independently to exist; thus there is no justication for proposing an additional nasal segmental phoneme which would only be found in nasal morphemes in interjections. An alternate analysis is available: one only has to specify that lexical nasality associates to vowels, instead of consonants, in interjections. There is little or no evidence for nasal prosody in interjections and onomatopoeia (once they are sufxed they enter into the realm of core lexemes), so that one must also specify that vowels are marked nasal or unmarked for nasality. In the case of onomatopoeia one can either specify vowels as nasal, or whole morphemes; the latter is preferable, given the close relationship of onomatopoeia to the core lexicon. Thus the difference between noncore lexemes and core lexemes with respect to lexical nasality is manifested as a difference in its segmental manifestation; no other explanation is necessary.

nasal prosody in chimila

33

Morpheme tier

M |\

M | \ \ [+nasal] m | | / | [-cons] | | -k
Fig. 4

Moraic tier

| | /|

[-nasal]

Root nodes

[-cons] | | [u

a]

5.3. Variation in realization between voiced stops and nasals. Where does this variation t in the general picture of lexical nasality in Chimila? Some roots which undergo this alternation, such as /d/ pull toward (something long) and /gA/ of /gA-t?-ka/ abandon can be independently shown to be nasal. Walker (1998) states that languages in which both /t/ and /d/ [are] realized as [n] under nasalization are unattested and suggests that this can be understood as a consequence of the highly neutralizing effect of such an outcome (1998:116, n. 15). The alternation may indeed be a synchronic remnant from a time when voiced stops in roots were realized as nasal consonants at the same point of articulation when undergoing nasal prosody. The synchronic occurrence of the alternation shows that the prosody is now optional morpheme-internally and occurs only in contexts where the voicing (i.e., phonemic contrast) of the underlying stop is recoverable. This is certainly the case for sufx allomorphs where nasals alternate with voiced stops within stems. In fact, the ternary specication of nasalization at the morpheme level accounts for patterns of sufx variation within stems and allows one to determine underlying lexical forms. In addition to phonological factors (lexical tone and nasality), the specication of the root with regard to transitivity controls allomorphs. When sufxes such as /-ta/ [-ta ~ -nta ~ -nda ~ -na] up, toward or /-t?/ [-te? ~ -nte? ~ -ne?] distributed (both [0nasal]) form stems with roots unspecied for transitivity, allomorphs analogous to [-nta] occur following vowels bearing high tone (53a), and those analogous to [-nda] occur following oral syllables bearing nonhigh tone (53b). Those analogous to [-na] occur following nasal roots specied for transitivity, either transitive (53c) or intransitive (53d), and alternatives analogous to [-ta] occur elsewhere for all roots (53a53d ):

34

international journal of american linguistics

(53a) /pta/ /pNata/ (53b) /tyi?da/ /tyi?Nata/ (53c) /wIta/ /wIta ditake/ (53d ) /wY-ta/ /wY-Na-ta/

[pnt] [pNat] [c?nd] [c?Nat]

to shed skin (snake) pull (it) out! (blade of grass from sheath) be gutted to gut

[wn] to pour out [wko tnin] pour out (water) for me! [wn] [whNat] be wild, untamed scare, make wild

Some nasal sufxes have allomorphs analogous to (53a) if lexical tone immediately precedes them, such as /-t/ [-nta ~ -na] neg (see 17) or /-d?/ [-nta? ~ -nda? ~ -na?] adj (see 11b; note also /-b\d?/ [-mb\inta?] piece vs. /-b\b\id?/ [-mb\imb\ida?] pieces). Some do not, such as /-nE?/ [-ne?] remote past. Oral sufxes, such as /-t?/ [-nti? ~ -ti? ~ -ndi?] down, away, up, toward, demonstrate variants typical of unmarked sufxes, except that they block nasal prosody: [-nti?] occurs following vowels bearing high tone, [-ndi?] within stems following a root-nal glottal stop, and [-ti?] elsewhere. Allomorphs such as [-nta], [-nta?], [-nte?], and [-nti?] are ambiguous with respect to indicating nasality of the lexical root to which they are afxed. Verb roots occur with these allomorphs where one cannot be sure regarding the specication of the root for nasality (54a); others occur where one can be reasonably sure that the root is nasal (see 54b): (54a) [ponte?] /p-t?/ /be.lit-dispersed/ be lit [poNote?] /p-No-t?/ /light-stem-dispersed/ to light (54b) [ginti?] /gY-ti?/ /extinguished-away/ be extinguished [gihNa?a] /gY-Na-a/ /extinguish-stem-cont / calm down a crying child

nasal prosody in chimila

35

An alternative occurs for (54b) be extinguished, in which the root is specied for transitivity (in this case, [-transitive]): [gti?] (/gY-t?/) be extinguished. In this root lexical nasality is unexpressed phonetically, because of the interaction of the specication [-transitive] and the requirement that CV syllables bearing lexical tone must be heavy. One cannot escape the conclusion that the sufx allomorphs not only signal phonological information concerning the root and root boundaries but also lexico-morphological information concerning the root itself. 6. Conclusions. The existence of nasality in the Chimila lexicon has implications with respect to morphological analysis. First, it casts considerable light on the verb morphology; in fact, it has led to a breakthrough in determining the inventory of verb sufxes and their phonetic forms, also in distinguishing between similar morphemes, for instance, /-Na/ [-Na] stem (root unmarked for transitivity), /-ka/ stem (inherently transitive root) and /-ka/ [-ka ~ -Na] out, away, or /-wa/ [-wa ~ -Nwa] mid/nomz ([0nasal]) and /-wa/ [-wa ~ -kwa] pat ([-nasal]). Second, the phenomena reported to be associated with nasality (variation in realization between voiced stops and nasals, morphophonemic alternations, the glottal off-glide, and insertion of [t] or [k]) provide dependable diagnostics for recognizing nasal roots: in all cases where one of these occurs with respect to a given verb root, one or more of the other phenomena have eventually appeared in the data.24 The proposed analysis also has theoretical implications. In order to provide a unied account of the behavior of nasality and nasal prosody, I have assumed a ternary distinction for marking lexical nasality ([+nasal], [-nasal], [0nasal]); a ternary distinction is also useful in explaining some occurrences of the glottal off-glide, as well as the insertion of [t] and [k] in verb root margins, and morphological variants of verb sufxes. The issue of ternary features is under debate and bears on current debates concerning opacity, locality, feature specication, the nature of feature geometry, the structure of the lexicon, and interactions between phonology and morphologyall crucial in dealing with harmony systems and none of which I can explore in detail here. When one looks at other languages within the Chibchan language family, the presence of nasality in the Chimila lexicon comes as no surprise. Even though no contrast between nasalized and nonnasalized vowels appears in two related languages spoken in northern Colombia, Ika (Landaburu 2000, Tracy and Tracy 1973; 1976, and Frank 1990) and Tunebo (Headland

24 At this stage of the analysis, using one or more of the other criteria for identifying nasal roots and stems would add at least 20 verb roots to the list in Appendix A; see (30b) and (54b) for examples.

36

international journal of american linguistics

and Headland 1976 and Headland 1997),25 in three other northern Colombian Chibchan languages nasalized and nonnasalized vowels contrast: Bar (Mogolln Prez 2000), Damana (Williams 1997), and Kogi (Gawthorne and Hensarling 1984).26 Constenla (1981) reports nasal vowels in contrast to oral vowels for the following Central American Chibchan languages: Cabcar (also Margary Pea 1982 and Jones 1983:154), Bribri (also Schlabach 1974), and Terraba. A contrast between oral and nasal vowels is reported by other researchers for the additional languages Bocot de Chiriqu (Gunn and Gunn 1974 and Margary Pea 1988), Guaym Movere (Kopesec and Kopesec 1974 and Abarca Gonzlez 1985), Teribe (according to Quesada 2000 and others, best considered to be the same language as Terraba) (Koontz and Anderson 1974 and Oakes 2001), and Rama (Holt 1986).27 The analysis proposed here locates Chimila with other Chibchan languages in which nasal vowels contrast with oral vowels. At the same time the behavior of nasality and nasal prosody in Chimila is unarguably unique, even though it exhibits characteristics typical of well-known phonological systems which include nasality and nasal prosody. Its uniqueness is found in the constraints on the segmental manifestation of nasality in core lexemes, and the interaction of these constraints with the prosodic system, lexical tone, syllable structure, and the morphology. APPENDIX A Lexically Nasal Verb Roots and Stems
The phonetic form is as pronounced phrase-internally. The following lexical stemforming morphemes occur with the verbs listed below: /-Na/ /-No?/ /-ka/ alongside, parallel to transitive stem away, out /-kwa/ /-o/ /-ta/ extended area transitivizer up, toward

25 Tunebo contrasts an oral bilabial glide and a nasalized bilabial glide (Headland and Headland 1976). 26 For Damana, Trillos (2000) reports /u/ in contrast with /u$ / but no contrasts for other vowels; she observes that /i,a,o,u/ nasalize contiguous to nasals and when they bear stress (2000:752). For Kogi, Ortiz Ricaurte (2000) reports no vowel contrasts but hypothesizes that nasality functions as a suprasegmental. 27 Wheeler (1972), Levinsohn (1975), and Frank (1988) do not reconstruct this contrast for the proto-language giving rise to northern Colombian Chibchan languages, although Frank observes that the proto-language most likely had nasalization as a suprasegmental feature operating on the word level (1988:18). Another investigator suggests proto-nasalized vowels as a possible motivation for some of the reexes of *m in Kogi (Jackson 1990). In his reconstruction of Proto-Chibchan, Constenla (1981) includes nasality as a suprasegmental phoneme. Although the data available to him were not nearly as reliable as that available to more recent researchers, he includes all the northern Colombian Chibchan languages mentioned above in his reconstruction and proposes possible reexes for nasalization in these languages, including Chimila.

nasal prosody in chimila

37

/-ka/ /-ke?/ /-kri/ a a-ta a-ka by by-ta by-t ?-k\i by-t ?-kwa b\a b\a-ta de de-ta d-ta $ d-ka $ do-ta do ekuna d d-ta

transitive stem aside elevated ?ah ?ana ?aNa mbih mbina m bitik\i m bitikwa
mb\ah

/-t?/ /-t ?/ /-wa/

distributed down, away from middle voice (mid)

Lexical Root or Stem

Phonetic Realization

Gloss
go return leave, go away roll into a ball cut head of grain off stem wind string in ball throw/shoot into the ground begin wake up say, mention get out of harms way to skin cut off a piece from enter swallow stretch ones self pull toward (something long) get divorced (two people) abandon abandon be lying down; be (something big) fetch water take off clothes step aside be brooding (hen) be hanging (something long) ascend vertically leak, escape (liquid) get away, escape (animate) creep creep along smoke (cigarette) eat with a stick (thick liquid) grab, catch grasp with hands be bright, lit up, blond

b\ana deh n dena ndina n diNa n dona n h do ?Ekuna ndo n dona


m

ka ga-t?-No?-ka ga-t ?-ka ga-t ?-o g()-wa gv-ka gv-ta g\v-ka gw Na y-t? ha-ta-ka ha-ka jy jy-ta j j-ta ka-ka-t ? ka-ta ka-t?

aka ganeNoka gatika N gato N ga()Nwa


N g

guNa

Nguna Ng\uNa N w g aa Naa

?ine? hanaka haNa hi hina hoh hona kakati? kana kane?

38
kNo ha-wa kvu$-ta kw-ta kw kw kw-ta kw-ta-ka kwa-ka (wNa) ly-ta ma-ta mo mo-ta-k\i na-k\i na-ta-k\i

international journal of american linguistics

koNo haNwa kuuna kwana kwah kwah kwana kwanaka kwaNa (waNa) lina maana moh monak\i naaNg\i naanak\i no ?oh ?ona \anaka \uh sah sasana sike? sinakwa suNa tona tuNwa tuna tuNwa uka una wah wana waNa wina winaka wina yah yana yuh yuna

n o o-ta \-ta-ka \u$ s sasa-ta sy-ke? sy-ta-kwa sv-Na t-ta tv-wa tv-ta tu$-wa u$-ka (root-stem) u$-ta wa wa-ta w-ka w-ta $ w-ta-ka $ wy-ta ya ya-ta yv yv-ta

fetch water, arrive at arroyo to milk approach be on, at (mass or group) to pour pour into; to bud out (tree) pour onto ground throw into stack wash clothes pass by carry hanging carry crossways on originate from; be vertical originate from (higher elevation) hear roast fry slide toward be unkempt, rufed grind pound earth rm pound on tie be blindfolded cook, boil shuck corn be alive sprout see drink drink through straw be there be abundant throw out pour out water pull out of be wild, untamed exist; live be born look for insert horizontally

nasal prosody in chimila

39

REFERENCES Abarca Gonzlez, Roco. 1985. Anlisis fonolgico del guaym movere. Estudios de Lingstica Chibcha, Serie Anual, 4:7 46. San Jos: Universidad de Costa Rica. Barnes, Janet. 1996. Autosegments with three-way lexical contrasts in Tuyuca. IJAL 62:3158. Beddor, Patrice Speeter. 1993. The perception of nasal vowels. Phonetics and Phonology 5: Nasals, Nasalization, and the Velum, ed. Marie K. Huffmann and Rena A. Krakow, pp. 171 96. New York: Academic Press. Blevins, Juliette, and Andrew Garrett. 1992. Ponapean nasal substitution: New evidence for rhinoglottophilia. Proceedings of the Berkeley Linguistics Society 18:221. Cho Yu, Young-mee. 1993. The phonology and phonetics of voiceless vowels. Proceedings of the Berkeley Linguistics Society 19:6475. Constenla, Adolfo. 1981. Comparative Chibchan phonology. Ph.D. dissertation, University of Pennsylvania. Frank, Paul. 1988. Proto-Arhuacan phonology. Ms. . 1990. Ika Syntax: Studies in the Languages of Colombia I. Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics 93. Dallas: Summer Institute of Linguistics and University of Texas at Arlington. Gawthorne, Linda, and Grace Hensarling. 1984. Fonologa del cogui. Sistemas Fonolgicos de Idiomas Colombianos 5:356. Bogot: Ministerio de Gobierno and Instituto Lingstico de Verano. Gordon, Matthew. 2002. A phonetically driven account of syllable weight. Language 78:5180. Gunn, Robert D., and Mary R. Gunn. 1974. Fonologa bocot. Lenguas de Panam, vol. 1, Sistemas Fonolgicos, ed. Patricia Baptista, pp. 31 48. Panam: Instituto Lingstico de Verano and Instituto Nacional de Cultura-Repblica de Panam. Hall, T. Alan. 2000. Phonologie: Eine Einfhrung. New York: Walter de Gruyter. Hayes, Bruce. 1995. A Metrical Theory of Stress. Chicago: University of Chicago Press. Headland, Edna. 1997. Diccionario Bilinge Uw Cuwa (Tunebo)Espaol EspaolUw Cuwa (Tunebo), con una grmatica uw cuwa (tunebo). Bogot: Instituto Lingstico de Verano. Headland, Paul, and Edna Headland. 1976. Fonologa del tunebo. Sistemas Fonolgicos de Idiomas Colombianos 3:1726. Bogot: Ministerio de Gobierno and Instituto Lingstico de Verano. Holt, Dennis. 1986. The development of the Paya sound system. Ph.D. dissertation, University of California, Los Angeles. Jackson, Robert T. 1990. Comparative phonology and grammar of the Arhuacan languages: Sierra Nevada de Santa Marta. Ms. Jones, Aziel W. 1983. El verbo cabcar: Ncleo con satlites. Amrica Indgena 43:14754. Klamer, Marian. 2002. Semantically motivated lexical patterns: A study of Dutch and Kambera expressives. Language 78:25886. Koontz, Carol, and Joanne Anderson. 1974. Fonologa teribe. Lenguas de Panam, vol. 1, Sistemas Fonolgicos, ed. Patricia Baptista, pp. 4969. Panam: Instituto Lingstico de Verano and Instituto Nacional de Cultura-Repblica de Panam. Kopesec, Michael F., and Bonnie M. Kopesec. 1974. La jerarqua fonolgica del guaym. Lenguas de Panam, vol. 1, Sistemas Fonolgicos, ed. Patricia Baptista, pp. 1730. Panam: Instituto Lingstico de Verano and Instituto Nacional de Cultural-Repblica de Panam. Landaburu, Jon. 2000. La lengua ika. Lenguas indgenas de Colombia: Una visin descriptiva, ed. Mara Stella Gonzlez de Prez and Mara Luisa Rodrgues de Montes, pp. 733 48. Santaf de Bogot, Colombia: Instituto Caro y Cuervo.

40

international journal of american linguistics

Levinsohn, Stephen H. 1975. El bokot, el guaym y el teribe, respecto al proto-chibcha. Lenguas de Panam, vol. 2, Observaciones preliminares sobre los sistemas gramaticales de las lenguas Chibchas, ed. Stephen H. Levinsohn, pp. 318. Panam: Instituto Lingstico de Verano and Instituto Nacional de Cultura-Repblica de Panam. Margary Pea, Enrique. 1982. Descripcin del sistema fonolgico de un idiolecto del cabcar de Ujarrs. Estudios de Lingstica Chibcha, Serie Anual, 1:33 44. San Jos: Universidad de Costa Rica. . 1988. La nasalizacin en el bocot de chiriqu. Estudios de Lingstica Chibcha, Serie Anual, 7:6573. San Jos: Universidad de Costa Rica. Malone, Terry. 2000. Interpreting CECIL frames: Examples from Chimila. Notes on Linguistics 3:67100. . 2001. Canonically long and geminate vowels in Chimila (Chibchan). Southwest Journal of Linguistics 20:71117. . 2004. Classiers in Chimila (Chibchan). Sprachtypologie und Universalienforschung 57:144201. . 2006. Tone and syllable structure in Chimila. IJAL 72:157. Mogolln Prez, Mara Cristina. 2000. Fonologa de la lengua bar. Lenguas indgenas de Colombia: Una visin descriptiva, ed. Mara Stella Gonzlez de Prez and Mara Luisa Rodrguez de Montes, pp. 71927. Santaf de Bogot, Colombia: Instituto Caro y Cuervo. Oakes, Perry J. 2001. A description of Teribe phonology. SIL Electronic Working Papers 20013. <www.sil.org>. Ortiz Ricaurte, Carolina. 2000. La lengua kogui: Fonologa y morfosintaxis nominal. Lenguas indgenas de Colombia: Una visin descriptiva, ed. Mara Stella Gonzlez de Prez and Mara Luisa Rodrguez de Montes, pp. 75780. Santaf de Bogot, Colombia: Instituto Caro y Cuervo. Parker, Steve. 1999. A sketch of Iapari phonology. IJAL 65:139. Piggott, G. L. 1992. Variability in feature dependency: The case of nasality. Natural Language and Linguistic Theory 10:3377. Quesada, J. Diego. 2000. Word order, participant encoding, and the alleged ergativity in Teribe. IJAL 66:98124. Schlabach, Raymond. 1974. Los fonemas del bribri del Valle de Talamanca. Amrica Indgena 34:35562. Steriade, Donca. 1994. Complex onsets as single segments: The Mazateco pattern. Perspectives in Phonology, ed. Jennifer Cole and Charles Kisseberth, pp. 20291. Stanford, Calif.: CSLI Publications. Tracy, Hubert P., and Martha Tracy. 1973. Fonemas del ica (arhuaco). Sistemas Fonolgicos de Idiomas Colombianos 2:5770. Bogot: Ministerio de Gobierno and Instituto Lingstico de Verano. . 1976. Notas sobre fonologa y ortograa ica (arhuaco). Sistemas Fonolgicos de Idiomas Colombianos 3:143 47. Bogot: Ministerio de Gobierno and Instituto Lingstico de Verano. Trillos Amaya, Mara. 1997. Categoras gramaticales del ette taara, lengua de los chimilas. Lenguas Aborgenes de Colombia: Descripciones 10. Santaf de Bogot, Colombia; Colciencias and Universidad de los Andes. . 2000. Sntesis descriptiva de los sistemas fonolgico y morfosintctico del damana. Lenguas indgenas de Colombia: Una visin descriptiva, ed. Mara Stella Gonzlez de Prez and Mara Luisa Rodrguez de Montes, pp. 74956. Santaf de Bogot, Colombia: Instituto Caro y Cuervo. Um, Hye-Young. 1996. The distribution and representation of laryngeals. Proceedings of the Berkeley Linguistics Society 22:41728.

nasal prosody in chimila

41

Walker, Rachel Lea. 1998. Nasalization, neutral segments, and opacity effects. Ph.D. dissertation, University of California, Santa Cruz. Wheeler, Alva. 1972. Proto Chibchan. Comparative Studies in Amerindian Languages, ed. Esther Matteson, pp. 93108. The Hague: Mouton. Williams, Cindy. 1997. Damana phonology. Ms. Zoll, Cheryl. 1996. Parsing below the segment in a constraint based framework. Ph.D. dissertation, University of California, Berkeley.

You might also like