Overview
Coatzospan Mixtec is best known in the phonological literature for the regressive nasal harmony system by which it marks the second person familiar, a problem which has given rise to a number of analyses that have focused on how to best characterize the blocking effect of some but not all consonants (cf. Poser 1980, Cole 1987, Trigo 1988, Piggott 1989,1992, Gerfen 1992, 1994, Homer 1995). In addition to morphological harmony, however, CM contains two other sources of vowel nasalization: 1) there are contrastively nasal vowels, and 2) vowels are predictably nasalized following nasal consonants. To this point, no one has offered a comprehensive account of the entire range of nasalization facts in CM.
This chapter seeks to remedy the situation. As with glottalization in Chapter 3, I provide a detailed consideration of both the phonetics and phonology of nasalization in the language. In so doing, my goals are threefold. First, I provide phonetic data in the form of nasal airflow traces in order to offer for the first time a degree of descriptive precision unavailable to impressionistic transcription. Secondly, these data serve as the concrete underpinnings of a single, unified Optimality Theoretic (Prince and Smolensky 1993, McCarthy and Prince 1993b) phonological analysis of the range of CM nasalization facts. Finally, the data inform the discussion of the phonetics/phonology interface, especially as it regards the relationship between physiological data and phonological feature specification.
The chapter is organized in the following manner. In §4.1, I provide a basic description of the nasalization facts, presenting data sets and illustrative nasal flow data. §4.2 offers a phonological analysis in which the three varieties of nasalization are accounted for within the context of a partially ranked set of Optimality Theoretic constraints. In §4.3, I discuss the implications of CM nasalization for the interface of phonetics and phonology. Specifically, I discuss the role of phonetic grounding (Archangeli and Pulleyblank 1994a) in the phonological analysis, and, as I have noted, the relevance of the CM flow data to claims regarding the relationship between phonetic implementation and phonological feature specification (cf. Keating 1990b, Huffman 1989, 1993, Cohn 1990, 1993a). In §4.5, I present my conclusions.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
The nasalization of the initial vowels in data in (3) can be attributed to the perseveration of nasality from the preceding nasal consonant.
Note that there is one token for S1 in which the velum is lowered significantly before the onset of the nasal consonant.
Recall from Chapter 3 that glottal stop does not constitute a segment in CM. Rather, glottalization, when present, is a property of the initial vowel of a couplet. There is thus no issue of glottal stop “transparency” here.
Poser (1980) was the first to observe that within an autosegmental framework, there was no need to establish a separate set of contrastive nasal vowels, given that the distribution of lexically nasal vowels is predictable and can be accounted for by the postulation of a lexically floating nasal autosegment that spreads from right to left in the word.
Note that all of the forms in (11) contain the voiced interdental fricative [8]. Neither the voiced fricative [ß] nor prenasalized stops appear in medial position in couplets with contrastively nasalized vowels, though they do surface medially in the context of second person familiar nasalization (see figure 17, this chapter). This issue is addressed in §4.2.2 below.
The increase in flow at the onset of the fricative can be attributed to a combination of sustained velum lowering during the transition from vocalic to fricative stricture. That is, as the stricture is formed for the fricative, less air will escape through the oral channel of the vocal tract. Again, see §4.3 for discussion of phonetic implementation.
As I note in Chapter 1, since flow drops rapidly at the release of oral closure for all stops from what appears to be a consequence of the experimental setup, the distinction between the oral release of prenasals flanked by nasal vowels and the nasal release of full nasal stops in the same context is hard to discern from the flow traces. In forms such as (19), however, it is clearly audible.
Specific proposals within a rule-based framework include Poser (1980), Cole (1987), Trigo (1988), Piggott (1989, 1992), Gerfen ( 1992, 1994 ).
Pending further discussion of phonetic implementation in §4.3 below, I assume that intervening voiced fricatives are not phonological targets for nasalization.
Note that the notion of edgemost mora is itself derivable from precedence relations (see Archangeli and Pulleyblank 1994a ). For example, we can define the leftmost mora of a root as a g that is 1) associated to a vowel in the root and 2) not preceded by any other g’ in the root. Likewise, we can define the rightmost mora as a mora in the root that is not followed by another mora in the root. For convenience, I will simply refer to rightmost and leftmost.
Another means of looking at this issue is to say that satisfaction of moraic alignment is mediated by the head or most sonorous element of the mora. In CM, in which there are no codas, this will always be a vowel.
See also Myers (1991) for a general discussion of the role of constraints, and Calabrese (1995) for an alternative view from the one taken here.
As Diana Archangeli has reminded me, another type of FCC-based opacity is one in which a feature [F] targets only anchors specified for a feature [G]. In this case, [F] may spread to an adjacent anchor that is associated to [G] and subsequent spread will be licit only if the next adjacent anchor is specified for [G]. Important for our urposes is that FCC-based approaches halt spreading not by banning crossed association lines, but by feature cooccurrence constraints, together with a ban on the skipping of possible targets.
For a general critical perspective on the NCC, see Coleman and Local (1991).
Cohn (1990, 1993a) offers phonetic arguments in favor of [-nasal]. I return to this issue in §4.3 below. See also Trigo (1993) for a kind of middle ground position in which [nasal] is sometimes binary and sometimes equipollent.
Even in pre-OT terms, there is no evidence that the specification of [-nasal] for opaque consonants derives from any principled theory of specification (cf. Kiparsky 1982, Archangeli 1984, 1988, Steriade 1987, Clements 1988, Archangeli and Pulleyblank 1994a ). The best motivation appears to come from Contrastive Underspecification (Steriade 1987), with voiceless stops specified as [-nasal] in order to encode the contrast with nasal stops. However, this fails to account for the opacity
of /f/ in forms such as ndi?’fé ‘shoe’. Since nasality is non-contrastive for fricatives, there is no motivation for specifying /f/ as [-nasal]; the only motivation is that /f/ blocks harmony. See Steriade (1995) for a critique of the logic of Contrastive Underspecification and Mohanan (1991) for arguments against Radical Underspecification.
See also Smolensky (1993) and Suzuki (1995) for sequential constraints.
See House and Stevens (1956), Fujimura and Linqvist (1971), Maeda (1982, 1993) for the acoustics of nasal vowels. See also Fujimura (1962), Fant (1970), Kurowski and Blumstein (1993) on nasal consonants.
Departing from the rule-based analysis deriving opacity from the voicing status the opaque consonants of Gerfen (1992, 1994), Homer (1995) treats opacity in terms of nasality and voicelessness within the Optimal Domains model of Cole and Kisseberth ( 1995 ). My proposal is distinct in that I do not adopt Cole and Kisseberth’s specialized theoretical machinery of domain construction, nor do I assume, as does Homer, that regressive vowel harmony targets the intervening consonants themselves. Additionally, Homer’s analysis does not attempt to account for the entire range of nasalization facts in the language.
I have found two exceptions: puitii ‘sand’, which may reflect the fusion of the
morpheme púú meaning ‘earth’ and some other morpheme, and m“uinde ’prickly pear’.
See Archangeli and Pulleyblank (1994a) for extensive discussion of rule parameters within an autosegmental framework. Note that iterativity would not be motivated for forms such as pii ‘hat’, under the assumption that these contain a single vowel root associated to two moras.
For more on the behavior of prenasalized stops see §4.2.2.2 below.
More broadly, we might profitably think of the phonologization of coarticulation as falling under the rubric of constraint families such as Persevere and Anticipate for contextual assimilation and *Persevere and *Anticipate for forced distinctiveness along a phonetic dimension.
I return to the issue of the possible motivation for this constraint in the grammar in §4.3.
Alternatively, we might assume that there is a single token of [nasal] in (a) whose association skips the initial vowel. As I discuss in §4.2.2.4 below, I assume that such a configuration would incur a fatal violation of *GAP.
I have found two exceptions: pi?t31 ‘sand’, which may reflect the fusion of the morpheme púú meaning ’earth’ and some other morpheme, and miiïnde ’prickly pear’.
I also assume an inviolate OCP for [nasal], thus barring the possibility of avoiding gapping by introducing multiple tokens of the OCP. I return in more detail to this issue below in the discussion of prenasalized stops.
Note that my choice of naja as the input form is arbitrary. I make no assumptions about the value of [nasal] for the input vowels here. What is important is that the form contain an initial In/ and a non-nasal medial consonant in order to illustrate the class of surface forms that our constraints will not generate.
Note that this analysis would also require a constraint banning the association of [nasal] to continuants (cf. Cohn 1989, 1993b) in order to rule out the association of [nasal] to the medial fricative and the final vowel, thus circumventing the gapping problem.
This is the position taken by Iverson and Salmons (1996) for prenasalization in the Mixtec languages in general.
I have found two exceptions: ndindl ‘honey’ and ndi?’fe’shoe’, though the former
appears to derive from the fusion of ndu?Ute ‘water’ and ßïjï ’sweet’. The latter may also be diachronically derived from two roots. The CM form for ’foot’ is Je?e, though
the lack of nasalization in this form leaves the presence of the nasal vowel in ndi?’Jë ‘shoe’ unexplained.
As I note above, there is one exception to this generalization: mi?lnde ‘prickly pear’.
That such configurations should constitute gapped structures is not surprising. As Steriade (1993) points out, there are no known cases of rightward nasal harmony triggered by prenasalized stops.
See Archangeli and Pulleyblank (1994a) for a distinct view in which gaps in domains of association comprising both root and prosodic (part ii above) anchors are calculated separately for each type of anchor.
In fact, this definition is probably too restrictive, given that any vowel-to-vowel harmony process that is triggered by a consonant will incur a *GAP violation. Within the context of OT, however, such a situation is analyzable in terms of a violable notion of gapping, under pressure from other, more highly ranked constraints.
The formulation and status of the OCP is subject to debate. See, for example, Archangeli and Pulleyblank (1994a), who provide a more complex calculation based on tier and anchor adjacency. See also Odden (1988) for arguments against the universality of the OCP and Pierrehumbert (1992) for arguments that calculation of the OCP is gradient and based on global notions of similarity rather than simple tier adjacency.
Other possible candidates are *nâßâ and nâtâ in which both AL-R and *GAP can be satisfied via the association of [nasal] to the medial consonant. I assume that such forms are ruled out by phonetically motivated constraints banning the phonological association of [nasal] to continuant consonants (cf. Cohn 1989, 1993b) and voiceless segments (see Pulleyblank 1989, Gerfen 1993, 1994 ). To simplify the discussion above, I ignore such candidates, though I return to these issues in the discussion of phonetic implementation below.
This is subject to some exceptions. For example, within a sentence, an oral vowel might follow a nasal vowel if a speaker makes a clear pause between a word ending in a nasal vowel and a following word is V-initial.
Given the high ranking of *NV in the hierarchy, I no longer assume the ranking AL-R » *vN is motivated by forms such as núni ‘corn’ (see figure 43). Rather, the nasalization of V1 in such forms can be attributed both the need to satisfy *GAP/OCP and *NV. AL-R can thus be freely demoted in the hierarchy.
Note that I assume that the loss of [nasal] for a prenasalized stop will result in a surface [t] in candidates (d-e). This is because the language contains no plain voiced stops. There is a tacit assumption here that voiced stops must be nasal, which could be encoded as a constraint on the representation of stop voicing. This issue is tangential to the point at hand, since both surface forms are more optimally paired with respectively identical inputs than with [ndaa] as in the tableau. Note also that under the assumption that [aal is a single vowel root, only one violation of Ident[nasal] is assessed in candidate (a).
This idea is similar to ideas advanced in Piggott (1992). Note that for CM, the bilabial sonorant “stop” would not be /mb/ but the fricative /13/.
I thank Joseph Salmons for discussion of this point.
In making claims for all of the Mixtec languages, it is not clear how well any particular variety of Mixtec is characterized when Marlett’s account is subjected to close scrutiny.
Note that (74c) and (75c) would also violate the even more highly ranked conditional constraint L - R.
This provides further evidence that the scope of AL-R and AL-L is root nasality, as argued in the discussion of *NV above.
Alternatively, we might explore the hypothesis that *vN obtains only of root nasal specifications; either approach, however, must explicitly recognize the distinction between a root [nasal] specification and [nasal] corresponding to the 2FAM morpheme. I return to this issue in more detail below regarding the status of *NC.
By contrast to McCarthy (1979) and Cole (1987), however, I assume within the context of an Optimality Theoretic grammar that there is no derivational ordering between association and subsequent plane conflation. Rather, I assume that (85) is the representation of the output candidate.
As McCarthy (1996) notes in pursuing this idea in Rotuman, it is reasonable on independent grounds to assume that features are subject to identity relations independent of their segments, since faithfulness to features is a viable means of understanding stability effects under segmental deletion. McCarthy also cites Lombardi (1995) for arguments in favor of the necessity of direct feature correspondence. See also Orgun (1995) on featural faithfulness in correspondence theory.
As I note above, this case is interesting in that the voiceless consonant will arrest the leftward extension of 2-FAM nasalization, while at the same time we cannot attribute the presence of the root [nasal] vowel to the inviolate nature of *NV.
Recalling the discussion of transparency in prenasalized stops above, I assume that the nasalization of V1 in (b) results from the association of the 2-FAM morpheme. Association of the lexical [nasal] would constitute a fatal violation of the ranking *NC » AL-Left for roots.
Note that there is an alternative view of the optimal candidate in (a), in which the 2-FAM token of [nasal] is associated to both the final vowel of the verb and to the final vowel of the adverb. Given the constraints, I assume that such a form would be ruled out, as it would incur two violations of *NC, one corresponding to the *NC triggered by lexical nasalization, and the other triggered by the association of 2FAM [nasal] to the final vowel of the verb.
Though I have focused on the issue of 2-FAM nasalization and *NC, note that Max[nas] must hold in ’ + C sequences independently of 2-FAM constructions, i.e. in syntactic sequences comprised of input strings of roots.
But see §4.3.2.1 for evidence that voiceless fricatives can be produced with velum lowering.
We might also take the data from S3 to indicate that this speaker’s phonetic implementation involves the construction of a narrow window for a raised velum in the release phase of the stop, thus forcing a later onset of nasalization.
Note that Sts fifth token of this form appears to be a speech error. That is, the form exhibits a pattern of nasal implementation that corresponds to the non-2-FAM form kunú ‘to run’. This form can be found in the Appendix.
An interesting perceptual follow up study would involve delaying the onset of nasal acoustic cues to gauge the point at which speakers cease to perceive CM vowels as nasal. At the same time it is intriguing to note Ohala and Ohala’s (1993) discussion of the fact that high airflow segments such as aspirated stops and voiceless fricatives—sounds realized with a greater than normal glottal opening—give rise both synchronically and historically to spontaneous nasalization. This is attributed to the spreading or overlap of the wider glottal position of the consonant and the beginning of the following vowel, which nevertheless remains completely voiced. This overlap creates acoustic effects which mimic those of nasalization, thus triggering the percept of a nasal vowel that is physiologically realized with a raised velum. That is, although velum lowering is articulatorily antagonistic to the production of fricatives, the glottal configuration of fricatives can generate the percept of nasalization in a following vowel. Perhaps the anticipation of velum lowering here further enhances the nasal quality of the following vowel.
Of course, the earlier that they lower the velum during the preceding fricative closure, the less phonetically felicitous the fricative. But, in these cases, speakers appear to be willing make this sacrifice.
Note that this model allows for a number of possible interpolative paths through a window. However, the development of an explicit model of interpolation is still an outstanding issue for the theory. (cf. Pierrehumbert 1980, Keating 1990b, Huffman 1989, Cohn 1990 ).
Cohn’s English data do not include words such as seen. It would be interesting to see if her speakers exhibit anticipatory velum lowering prior to the release of the fricative in such cases.
The reader is referred to Cohn (1990) for the particulars of the phonological account of the facts.
I thank Patricia Keating (p.c.) for bringing this possibility to my attention.
For their part, I assume that voiceless stops have a raised velum target from the onset of closure that extends throughout the closure phase of the stop.
See, for example, Ladefoged and Maddieson (1996:111) for similar looking data on voiceless nasal stops in Burmese.
Perhaps this can be related to the notion of Lazy recently exploited by Kirchner (1995b), building on the work of Lindblom (1983) on the economy of speech gestures.
Recall that I assume that this /f/ has no voicing specification under the assumption that it is thus phonologically distinguished from the [-voice] opaque /f/ (see discussion in Chapter 3).
Note as well that the flow pattern on the token produced by S3 in (139) is additionally odd in that it exhibits an increase in nasal flow at both the onset and offset of the fricative, suggesting that the orality target at the right edge is not reached in this form, i.e. that the velum lowering gesture for the final vowel overlaps with the release of the fricative.
Note, however, that matters are complicated by the fact that S1 does not glottalize the preceding vowel. That is, she sometimes treats /f/ as voiceless with respect to nasalization, but not with respect to glottalization.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Gerfen, C. (1999). Nasalization. In: Phonology and Phonetics in Coatzospan Mixtec. Studies in Natural Language and Linguistic Theory, vol 48. Springer, Dordrecht. https://doi.org/10.1007/978-94-017-2620-7_4
Download citation
DOI: https://doi.org/10.1007/978-94-017-2620-7_4
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-5345-9
Online ISBN: 978-94-017-2620-7
eBook Packages: Springer Book Archive