Bartholomew Collection of Unpublished Materials SIL International - Mexico Branch

Language and Culture Archives

NOTICE This document is part of the archive of unpublished language data created by members of the Mexico Branch of SIL International. While it does not meet SIL standards for publication, it is shared “as is” under the Creative Commons Attribution- NonCommercial-ShareAlike license (http://creativecommons.org/licenses/by-nc- sa/4.0/) to make the content available to the language community and to researchers.

SIL International claims copyright to the analysis and presentation of the data contained in this document, but not to the authorship of the original vernacular language content.

AVISO Este documento forma parte del archivo de datos lingüísticos inéditos creados por miembros de la filial de SIL International en México. Aunque no cumple con las normas de publicación de SIL, se presenta aquí tal cual de acuerdo con la licencia "Creative Commons Atribución-NoComercial-CompartirIgual" (http://creativecommons.org/licenses/by-nc- sa/4.0/) para que esté accesible a la comunidad y a los investigadores. Los derechos reservados por SIL International abarcan el análisis y la presentación de los datos incluidos en este documento, pero no abarcan los derechos de autor del contenido original en la lengua indígena.

Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores1, 2 R. David Klint SIL International 1 Introduction

Mazatec is a Mexican language with 12-20 variants spoken in the La Cañada area of Oaxaca. Many variants show asymmetries in the laryngeally modified consonants of the phonemic inventory. Specifically, the laryngeally modified consonants in the phonemic inventory of Íénná, Mazatec of Mazatlán Villa de Flores, ISO 639-3 = vmz, mazateco del suroeste (INALI 2016), are asymmetric.

Table 1 Obstruents and resonants in Íénná

Bilabial Alveolar Postalveolar Palatal Velar Glottal Nasals m m n̥ n ɲ̊ ɲ Glottalized nasals ˀm ˀn ˀɲ Stops t d k ⁿg ʔ Aspirated stops tʰ ⁿtʰ kʰ Glottalized stops ˀⁿd Affricates ts dz tʃ dʒ Aspirated Affricates tsʰ ⁿtsʰ tʃʰ ⁿtʃʰ Glottalized affricates tsˀ tʃˀ Fricatives s ʃ h Glottalized fricatives sˀ Approximates ɸ̞ β̞ j ̊ j Glottalized Approximates ˀw ˀj

1 The abbreviations used in this paper as follows:1 = first person, 2 = second person, 3 = third person, excl = exclusive, incl = inclusive, pl = plural. 2The Ethnologue gives the code vmz and the name mazateco de Mazatlán to the language known by it's speakers as Íénná (Lewis et al. 2016). The Instituto Nacional de Lenguas Indigenas (INALI) refers to it as mazateco del suroeste, and gives the local name as ienra naxinandana nnandia (INALI 2016). It is important to note that despite the word najndiaámeaning center in Mazatec, the language which is named mazateco del centro is not spoken in Najndia̱á, the Mazatec name for Mazatlán Villa de Flores. According to INALI, mazateco del centro is spoken in the municipalities of Huautepec, Huautla and Santa María Asunción, which corresponds to the variant of Huautla de Jiménez, which has the ISO-63 code mau Lewis et al. (2016).

1 2 Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores The Íénná phonemic inventory includes post-aspirated obstruents, [tʰ, ⁿtʰ, kʰ], voiceless resonants [m̥, n̥, ɲ̊, ɸ̞ , β̞ ], post-glottalized obstruents [tsˀ , tʃˀ], and pre-glottalized resonants [ˀw , ˀj ]. It includes neither pre-glottalized obstruents such as [ˀt, ˀts], nor post-glottalized resonants such as [mˀ, nˀ, ɲˀ]. Ascribing non-modal voicing to a morpheme feature removes the otherwise unmotivated asymmetry in the phonemic inventory of Íénná and provides evidence that non-modal voicing as a morphemic feature may be an areal phenomenon of Mesoamerican languages, e.g. [+nasal] and [+CG] in Mixtec (Mar- lett 1992, Macaulay & Salmons 1995), [+CG] in Totonac (Aschmann & Wonderly 1952), Trique (Hollenbach 1984) and Tepehua (Watters 1987), and [+nasal] in Me'phaa (Marlett y Neri 2012).

Table 2 Morphemic structure

Glottalized stem Aspirated stem /síG/ [sˀi] party /tíH/ [tʰí] round /tʃáG/ [tʃˀá] carry /kāH/ [kʰā] iteration /tsūG/ [tsˀū] huipil dress /tsáH/ [tsʰá] happy /nījāG/ [nīˀjā] house /β̞īH/ [ɸ̞ī] go /tīβ̞āG/ [tīˀβ̞ā] he is carrying /jàH/ [j ̊à] eagle /nāmíG/ [nāˀmí] priest /nūH/ [n̥ū] night

A morphemic feature analysis of non-modal voicing avoids the complex and typologically unusual syllable structure required to analyze it as independent segments (Pike & Pike 1947), the intricate featural system required to analyze it as a secondary feature (Steri- ade 1994), and avoids the redundancy of postulating breathy and laryngealized features on both vowels and onsets Golston & Kehrein (1998). 2 Previous analyses

Previous analyses, accounting for the assymetrical pattern of non-modal voicing in a closely related Mazatec variety (mau), have included phonemic segments (Pike & Pike (1947), García et. al. (2014)), vowel features (Golston & Kehrein (1998)), and secondary features (Steriade (1994), García et. al. (2014)). The analysis of non-modal voicing as a morpheme feature accounts for asymmetry in the phonemic inventory without requiring non-modal secondary features or typologically rare syllable structures.

Table 3 Previous analyses of laryngeal contrasts in Huautla Mazatec

gloss Pike & Steriade Golston & García et. Pike (1994) Kehrein al. (2014) (1947)3 (1998) Pre-laryngealization rainbow ʔja4 ˀja4 ˀja4 ˀja4 3We adapt Pike and Pike's data into the API for. Additionally, we mark tone with 1 being the lowest tone and 4 being the highest tone, as opposed to the their data which has 4 as the lowest and 1 as the highest. Aspiration 3 gloss Pike & Steriade Golston & García et. Pike (1994) Kehrein al. (2014) (1947)3 (1998) Post- he jʔa³ jˀa³ ja̰³ ja̰³ laryngealization carries pre-aspiration remains hka³4 ʰka34 k̤a³34 hka³4 post-aspiration smelly khã³ kʰã³ kã̤³ kʰã³

In a morpheme feature analysis, stems rather than segments are marked with the non- modal features extended glottis [+EG] or constricted glottis [+CG]. The non-modal voicing is realized at the leftmost edge of the resonant on the last syllable of the stem. 3 Aspiration

Aspiration, in Íénná, can occur by itself in onset position, or it can precede or proceed another consonant, which we describe as pre-aspiration or post-aspiration respectively. The distribution of aspiration on onsets with resonants is distinct from aspiration with obstruents.

Table 4 Onsets involving aspiration in Íénná

Pre-aspirated Post-aspirated /h/ hó two /t/ htī itch thī round /ⁿd/ n̥djú whirlpool ⁿthò̥ wind /k/ hku his head khě different /ⁿk/ n̥gū one — /kʷ/ hkʷjé rough kʷḁ̀ issue/problem /ⁿgʷ/ — — /ts/ htsjē4 guava tshá possum /ⁿd͡z/ n̥djà too salty ⁿtshḁ̄ his hand /t͡ʃ/ —5 tʃho̓: egg ̥ /ⁿd͡ʒ/ n̥dʒa exists ⁿtʃhḁ̄ he speaks /n/ n̥nú owl — /m/ m̥má black — /ɲ/ ɲ̊ɲú night — /β̞/ ɸ̞ī goes — /j/ j ̊ē, hje fat —

4For many speakers, the pre-aspiration has assimilated to the place of the affricate, so the /hts/ sequence is pronounced /sts/. 5There are no recorded words of the sequence /htʃ/. It is probable that the same process of assimilation underway in the sequence /hts/ to /sts/ has also assimilated /htʃ/ to /ʃtʃ/. 4 Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores

6For the majority of speakers, the semivowel /j/ is pre-aspirated in an aspirated morpheme, and the glottis is not extended during the full production of the resonant, [hj], but in Íénná spoken in Barrio Guadalupe, the glottis is extended throughout the production of the resonant, [j ̊]. In figures i, ii and iii we can see the difference in frication and resonance between the glottal fricative in /hàmà/ [hàmà] root, the palatal semivowel /já/ [já] tree and the voiceless semivowel /jã̄H/ [j ̊ã] three.

Figure i Spectrogram of /hàmà/ root Glottalization 5 García et. al. (2014) in their phonetic study of the Mazatec spoken in Río Santiago give acoustic evidence that the pre-aspiration of stops should be considered the full segment /h/.

The syllable structure of Íénná, which accepts as its maximal syllable C1C2VV, where C1 is a

fricative and C2 is a stop or affricate, also supports their claim. While the syllable structure of the Río Santiago variant is slightly different, masking this trend, in Íénná, pre-aspiration of stops patterns along with the other fricatives and can precede a stop or affricate, but no other consonant. The pre-aspiration of resonants, on the other hand, coincides with the articulation of a portion of the segment in a way articulatorily impossible with obstruents. On nasals, the first portion on the segment is pronounced voicelessly, and voicing begins during the production of the nasal stop. On semivowels, /β̞/ is completely devoiced resulting in [ɸ̞], and /j/ is either partially or totally devoiced to [hj] or [j] ̊ 6 depending on the town.7

Table 5 Aspiration on different syllable types, where R is resonant and O is obstruent

/RV/H → [R̥V] or [ʰRV] /OV/H → [OʰV]

If the pre-aspiration of obstruents is analyzed as the segment /h/, a pattern of post-aspiration of obstruents and devoicing/preaspiration of resonants emerges. I propose that it is the morpheme itself which is specified as [+EG] and that the aspiration will be realized at the left edge of the resonant portion of the syllable. When the onset of the syllable has an obstruent, the aspiration is realized at the right edge of the onset, and when the onset is a resonant, the aspiration is realized at the left edge of the onset. Following Marlett's (1992) convention of marking morphemic nasalization with a superscript N,N , I mark morphemic aspiration with a superscript H, H, and morphemic glottalization with a superscript G, G. The creation of voiceless resonants through overlap of morphemic aspiration with the resonant onset gives rise to a crosslinguistically rare class of voiceless resonants, with /j ̊/ in 2.2% , /ʍ/, en 3.5%, of the languages of the world Maddieson (1984). The robust class of these phones in Íénná merits further investigation. 4 Glottalization

The distribution of glottalization in the onsets of Íénná is more complicated than that of aspiration, both due to a contrast system in decline, when compared to its sister variants, as well as due to separate phonotactic interactions which obscure the pattern. However, a basic pattern of post-glottalization of obstruents and preglottalization of resonants is quite clear. As seen in 6, many consonants do not interact with glottalization at all, and preglottalization of /ⁿd/ in words such as ʔⁿdjē hears does not follow the pattern of preglottalization of other obstruents. This pattern is distinct from the Huautla/Río Santiago and Jalapa patterns which include pre- and post- glottalization of both obstruents and resonants.

Figure ii Spectrogram of /já/ tree

Figure iii Spectrogram of de /jã̄H/ [j ̊ã]́ three

We see here the difference between the three phones [h], [j] and [j ̊]: the [h] has a larger amount of white noise, the diffusion of sound throughout all frequencies, visible in the white space in the spectrogram; the [j] has a larger mount of resonance, visible in the dark bands of formant frequencies; and the [j ̊] is more resonant

than the [h], but less resonant than the [j]. Nevertheless, the acoustic energy clearly gathers around the F2,F3,

and F4 formants, showing the acoustic correlate of the place of the voiceless approximate. 7Kirk (1966) in his reconstruction of Proto-mazatec, proposes /*ʃʰ/ as the proto-phoneme for this cognate in Íénná. In his crosslinguistic study of affricated fricatives, however, Guillaume Jacques (2011):10 proposes 6 Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores Table 6 Onsets involving glottalization in Íénná

Pre-glottalized Post-glottalized /ʔ/ ʔō five /t/ — kētˀāsú legó a pasar /ⁿd/ ˀⁿdjē hears — /k/ — kˀjẽ̄ die /kʷ/ — kʷí̤ will be named /ⁿg/ — ⁿgˀjá shadow /ts/ — tsˀū huipil /ⁿdz/ — — /t͡ʃ/ — tʃˀá carries /ⁿd͡ʒ/ — — /s/ — sˀí party /ʃ/ — — /h/ — — /n/ nāˀní mala mujer plant — /m/ ˀmi is called — /ɲ/ ˀɲú forcefully — /β̞/ tīˀβā he is carrying — /j/ nīˀjá house —

I show loss of glottalization in a brief comparison of speakers whose linguistic conser- vatism correlates roughly to age.

Table 7 Glottalization in four speakers

CGZ ACM ACF AMX man man man woman 54 years old 54 years old 30 years old 33 years old /tsu²G/ huipil dress [tsˀu²] [tsˀu²] — [tsu²] /tsje²G/ lazy [tsˀje²] [tsje²] [tsje²] [tsje²] /siG/party [sˀi³] [sˀi³] [si³] [si³] /namíG/ priest [na²ˀmi³][na²mi³][na²mi³][na²mi³] /naniG/ mala mujer plant [na²ˀni³][na²ˀni³][na²ni³][na²ni³] /namiG/ house [ni²ˀja³][ni²ˀja³][ni²ja³] [ni²ja³] /βaG/ carries [ti²ˀβ̞a²][ti²β̞a²][ti²β̞a²][ti²β̞a²]

The preglottalization of the alveolar prenasalized stop [ʔnd] can not be explained by loss of glottalization. The solution to this irregularity is due to a pattern of denasalization that that the proto-phoneme should be /j ̊/, which is indeed the Íénná pronunciation of every word from Kirk's list, /j ̊ã¹/ 'eagle', /j ̊ã²/'three', /j ̊õ/'six', /j ̊e¹/'young corn fruit/jilote', /j ̊e²/'fat', y /j ̊e³/'sin' but not in the word /hĩ/ 'eight', which is pronounced as a glottal fricative and not a voiceless semivowel. Allomorphic evidence 7 has been noted in several Mazatec variants. The flap [ɾ] only appears in postonic position before non-nasal vowels, in native Mazatec words (Golston & Kehrein 1998, Beal 2011, Filio 2014). Williams (2004) proposes that this is due to denasalization of the nasal stop /n/ before oral vowels in postonic position, and further argues that some surface level instances of [nd] in pretonic position are underlyingly /n/ and are being denasalized, or rather, post- oralized to [nd] before oral vowels. With this in mind, the unexpected preglottalization of [nd] is understood as the preglottalization not of /ⁿd/, but of /n/. The verb [ʔndjē] hears is underlyingly /njēG/. The morphemic glottalization is assigned to the left edge of the resonant portion of the vowel and the phonetic realization of the /n/ is [nd].

Table 8 Preglotallization of [nd]

Glottalization applied at left edge of resonant portion /niē/G → [ʔndjē] /n/ post-oralized before oral vowel

5 Allomorphic evidence

There is evidence from morphemic alternation that aspiration and glottalization are assigned at the morphemic rather than segmental level as well. Here we give evidence from three distinct morphemic processes: tense stem alternation, possession, and imperative.

Table 9 Tense paradigm for [ˀmi] is called

Prefix Stem Surface form Gloss β̞íGN [ˀmí] is called k- β̞íGN [kw̤i] will be called

A morphemic analysis of aspiration also helps make the inalienable possession paradigm more transparent.

Table 10 Paradigm for [ⁿtsʰā] hand

ⁿdza᷅ my hand ⁿdze your hand ⁿt͡ʃʰḁ his/her/their hand ⁿd͡zǎ: our hand inclusive ⁿd͡zahì our hand exclusive ⁿd͡zǒ: your hand plural 8 Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores All members of the paradigm for the word hand begin with a prenasalized affricate, and the third person has a voiceless prenasalized portion and an aspirated release. The tone contours and the shape of the final vowels can be analyzed as the coalescence of the person enclitics -ā 1,-hī 2,-́ incl,-hì excl,-hú 2pl, but the aspiration of the third person is not so clear. However, just as stems are specified as [+EG] rather than a particular segment in the stem, the third person possessive morpheme in this paradigm is a autosegmental [+EG] that has no segmental value. The morphology of the third person possessive form would be /ⁿd͡zā/ hand + /H/3 = [ⁿt͡sʰā]. Another morpheme which contains no segmental content is the imperative (in some verbal paradigms), which consists of a low tone and the feature [+EG].

Table 11 Mood paradigm [nikét͡sjēhī] you think

Mood Root Surface form Gloss ∅- indicative nikét͡sjēhī [nīkét͡sjēhī] you think H- imperative nikét͡sjēhī [nìkʰét͡sjēhī] think! káta- optative nikét͡sjēhī [kátānīkét͡sjēhī] may you think

6 Evidence from other variants

This morphemic analysis of [+EG] and [+CG] is a useful way to look at the variation of non-modal voicing in other Mazatec variants as well. As Macaulay & Salmons (1995) show in their study of glottalization in Mixtec, interrupted vowels and vowels ending in glottal stops can be seen to be the result of a phonological rule to "associate the feature [+CG] to the timing slot corresponding to the leftmost vowel of the couplet." In the context of Mazatec, we can look at the variation in the position of the glottal stop in different variants as differences in the realization of [+CG] when applied to the morpheme. For example, García et. al. (2014) describe in the Mazatec variant spoken in Río Santiago both post- glottalization and preglottalization of resonants, a pattern which is not possible in Mazatlán. As examples of post-glottalized resonants, they give mʔe he dies and jʔa he carries.A comparison of cognates of these two words, drawn from Kirk (1966), shows variation.

Table 12 Cognates of Glottalized resonants (Kirk 1966)

Proto-mazatec *ʔme² die/be sick *ja carries Ayautla tiʔme¹ he's dying jaʔá carries Jalapa de Diaz ti4mʔe:³ he's dying jʔá carries Mazatlán Villa de Flores ka³βe²ja² murió ja² carries Río Santiago mʔe² muere jʔa³ carries Santo Domingo tiʔmé he's sick jʔa³ carries Soyaltepec khi²ʔme:² he's sick/dying jaʔá, jʔá³ carries San Lorenzo Cuauntecuitla joʔó carries Chiquihuitlán kaʔmé he died San Pedro Ixcatlán kiʔmẽ́ Evidence from other variants 9 The majority of variants pre-glottalize the nasal /m/, with Jalapa de Diaz and Río Santiago being the only variants to post-glottalize. This variation is accounted for by postulating two distinct rules for application of [+CG] to the root:

Resulting in pre-glottalization: Associate the feature [+CG] to the leftmost edge of the resonant portion of the root.

Resulting in post-glottalization: Associate the feature [+CG] to the leftmost edge of the nucleus of the root. .

These rules, however, do not accurately predict the position of the glottal stop on roots with a /j/ onset, as seen in 12. The approximate is either post-glottalized or realized in the middle of the vowel, and in no case is it pre-glottalized. There are no doubt differing phonotactic constraints on the realization of [+CG] and a more robust cross-variant analysis is no doubt necessary to discover the specific interactions. Further evidence of [+EG] associated with the root can be seen in the verb classes de- scribed by C. Jamieson (1996:290), whose preverbs alternate depending on the stem used. Table 13 shows a selection of verb classes relevant to the discussion at hand.

Table 13 Preverb paradigm C. Jamieson (1996)

1st person singular 2nd person Verb and 3rd person and 1st person plural class present future present future 1. βe- kwe- βe- kwe- 2. βa- kwa- βa- kwa- 3. βo- sko- tʃo- tʃo- 4. βu- sku- tʃu- tʃu- 7. ɸa- khwa- ɸa- khwa-

As seen in section5 in Íénná, the future tense morpheme can be understood as /k-/ and when added to the verb stems, the /β/ is resyllabified as part of the nucleus and pronounced as the more open semivowel /w/. The preverbs for future tense for class 7 have a post- aspirated /k/ which corresponded to the voiceless bilabial fricative /ɸ/. The only difference between class 1 verbs and class 7 verbs is the feature [EG]. Class 1 verbs are specified [-EG] and class 7 verbs are specified [+EG]. When [+EG] is applied to the present preverb /βa-/, the glottis is extending during the resonant portion of the morpheme resulting in [ɸa-] and when the future morpheme /k-/ is added to make the preverb /kβa-/, the /β/ is resyllabified as a semivowel and [+EG] is realized once again at the leftmost portion of the preverb, [khwa-]. 10 Non-modal voicing as morphemic features in Íénná, Mazatec of Mazatlán Villa de Flores 7 Conclusion

The non-modal voicing features [+EG] and [+CG] are applied at the leftmost portion of the resonant segment of the final syllable in the morpheme in Íénná. Resonant onsets are realized as voiceless or pre-aspirated, and obstruent onsets are realized as post-aspirated. This pattern is most easily captured by postulating that glottal features are assigned at the morphemic level rather than the segmental level. This analysis also simplifies the seemingly allomorphic alternation of preverbs seen in Mazatec, making the future tense a composition of the future morpheme /k-/ plus the preverb /βV-/. The alternation of [ɸa-] ~ [kʰwa] is the addition of future /k-/ to the preverb /βaH/ where the preverb is aspirated at the morpholog- ical level. This morphemic feature analysis will no doubt prove useful in a generalization of the notoriously complicated verb paradigms of Mazatec.

Referencias

Aschmann, Herman and William L. Wonderly. 1952. Affixes and implicit categories in Totonac Verb Inflection. International Journal of American Linguistics 18:130-145. Beal, Heather D. 2011. The Segments and Tones of Soyaltepec Mazatec. Ph.D. dissertation. (Arlington). The University of Texas at Arlington. Filio García, Israel. 2014. Esctructura Fónica y acercamiento gramatical del mazateco De Nàjndiàá. Tésis de maestría. (Distrito Federal). Centro de investigaciones y estudios superiores en antropología social. García García, Eloy, Mario E. Chávez Peón, y Gilles Polian. 2014. Contrastes laríngeos en el mazateco de Río Santiago (Huautla). Paper presented at the Coloquio sobre lenguas otomangues y vecinas 6, Oaxaca, México. Golstein, Chris y Wolfgang Kehrein. 1998. Mazatec Onsets and Nuclei. International Journal of American Linguistics Vol. 64, No. 4:311-337. Hollenbach, Barbara E. 1984. The phonologly and morphology of tone and laryngeals in Copala Trique. Ph.D. dissertation. University of Arizona. Instituto nacional de lenguas indigenas. 2016. Catálogo de las lenguas indígenas nacionales: Variantes lingüísticas de México con sus autodenominaciones y referencias geoestadísticas. (http://www.inali.gob.mx/clin-inali/html/v_mazateco.html). (accessed 6 de marzo, 2016). Jacques, Guillaume. 2011. A panchronic study of aspirated fricatives, with new evidence from Pumi. Lingua 121 (9):1518-1538. Jamieson, Carole Ann. 1996. Diccionario de Mazateco de Chiquihuitlán, Oaxaca. Tucson, AZ: Instituto Lingüístico de Verano. Kirk, Paul L. 1966. Proto-Mazatec Phonology. Ph.D. dissertation. (Seattle, WA). University of Washington. Lewis, M. Paul, Gary F. Simons, y Charles D. Fennig. 2016. Ethnologue: Languages of the World, Nineteenth edition. (http://www.ethnologue.com). (accessed 6 de marzo, 2016). Maddieson, Ian. 1984. Patterns of sounds. Cambridge: Cambridge University Press. Macaulay, Monica y Joseph C. Salmons. 1995. The Phonology of Glottalization in Mixtec. International Journal of American Linguistics volume:38-61. Marlett, Stephen A. 1992. Nasalization in Mixtec languages. International Journal of American Linguistics 58:425-435. Marlett, Stephen A. & Plácido Neri Remigio. 2012. Me'paa de Acatepec. (http://lengamer.org/publicaciones/trabajos/me'paa_de_Acatepec_afi.pdf). (accessed 24/01/2017). Pike, Kenneth L. y Eunice Pike. 1947. Immediate constituents of Mazatec syllables. International Journal of American Linguistics 13:78-91. Steriade, Donca. 1994. Complex onsets as single segments: the Mazateco pattern. In J. Cole and C. Kissebert, eds. Perspectives in Phonology. collection_volume:203-291. Stanford, California: Center for the Study of Language and Information. Referencias 11

Watters, James. 1987. Underspecification, Multiple Tiers and Tepehua Phonology. In eds. Anna Bosch, Barbara Need, and Eric Schiller, eds. Papers from the 23rd Annual Regional Meeting of the Chicago Linguistic Society: Part Two: Parassession on Autosegmental and Metrical Phonology 23:388-402. Chicago: Chicago Linguistic Society. Williams, Cindy. 2004. Post stop nasalization in Ayautla Mazatec. Paper presented at the Conference on Oaxacan and Oto-Manguean Languages-1, University of California at Berkeley.