1

1 PHONETIC AND PHONOLOGICAL RESEARCH ON NATIVE AMERICAN

2 LANGUAGES: PAST, PRESENT, AND FUTURE1

3

4 MATTHEW K. GORDON

5

6 UNIVERSITY OF CALIFORNIA, SANTA BARBARA

7 2

8 TABLE OF CONTENTS

9

10 1. Introduction

11 2. Early research on the sound systems of Native American languages

12 3. Instrumental phonetic research on Native American Languages

13 4. Phonological research on Native American Languages

14 5. Four case studies of the and of Native American Languages

15 5.1. Iambic metrical structure

16 5.1.1. Iambic lengthening in Chickasaw

17 5.1.2. High tone assignment in Creek

18 5.2. Weight-sensitive tone

19 5.3. Prosodic

20 5.4. Phonetics and phonology in endangerment and revitalization

21 6. Prospects for the future

22

23

24

25

26

27

28

29 3

30 1. Introduction

31

32 The study of Native American languages has played a prominent role in the development

33 of modern phonetic and phonological description and theory. Dating back to pioneering

34 work in the early 20th century by Franz Boas, Edward Sapir, and Pliny Earl Goddard,

35 detailed case studies of native languages of the Western Hemisphere have been integral

36 to fleshing out the typological variation in sound systems that serves as the basis for

37 theory construction and evaluation. Many of the insights of modern phonetics and

38 phonology crucially rely on the discovery in Native American language of features not

39 found in the Indo-European and Uralic languages providing the fodder for the pioneering

40 linguistic research of the 19th century. The advancements in phonetics and phonology

41 inspired by the study of Native American languages span diverse subfields, including

42 acoustic, articulatory, and auditory phonetics and phonological description and theory.

43 The breadth of methodologies attested in research on indigenous languages of the

44 Americas is matched by the wide range of phenomena that have been investigated. The

45 combination of the phonetic and phonological diversity and, in many cases, complexity

46 of Native American languages coupled with their relatively understudied status ensures

47 their continued importance in the enhancement of our understanding of phonetics and

48 phonology. On a practical level, new avenues of research on the phonetics and phonology

49 of Native American languages have also emerged as communities engage in language

50 revitalization efforts. 4

51 This paper provides an overview of how research on Native American languages

52 has contributed to advances in the fields of phonetics and phonology, spanning the period

53 shortly before the publication of IJAL’s inaugural issue in 1917 to the present. In order

54 to operationalize the presentation, discussion will be organized around the division

55 between instrumental phonetic studies and descriptive and theoretical research in

56 phonology, even though in practice the two threads of research are closely intertwined.

57 Work on languages spoken throughout the Americas will be discussed with emphasis on

58 languages of the Southeastern , which serve as the basis for three presented

59 case studies attesting to the important contributions of this linguistic area to our

60 knowledge of phonetics and phonology. A fourth case study is devoted to an increasingly

61 prominent area of work: the study of phonetic and phonological characteristics of

62 speakers, semi-speakers and learners of endangered languages.

63

64 2. Early research on the sound systems of Native American languages

65 In his introduction to the inaugural issue of IJAL in 1917, Franz Boas recognized the

66 relatively impoverished state of knowledge of Native American languages (especially

67 those of Central and South America) at that time but also expressed hope that recent

68 progress in the scope of material collected and the methodologies employed in its

69 collection augured well for future research. He also was well aware of the fragile status of

70 many languages owing to the rapid encroachment of socially and culturally dominant

71 European languages, contact with which threatened to dramatically alter the indigenous 5

72 languages of the Americas, e.g. the strength of the glottalized stops in languages of the

73 Pacific coast (Boas 1917a:2).

74 At the time of IJAL’s first edition, there had already been some descriptive

75 phonetic and phonological work on Native American languages of North America that

76 was considerably more rigorous than early cursory descriptions from explorers,

77 missionaries and traders. Research at the turn of the 20th century focused on descriptions

78 of particular languages and the synthesis of material from multiple languages to establish

79 genetic relationships between languages. Boas himself had published extensively on

80 Native American languages, including chapters on Tsimshian, Chinook, and Kwakiutl in

81 his own edited volume Handbook of American Indian Languages (Boas 1911), which

82 also contained descriptions of Tlingit, Haida, Fox, Maidu, Dakota, and . Sapir

83 was also already a prominent figure in language documentation and historical-

84 comparative by the time of IJAL’s inauguration, having published language

85 descriptions and comparative work on Chinookan, Wakashan, Uto-Aztecan, Na Dene,

86 and Takelma.

87 As portended by Boas, IJAL would play an integral role in the promotion of

88 research on Native American languages, including in the areas of phonetics and

89 phonology. Boas himself was at the vanguard of advancing IJAL’s agenda, publishing in

90 the first issue a description of the now extinct Pochutec variety of Nahuatl, including

91 several pages of phonology (Boas 1917b).

92 Boas’ paper on Pochutec would set a template for many future papers in IJAL that

93 provide language descriptions, which characteristically include a section devoted to 6

94 phonology. This traditional type of grammatical overview paper has been supplemented

95 with an increasing number of publications focused on particular phonetic or phonological

96 topics. Concomitant with the narrowing in scope of individual papers, progressively more

97 research is also showcased in non-areal journals publishing research in phonetics and/or

98 phonology. Through these publications as well as through early language descriptions

99 that still retain their importance, phonetic and phonological research on Native American

100 languages has played a prominent role in shaping the fields of phonetics and phonology.

101 The remainder of the paper summarizes these contributions from an historical

102 perspective, starting in the next section with phonetics.

103

104 3. Instrumental phonetic research on Native American Languages

105

106 In his introduction to IJAL, Boas (1917a) singles out for commendation the phonetic

107 fieldwork conducted by Pliny Earle Goddard at the turn of the 20th century on the Pacific

108 Coast Hupa and Kato (Goddard 1907; 1912). Goddard’s work was

109 ambitious in its employment of an impressive array of instrumental techniques to

110 investigate the aerodynamic and articulatory characteristics of both languages, Hupa

111 more comprehensively than Kato. His methodologies included photography to assess the

112 position of the mouth during the production of , palatography to examine the

113 location of contact between the tongue and the roof of the mouth, and kymography, to

114 investigate fluctuations in air pressure during speech. The kymograph could be outfitted

115 with different fittings to analyze data produced at different sources, including the mouth, 7

116 the nose, and the larynx. In a later study, Goddard (1928) investigated pitch variation to

117 demonstrate that Hupa (like other Pacific Coast Athabaskan languages and unlike

118 Southern Athabaskan and many Northern Athabaskan languages) does not use tone to

119 mark lexical contrasts. Goddard’s use of multiple instrumental methodologies was far

120 ahead of its time, only reemerging again at the end of the 20th century albeit with far

121 more sophisticated tools (see below).

122 Figure 1 illustrates two representative palatograms from Goddard (1907). The one

123 on the left shows the contact pattern on the roof of the mouth for the denti-alveolar /t/ in

124 /taw/ ‘no’, while the one on the right shows contact for the palatalized velar in /kʲʰa/

125 ‘dress’. In both palatograms, the area depicted is the region in front of the velum;

126 palatography is thus not an effective methodology for examining contact patterns for

127 posterior articulations. The contact between the tongue and the roof of the mouth,

128 represented as the dark region, is fairly broad in the front-back domain for the denti-

129 alveolar, extending from the alveolar ridge all the way forward to the back of the upper

130 teeth. For the palatalized velar stop, contact extends along both sides of the mouth

131 forward in a triangular pattern indicative of a raised tongue body associated with the

132 secondary palatalization gesture. The velar closure does not show up in the palatogram

133 since it is behind the depicted region.

134 8

135

136 Figure 1. Palatograms of the denti-alveolar stop in /t/ in /taw/ ‘no’ (on left) and the

137 palatalized velar in /kʲʰa/ ‘dress’ (on right) (Goddard 1907).

138

139 Figure 2 shows kymography traces from the mouth (the top line) and nose (the

140 bottom line) in the word /tʰaʔnaːn/ ‘water, to drink’ (Goddard 1907).

141 9

Oral

Nasal 142

143 t ʰ a ʔ n aː n 144

145

146 Figure 2. Kymography traces for the oral (top) and nasal (bottom) air in the word

147 /tʰaʔnaːn/ ‘water, to drink’.

148

149 The general relationship between oral and nasal flow is evident in the figure: as air comes

150 out through the nose in the production of the two nasals, it ceases to vent through the

151 mouth, although there is a brief transitional period of coarticulation between the nasal and

152 adjacent vowels during which air escapes through both the nose and mouth. The glottal

153 stop is apparent from the cessation of airflow through either the nasal or oral cavities. The 10

154 figure suggests venting of air through the nose during the oral stop at the beginning of the

155 word, plausibly reflecting passive breathing prior to phonation.

156 Although Goddard’s work provided insight into many characteristics of the sound

157 systems of Hupa and Kato, the limitations in both the instrumentation of the time and in

158 Goddard’s phonetic expertise yielded some descriptive inaccuracies (see Gordon 1996 for

159 a more recent phonetic study of Hupa). For example, in grappling with the description of

160 the ejective /t’/, Goddard correctly characterizes both its long lag -onset-time and

161 (less confidently) the glottalic nature of the airstream mechanism, but also incorrectly

162 suggests that it is produced with ingressive airflow:

163

164 “Hupa has another t formed in the same tongue position, but having quite

165 a different quality. It appears to lie between d [its voiceless unaspirated

166 counterpart] and t [its aspirated counterpart], and is at first distinguished

167 from them with great difficulty. It differs from d that there is a definite

168 period of time after the breaking of the contact before sonancy begins. It

169 differs from t in that it lacks the aspiration. In fact the breath seems to be

170 drawn in rather than forced out. This does not appear to be done from the

171 lungs but from the mouth, either by the sudden withdrawing of the tongue

172 enlarging the buccal cavity, or more probably by a closure of the glottis”

173 (Goddard 1907:14; italics mine).

174 11

175 The development of the sound spectrograph during World War II offered new

176 opportunities for the instrumental investigation of speech. In a series of four papers,

177 Hickerson (1958a; 1958b; 1959a; 1959b) employed spectography to conduct a

178 comprehensive study of both segmental and suprasegmental properties of Shawnee.

179 Hogan (1976) is a spectrographic study of quantitative and qualitative properties of Dëne

180 Sųłiné ejectives. He includes representative spectrograms illustrating the ejective

181 and afficates, which are characterized by a long lag voice-onset-time, similar to

182 those in Hupa.

183 Lindau’s (1984) cross-linguistic acoustic study of ejectives represents a

184 significant methodological advancement in that it includes data from multiple speakers of

185 the same language (including nine Navajo speakers), which allows for teasing apart

186 properties that are characteristic of the population of speakers as opposed to idiosyncratic

187 properties of someone’s idiolect. Dart’s (1991; 1993) phonetic research on the coronal

188 contrasts of Tohono O’odham is another relatively early multi-speaker study employing

189 spectral data, which Dart uses in conjunction with palatography and its inverse technique

190 linguography, in which contact patterns on the tongue are investigated.

191 A concentrated body of research on phonetic properties of Native American

192 languages emerged during the 1990s as part of the NSF Sounds of the World’s

193 Languages grant awarded to Peter Ladefoged and Ian Maddieson at UCLA (where

194 Lindau and Dart also conducted their research). Languages (from North, South, and

195 Central America) described in a series of UCLA Working Papers in Phonetics volumes

196 (and subsequently, in most cases, in other journals) include Navajo (de Jong and 12

197 McDonough 1993, McDonough and Ladefoged 1993, McDonough et al. 1993,

198 McDonough and Austin-Garrison 1994), Tlingit (Maddieson et al. 1996), Montana Salish

199 (Flemming et al. 1994) Western Apache (Potter et al. 2000), Jicarilla Apache (Tuttle

200 2000), Hupa (Gordon 1996), Aleut (Cho et al. 1997), Chickasaw (Gordon et al. 1997),

201 Jalapa Mazatec (Silverman et al. 1994), Banawá (Ladefoged and Everett 1996) and Wari’

202 (MacEachern et al. 1996).

203 These studies are quantitative and rely on data from several speakers of both

204 genders, though the number of speakers is small in the case of some severely threatened

205 languages. The papers include descriptions of basic phonetic properties, such as

206 formant frequencies, durational characteristics of and vowels, and, in the case

207 of tone languages, fundamental frequency values. They also include qualitative and

208 quantitative analyses of typologically unusual properties found in the particular language

209 under investigation. Topics of interest targeted by the studies include (among many

210 others) the ejective fricatives of Tlingit, the phonation contrasts of Jalapa Mazatec, the

211 velar vs. uvular distinction in Aleut, the clusters of Montana Salish, and the

212 atypical vowel of Wari’, which contains more front rounded vowels than back

213 rounded vowels.

214 The studies employ diverse methodologies. For example, Maddieson et al. (1996)

215 use acoustic and aerodynamic data in tandem to infer the articulatory characteristics of

216 the typologically rare Tlingit ejective fricatives, which are found in only 10 of 451 (2.2%)

217 languages in the UPSID survey (Maddieson and Precoda 1990). In their study of

218 Montana Salish, Flemming et al. (1994) complement their acoustic data with several 13

219 types of aerodynamic (oral pressure and flow and nasal flow) and physiological

220 (electroglottography) data.

221 The data from the Ladefoged and Maddieson project also have been pooled and

222 employed in typological studies of phonetic properties. For example, Cho and Ladefoged

223 (1999) is a survey of voice-onset-time (VOT) in 18 languages targeted for study by the

224 grant, the majority of which (11) are Native American languages. One of the goals of

225 their study is to explore the relationship between VOT values in languages lacking a

226 contrast between unaspirated and aspirated stops and in languages possessing an

227 aspiration contrast. One working hypothesis guiding their research is that VOT values for

228 unaspirated stops are smaller in languages with an aspiration contrast due to the need to

229 maintain a perceptually salient distinction with the aspirated series.

230 Figure 3 depicts average VOT values for the velar stops (the place of articulation

231 most broadly represented in the database) in the surveyed Native American languages.

232 The languages on the left lack an aspiration contrast for voiceless stops, whereas those on

233 the right have the contrast.

234 14

180 160 140 120 100 Unaspirated 80 60 Aspirated 40 20 0

235

236 Figure 3. Voice-onset-time (VOT) values (in milliseconds) for velar stops in 11 Native

237 American languages (Cho and Ladefoged 1999).

238

239 As figure 3 shows, although VOT values for the unaspirated series are generally smaller

240 (as predicted) in languages with an aspiration contrast, this pattern is not universal. VOT

241 values for the unaspirated stops in Hupa and Navajo, which both contrast aspiration, are

242 thus either equivalent to or larger than values for the unaspirated stops in Chickasaw and

243 Banawá, both of which lack an aspiration contrast. It is also striking that VOT values for

244 the two Aleut varieties, which lack an aspiration contrast, are roughly equivalent to (or

245 even larger than) VOT values for the phonemic aspirated stops in three of the five

246 languages with an aspiration contrast. The considerable cross-linguistic variability

247 observed in Cho and Ladefoged’s (1999) study underscores the extent to which the

248 phonetic realization of a phonological property may differ across languages, often in 15

249 ways that are not necessarily predictable from the structure of a language. Nevertheless,

250 despite the cross-linguistic variation in VOT values, the phonetic distinction between

251 unaspirated and aspirated stops is clearly maintained in all languages in Cho and

252 Ladefoged’s study with contrastive aspiration: VOT values for the aspirated series are

253 thus at least twice as long as those for unaspirated stops and, in some languages, four or

254 five times as long. Another interesting finding of the Cho and Ladefoged (1993) study is

255 that in three of the four languages surveyed with both an aspiration contrast and ejective

256 stops (Navajo, Western Apache, and Tlingit but not Hupa), VOT values for the ejectives

257 fall between those of their unaspirated and aspirated counterparts, suggesting that VOT

258 may be used as a supplemental cue to the identification of ejectives.

259 Recent work on Native American languages has built on phonetic research of the

260 20th century by employing more sophisticated techniques for directly accessing

261 articulatory data, rather than merely inferring it from acoustic and aerodynamic

262 measurements. Esling et al. (2005) use laryngoscopy in their study of Nuuchahnulth to

263 examine the production of glottal stop, glottalized , and pharyngeals. McDowell

264 (2004), Namdaran (2006), and Hudu (2008) employ ultrasound to investigate the

265 phonetic realization of vowel retraction triggered by uvulars and pharyngeal consonants

266 in Interior Salish languages. Gick et al. (2012) also use ultrasound data in their

267 examination of voiceless vowels in Oneida and Blackfoot, finding that articulatory

268 reflexes of these vowels remain even in the absence of acoustic traces. They further show

269 in a perception experiment involving Blackfoot listeners that listeners are perceptually 16

270 unable to distinguish the identity of vowels that have been acoustically but not

271 articulatorily deleted.

272 The Gick et al. (2012) study belongs to a growing area of research in Native

273 American languages, and more generally the field of phonetics, employing perception

274 experiments in tandem with production data in order to assess the mapping between

275 articulation and perception. Wright et al. (2002) complement their production study of

276 Witsuwit’en ejectives with a perception study, finding that the contrast between ejective

277 and unaspirated stops is less reliably perceived than other distinctions including the

278 contrast between aspirated and unaspirated stops. This result plausibly is tied to the

279 characteristically short VOT values for ejectives in Witsuwit’en relative to those in the

280 four languages with ejectives considered in the Cho and Ladefoged (1999) study

281 discussed earlier.

282 Because of their overall more robust status relative to their neighbors to the North

283 and South, much of the work on perception has focused on languages of Central America.

284 Gerfen and Baker (2005), for example, examine the acoustic and perceptual correlates of

285 laryngealized phonation in the Otomanguean language Coatzospan Mixtec, finding that

286 drops in amplitude or fundamental frequency alone are sufficient to cue judgments of

287 laryngealization even in the absence of differences in spectral tilt, which is typically

288 regarded as the most salient acoustic reflex of non-modal phonation (Gordon and

289 Ladefoged 2001). Clopper and Tonhauser (2013) examine the perceptual correlates of

290 focal prominence in Paraguayan Guaraní. Crowhurst and Olivares (2014) is a perception 17

291 study that explores rhythmic groupings and the psychological underpinnings of rhythmic

292 biases in Betaza Zapotec (see section 5.1 for more research on rhythmic groupings).

293 One area of research emphasized by Boas (1917a) to be of great importance,

294 studies based on narrative and discourse data, has recently gained greater prominence in

295 the phonetic literature. Because naturalistic data collected outside of the laboratory

296 introduce a number of confounding factors that can make results difficult to interpret,

297 most quantitative phonetic studies on Native American languages (and more generally in

298 the field of phonetics) are based on elicited data. Discourse data, however, are critical to

299 gain a comprehensive understanding of prosodic properties such as prominence,

300 intonation, and prosodic constituency. Hintz’s (2006) paper on South Conchucos

301 Quechua and Gordon and Rose’s (2006) work on Emerillón are two acoustic studies of

302 correlates in Native American languages of South America that compare words

303 uttered in isolation with narrative data. Both studies find differences in the phonetic

304 realization of prominence and in its location between the two contexts, confirming the

305 need for more discourse-based studies of prosody. Several recent studies have used

306 narrative data to explore higher level prosodic units in Native American languages, e.g.

307 Beck and Bennett (2007) on , Berez (2011) on Ahtna, and Lovick and Tuttle

308 (2012) on Dena’ina. Along similar lines, Russell (2008) is an acoustic study of vowel

309 coalescence across word boundaries in Plains Cree based on data from two narratives.

310

311

312 18

313 4. Phonological research on Native American Languages

314

315 Research on languages of the Americas has also played an important role in advances in

316 phonological description and theory over the last century. Already at the turn of the 20th

317 century, Sapir was churning out remarkably detailed descriptions of phonological

318 patterns in a variety of languages. Sapir’s (1912) grammar of Takelma illustrates the

319 precision and scope of his phonological analyses, which comprised a broad range of

320 topics, including phoneme inventories, phonotactic restrictions, segmental alternations

321 (e.g. , dissimilation, deletion, degemination), and prosody. Many of his

322 phonological descriptions inform phonological theory even today, and the value of his

323 work is heightened by the fact that most of the languages he described are now either

324 moribund or extinct. Sapir’s (1930) description of the phonology of Southern Paiute, for

325 example, has sparked debate about syllable integrity in stress systems (Halle and

326 Vergnaud 1987, Hayes 1995), i.e. whether foot boundaries can split syllables, and the

327 characterization of the relationship between and segmental alternations

328 (McCarthy and Prince 1995, Gurevich 2000, Raimy 2000). Even Sapir’s textual materials

329 have been consulted in corpus-based phonological studies. Gordon and Luna (2004), for

330 example, analyze the location of stress for a subset of Sapir’s Hupa texts (Sapir and Golla

331 2001), extracting a number of statistical biases in stress placement that have analogs even

332 at later stages of the language.

333 Descriptive work on the phonology of Native American languages has continued

334 unabated over the last century, much of it represented in IJAL articles and publications. 19

335 Over time, as descriptive coverage of most languages has expanded, increasingly more

336 research has employed data from Native American languages to inform topical issues in

337 phonological theory. Research on Native American languages has played a prominent

338 role in the development and elucidation of many phonological frameworks and

339 formalisms, most recently the constraint-based framework of Optimality Theory (Prince

340 and Smolensky 1993). In many cases, data from older comprehensive grammars of now

341 extinct languages continue to receive extensive treatment in the theoretical literature. The

342 importance of Sapir’s Southern Paiute data for metrical stress theory was mentioned

343 earlier. Similarly, Stanley Newman’s (1944) grammar of Yokuts provided the material

344 for Archangeli’s (1984) seminal dissertation on non-linear phonology and

345 underspecification of features. Hoijer’s (1933) description of and shortening in

346 Tonkawa has likewise provided ample fodder for the study of metrical structure and

347 reduplication (e.g. Phelps 1975, Kenstowicz and Kisseberth 1979, Noske 1993,

348 Gouskova 2003; 2007).

349 The data from these classic works as well as from more recent studies of Native

350 American languages have been instrumental in broadening the typological coverage of a

351 wide range of phenomena, including phonemic inventories, phonotactics, syllable

352 structure, stress, tone, intonation, poetic metrics, prosodic morphology, and prosodic

353 constituency. I cite here just a few of the contributions of phonological research on

354 Native American languages to these topics. Pirahã, which has only seven consonants and

355 eight vowels (Everett 1986), possesses the smallest phoneme inventory in the world

356 (Maddieson 1984), while varieties of Eastern Chatino have perhaps the most complex 20

357 tone inventories in the world, contrasting as many as twelve tones (Cruz and Woodbury

358 2006, Villard 2009, Campbell and Woodbury 2010). On an historical level, Northern

359 Athabaskan languages, which diverge in their synchronic tonal reflexes of glottal

360 constriction, have played a pivotal role in our understanding of tonogenesis (Kingston

361 2005). The vowelless words of certain , e.g. Nuxalk (Bagemihl 1991),

362 present challenges to theories that assume the universality of syllables and sonority

363 sequencing principles governing phonotactics. Sibilant harmony systems of the type

364 found in Chumashan and Athabaskan have played a prominent role in the development of

365 theories of long-distance assimilation (Gafos 1999, Hansson 2001, Poser 2004, McCarthy

366 2007). Several intonation systems reported for languages of the Americas, e.g. Wari’

367 (Everett and Kern 1997) and Chickasaw (Gordon 2005a), violate the overwhelming

368 cross-linguistic tendency for the default intonational contour in declaratives to be

369 characterized by a terminal fall. Work by Hinton (1984; 1990) on Havasupai and

370 Fitzgerald (1998) on Tohono O’odham provide much needed breadth to typological

371 knowledge of linguistic properties of sung verse. The stress system of Pirahã (Everett

372 and Everett 1984, Everett 1986) has provided the impetus for research on onset-sensitive

373 weight (Topintzi 2004; 2010, Gordon 2005b) with further reinforcement from the

374 sensitivity of the minimal word requirement to onset complexity in Yakima Sahaptin

375 (Hargus and Beavert 2006). The prosodic system of Kashaya (Oswalt 1988, Buckley

376 1994; 2013) has necessitated expansion of the theory of extrametricality to include cases

377 at the left edge of prosodic domains. The scalar weight hierarchies of Asheninca (Payne

378 1990) and Nanti (Crowhurst and Michael 2005), which are simultaneously sensitive to 21

379 vowel quality, vowel length and the presence vs. absence of a coda consonant, present

380 challenges to the most commonly adopted theory of weight, moraic theory (Hayes 1989).

381 Native American languages of the Pacific Northwest have provided virtually the entire

382 corpus of material underlying the debate about the role of phonetic factors in analyzing

383 the phasing of laryngeal and oral gestures relative to each other in glottalized sonorants

384 (e.g. Plauché et al. 1998, Steriade 1999, Howe and Pulleyblank 2001, Shaw et al. 2005,

385 Bird et al. 2008, Bird 2011).

386

387 5. Four case studies of the phonetics and phonology of Native American Languages

388

389 This section presents four case studies illustrating ways in which work on Native

390 American languages have contributed to the fields of phonetics and phonology. The first

391 three case studies focus on prosodic features, including iambic metrical structure (5.1),

392 weight-sensitive tone restrictions (5.2), and prosodic morphology (5.3). The final case

393 study (5.4) deals with an area of ever increasing importance as the number of endangered

394 languages grows: phonological and phonetic characteristics of semi-speakers and their

395 acquisition by second language learners. Focus in the case studies, particularly the first

396 three, is on languages of the Southeastern United States, which have played a prominent

397 role in furthering our understanding of phonetics and phonology over the last century.

398

399

400 22

401 5.1. Iambic metrical structure

402

403 The most widely adopted metrical stress theory, that proposed in Hayes (1987; 1995)

404 assumes an asymmetric inventory of feet consisting of three types. According to this

405 theory, trochaic feet may, on a language-specific basis, either be weight-sensitive or not,

406 whereas iambic feet are universally weight-sensitive. The three types of feet in Hayes’

407 proposed inventory are shown schematically in (1), where the metrically strong element

408 is marked with an ‘x’.

409

410 (1) Asymmetric foot inventory (Hayes 1987; 1995)

411

Syllabic Trochee Moraic Trochee Iamb

x x x

(σ σ) (µ µ) (µ µ)

412

413 As (1) shows, canonical weight-sensitive feet, including both the moraic trochee and the

414 iamb, consist of two moras (assigned to vowels and, on a language-specific basis, coda

415 consonants), whereas weight-insensitive feet contain two syllables. Hayes grounds his

416 foot inventory in data from numerous psychological experiments in which listeners prefer

417 trochaic groupings for alternating loud and soft stimuli that are durationally balanced but

418 iambic groupings for alternating long and short stimuli that are balanced in terms of

419 loudness. Results of these experiments suggest an inherent sensitivity to duration in the 23

420 case of iambs but not trochees. Hayes finds linguistic support for this asymmetry between

421 the two foot types in his cross-linguistic survey of stress systems, showing that iambic

422 stress languages are overwhelmingly (if not exclusively) weight-sensitive, whereas

423 trochaic stress languages vary in their weight-sensitivity.

424 Virtually all of Hayes’ (1995) survey of iambic languages is based on languages

425 of the Americas (many of which are genetically diverse), including Hixkaryana, Creek,

426 Chickasaw, Choctaw, Delaware, Malecite-Passamaquoddy, Eastern Ojibwa, Menomini,

427 Potawatomi, Cayuga, Seneca, Onondaga, Yupik, Hopi, Sierra Miwok, and Maidu. The

428 Southeastern languages Chickasaw, Choctaw and Creek contribute some of the especially

429 key case studies illustrating iambic metrical structure. These are discussed in the next

430 two sections.

431

432 5.1.1. Iambic lengthening in Chickasaw

433

434 One of the pervasive features of iambic systems, exemplified by Chickasaw and

435 Choctaw, is iambic lengthening, a phenomenon whereby metrically strong phonemic

436 short vowels in open syllables lengthen to create a durationally unbalanced short-long

437 foot just like the preferred iambs in the psychological experiments discussed above.

438 Examples of iambic lengthening in Chickasaw (Munro and Ulrich 1984, Munro and

439 Willmond 1994; 2008, Gordon and Munro 2007) appear in (2), where lengthened vowels

440 are indicated by an IPA half-length symbol. As in other languages with weight-sensitive

441 iambs, feet in Chickasaw may consist of two light syllables, a single heavy syllable 24

442 (closed syllables and those with long vowels in Chickasaw), or a light followed by a

443 heavy syllable. In Chickasaw, unlike many other iambic stress language, a final light

444 syllable immediately following a strong syllable also constitutes a foot on its own. Note

445 that primary stress in the Chickasaw examples preferentially docks on a long or

446 lengthened vowel and otherwise on the final syllables in words lacking a long vowel.

447

448 (2) Iambic lengthening in Chickasaw

449

(aˈbiˑ)(kaˌtok) ‘s/he was sick’

(aˌsaˑ)(biˈkaˑ)(ˌtok) ‘I was sick’

(kiˈsiˑ)(liˌtok) ‘I bit it’

(ʧiˌkiˑ)(siˈliˑ)(ˌtok) ‘I bit you’

(ʧoˌkoʃ)(koˈmoˑ)(ˌtok) ‘s/he played’

(aˈsaˑ)(biˌka) ‘I am sick’

(ʧoˌkoʃ)(koˈmo) ‘s/he plays’

(aˈbiˑ)(ˌka) ‘s/he is sick’

(kiˈsiˑ)(ˌli) ‘I bite it’

450

451 A curious feature of languages with iambic lengthening is the apparent absence

452 (according to phonological descriptions) of lengthening in final syllables even if they are

453 open and metrically strong, e.g. (ʧoˌkoʃ)(koˈmo) not (ʧoˌkoʃ)(koˈmoˑ), (aˈsaˑ)(biˌka) not

454 (aˌsaˑ)(biˈkaˑ). The failure of lengthening to apply word-finally is particularly puzzling 25

455 given that final position is cross-linguistically a common locus of phonetic lengthening in

456 languages as diverse as Taiwanese (Peng 1997), Yoruba (Nagano-Madsen 1992),

457 Inuktitut (Nagano-Madsen 1992), English (Klatt 1975, Umeda 1975, Wightman et al.

458 1992), Hungarian (Hockey and Fagyal 1999), and Chickasaw’s Muskogean relative

459 Creek (Johnson and Martin 2001). One might thus expect final syllables to be most

460 conducive to iambic lengthening.

461 Gordon and Munro (2007) explore the puzzling absence of lengthening final

462 stressed syllables in a phonetic study of final vowel length under different metrical

463 conditions in Chickasaw. Since all final syllables in Chickasaw are metrically strong as

464 evidenced by stress patterns (Munro 1996, Gordon 2004), the relevant comparisons are

465 between final vowels that are part of a disyllabic foot consisting of two light syllables and

466 final vowels that consist of a single light (CV) syllable. Gordon and Munro (2007) also

467 compare vowels that are word-final but phrase-internal with those that are phrase-final

468 but not utterance-final and those that are utterance-final. They find that all final vowels

469 undergo lengthening regardless of the size of the foot to which they belong; thus, the final

470 vowel in (aˈbiˑ)(ˌka) ‘s/he is sick’ is equivalent in length to the final vowel in

471 (aˈsaˑ)(biˌka) ‘I am sick’ despite the fact that the rightmost foot is monosyllabic in the

472 first word but disyllabic in the second. They further discover that word-final lengthening

473 is cumulative across domains of different sizes: greatest utterance-finally, slightly smaller

474 phrase-finally but utterance-medially, and smallest in magnitude phrase-medially. Much

475 of the length (almost 40% averaged across speakers) of utterance-final vowels, however,

476 is associated with breathiness, unlike phrase-final and word-final vowels (that are not 26

477 utterance-final), which have considerably shorter breathy phases: 17% for phrase-final

478 vowels and 10% for word-final vowels. The cumulative nature of final lengthening and

479 breathiness across domains of different sizes is consistent with cross-linguistic patterns,

480 leading Gordon and Munro (2007) to suggest that the natural tendency for final

481 lengthening, which can be associated with slowing down of articulatory gestures without

482 necessarily any increase in gestural magnitude (see Edwards et al. 1991, Beckman et al.

483 1992), plausibly supersedes lengthening due to metrical prominence, which unlike final

484 lengthening is attributed to an increase in gestural magnitude.

485 Another interesting finding of Gordon and Munro’s (2007) study is that word-

486 medial lengthened vowels remain durationally distinct from phonemic long vowels (at

487 least for certain speakers). The phonemic long /iː/ in (tiː)(kãːʔ)(tiʔ) ‘dirt dauber (wasp)’

488 for you’ is thus longer than the lengthened /iˑ/ in (satiˑ)(kah)(bi) ‘I am tired’. Iambic

489 lengthening in Chickasaw thus falls in the class of near-neutralizing alternations, which

490 makes it problematic to capture using discrete units of weight like the mora. Under

491 Hayes’ (1995) account, iambic lengthening creates a heavy, i.e. bimoraic, strong syllable;

492 however, assuming that the lengthened vowel is bimoraic incorrectly implies that it is

493 equivalent in length to phonemic long vowels. It is hoped that future phonetic study of

494 iambic lengthening in other languages will determine whether it triggers near or complete

495 neutralization of underlying length distinctions. Study of other languages in which final

496 short vowels are only stressed when part of a disyllabic foot (unlike in Chickasaw) will

497 also reveal whether the final lengthening patterns in Chickasaw are potentially an artifact

498 of the final stress or are also observed in final unstressed vowels. 27

499 5.1.2. High tone assignment in Creek

500

501 Not all iambic languages manifest metrical strength through vowel lengthening. Another

502 Muskogean language, Creek, has a high tone that occurs in a location that is predictable

503 from iambic foot structure. As described by Haas (1977) and Martin (2011) and

504 instrumentally verified by Martin and Johnson (2002), the default high tone in Creek

505 words lacking a lexically specified accent spans from the leftmost to the rightmost

506 metrically strong syllable, either an initial heavy syllable containing a long vowel or a

507 coda consonant, or the second syllable if the first syllable is light. Metrical structure and

508 tonal patterns in Creek are illustrated in figure 4 for the words (atʃok)(ɬán)wa ‘daddy

509 longlegs’ and (apa)(taná) ‘bullfrog’. In the second form, the high tone span continues

510 from the second vowel all the way through most of the final vowel, which is metrically

511 strong. In the first form, in contrast, there is a precipitous fall in F0 starting at the

512 beginning of the final vowel, which is metrically unparsed.

513

514

515

516

517

518

519

520 28

521 522 523 524

175 525 150 H

) H ) z z H ( H

(

h 528 h c t c i t i P 529 P 530 100 100 0 5310.9756 0 0.6971 Time (s) 532 Time (s) 533 a tʃ o k ɬ a n w a a p a t a n a 534 535 Figure 4. Iambic high tone assignment in the Creek words (atʃok)(ɬán)wa ‘daddy

536 longlegs’ (on left) and (apa)(taná) ‘bullfrog’ (on right)

537 In a study of vowel quality and duration, Johnson and Martin (2001) find that

538 final vowels, both phonemically short and long, are centralized relative to their initial

539 counterparts despite being longer, a result that suggests that final position triggers both a

540 slowing down of articulatory gestures and a decrease in tongue displacement in Creek.

541 This combination of results is consistent with the hypothesis advanced above that final

542 lengthening in Chickasaw has a fundamentally different articulatory source from

543 lengthening associated with metrical prominence.

544

545 5.2. Weight-sensitive tone

546

547 Another phenomenon informed by research on languages of the Southeastern United

548 States is weight-sensitive tone, the restriction against contour tones on light syllables

549 observed in many languages (Clark 1983, Hyman 1988, Gordon 2001, Zhang 2002). If 29

550 contour tones are viewed compositionally as combinations of low plus high tones (in

551 either order), restrictions against them can be captured in theories of weight, such as

552 moraic theory, as a one-to-one upper limit on associations between tones and units of

553 weight (Woo 1969). In languages imposing this constraint on tone-to-weight mappings,

554 the configuration in (3b) with two tones linked to a single mora is not licit, whereas the

555 same pattern is permitted in a language that does not place weight-sensitive restrictions

556 on the licensing of contour tones. One-to-one mappings between tones and moras, as in

557 (3a) and (3d), are allowed in both types of languages as are mappings of a single tone to

558 multiple moras, as in (3c).

559

560 (3) Various tone-to-mora mappings in languages with and without weight-sensitive tone

561 (a) σ (b) σ (c) σ (d) σ

µ µ µ µ µ µ

T T T T T T Weight-sensitive tone √ * √ √ No weight-sensitive tone √ √ √ √ 562 563 Contour tone restrictions are illustrated by Koasati (Gordon et al. 2015), in which a

564 subset of carry a rising tone and this rising tone is (almost exclusively) limited to

565 syllables containing a long vowel (abbreviated CVV) or syllables closed by a coda

566 (abbreviated CVR). Examples of lexical rising tone in Koasati appear in (4),

567 followed by a pitch trace of a word with rising tone in figure 5.

568

569 30

570 (4) Lexical rising tone on CVV and CVR in Koasati

571

hopǒːni ‘a cook’

tǎlwa ‘singer’

pokko tǒːli ‘ball player’

naːɬǐːka ‘speaker’

mobǐːla ‘car’

athǒmma ‘Indian’

572 573 574 575

576 200 577 H ) z H (

578

h L c t i

579 P 580 581 125 0 0.8256 582 Time (s) 583 h o p ǒː n i 584 585

586 Figure 5. Lexical rising tone on the penult of the Koasati hopǒːni ‘a cook’

587

588 A similar confinement of phonological contour tones to CVV and CVR syllables

589 is also found in (Chafe 1976) and in Koasati’s linguistic relative Creek (Martin

590 2011). Caddo differs from Koasati, however, in permitting only falling and not rising

591 tones. 31

592 Oklahoma Cherokee also has a restriction against contour tones, which differs

593 from the Koasati and Caddo pattern in permitting contours only on syllables containing a

594 long vowel (Wright 1996, Uchihara 2013). Oklahoma Cherokee also has a more complex

595 tone inventory than the Koasati and Caddo data discussed thus far in featuring both rising

596 and falling tones. Examples of rising and falling tone on long vowels in Cherokee appear

597 in (5) (from Uchihara 2013:172).

598

599

600 (5) Contour tones on CVV in Oklahoma Cherokee

601

kʰijûːka ‘chipmunk’ 602

kakʰâːneha ‘he is giving him a living thing’

kʰawǒːnu ‘duck’

kaleːjʌˇːska ‘A long object is falling from an

upright position’

603

604 Because it is realized along the same physical dimension of fundamental

605 frequency, intonation in many languages is subject to similar restrictions on pitch

606 excursions to those governing lexical tone. For example, imperatives in Koasati are

607 marked with a terminal high-low fall in pitch, which is realized in its canonical form on

608 final syllables that end in a sonorant coda. (Koasati lacks final long vowels.) If, however,

609 the final syllable is not CVR (which also licenses low-high lexical rises) the high-low fall 32

610 is truncated to a simple high tone. The canonical and truncated versions of the terminal

611 high-low fall in Koasati imperatives are illustrated for the words ishol ‘you all take it!’

612 and halatk ‘grab on!’, respectively, in figure 6. Note that both words begin with a

613 lowered baseline pitch (not attributed to the imperative), which accounts for the initial

614 rise to the high tone.

615

250 275 H H ) ) z z H H ( (

h h c c t t i i P L P

100 100 0 0.5972 0 0.5456 616 Time (s) Time (s) 617 i s h o l h a l a t k 618 619 Figure 6. Falling tone in Koasati imperative ishol ‘you all take it!’ (on left) and simple

620 high tone in Koasati imperative halatk ‘grab on!’ (on right)

621

622 The Koasati/Caddo type (CVV, CVR heavy) and Cherokee type (CVV heavy)

623 tone restrictions are typologically very common in the world and find an explanation in

624 terms of phonetic considerations. The acoustic correlate of tone is fundamental

625 frequency, which is most effectively realized on a sonorous sound, where vowels provide

626 a slightly better backdrop than sonorant consonants. are least conducive to

627 supporting fundamental frequency information. Because contour tones involve a

628 fundamental frequency excursion, they generally require more time to implement than 33

629 level tones. For this reason, a long vowel provides the best docking site for a contour

630 tone, slightly better than a sequence of a short vowel plus sonorant consonant. Both CVV

631 and CVR are far superior in their tone-bearing capacity to either a short vowel in an open

632 syllable or one followed by an coda (see Zhang 2002 for further discussion).

633 Languages like Cherokee that restrict contour tones to CVV thus impose the most

634 stringent requirement on contour tones, whereas those like Koasati and Caddo that permit

635 contour tones on both CVV and CVR employ a slightly less strict weight criterion.

636 Predicted not to occur given phonetic grounding is a hypothetical language in which

637 syllables closed by an obstruent are able to support contour tones but those closed by a

638 sonorant are not or a hypothetical language in which CVR can carry contours but CVV

639 cannot.

640 Tonal restrictions may also be sensitive to the type of contour tone. Zhang (2002)

641 observes that rising tones are subject to more stringent restrictions than falling tones

642 cross-linguistically. Many languages thus have falling but not rising tones, while others

643 restrict rising tones more severely in terms of the contexts in which they may occur. This

644 asymmetry between the two types of contours also falls out from phonetic considerations

645 discussed in Zhang (2002): rising tones take longer to execute than falling tones. One

646 would thus predict that tolerance of rising tones in a language predicts the occurrence of

647 falling tones in that language.

648 The Caddo and Creek data discussed thus far display the predicted type of

649 asymmetry in contour tone licensing, since they possess falling but not rising tones.

650 Koasati, however, contradicts this pattern in having rising but not falling lexical tones 34

651 (though it does have falling intonational contours, as shown earlier), suggesting that the

652 distribution of contour tones is not necessarily always predictable on phonetic grounds

653 synchronically.

654 As it turns out, consideration of the tonal history of the Muskogean language

655 family to which Koasati and Creek both belong sheds some light on the apparently

656 anomalous behavior of the Koasati tone system synchronically. As Martin (2013) and

657 Gordon et al. (2015) show, rising tone in Koasati corresponds to falling tone in some

658 other , including Koasati’s close relative Alabama, e.g. Koasati

659 tǎlwa vs. Alabama tậlwa ‘singer’ (Gordon et al. 2015:115). Martin (2013) and Gordon et

660 al. (2015) suggest that falling tone reflects the original pattern and that the rising tone of

661 Koasati resulted historically from the typologically common phonetic phenomenon of

662 peak delay (e.g. Silverman and Pierrehumbert 1990, Prieto et al. 1995, Xu 2001),

663 whereby the fundamental frequency peak is shifted rightward phonetically. As a result of

664 peak delay, the syllable originally associated with the falling tone became phonetically

665 associated with an F0 rise, eventually being analyzed as carrying a rising tone instead.

666 The Koasati reanalysis provides an illustration of how one phonetically natural event, in

667 this case, peak delay, can trigger a typologically uncommon (and phonetically

668 unpredicted) phonological pattern, in this case, the occurrence of a phonological rising

669 but not a falling tone.

670

671

672 35

673 5.3. Prosodic morphology

674

675 Prosodic morphology is a cover term for a cluster of phenomena sensitive to the

676 phonological shape of morphemes and formally united in a research program initiated by

677 McCarthy and Prince (1986/1996). One property falling under the purview of prosodic

678 morphology is the class of templatic morphological operations characterized by the

679 selection of a prosodically defined morpheme. The most pervasive type of templatic

680 morphology is reduplication, a process of word formation (often conveying meanings

681 such as intensification, repeated action, or plurality) associated with the copying of a

682 morpheme or part of a morpheme. The examples in (6) illustrate the marking of plurality

683 through reduplication in Tunica data appearing in an early IJAL article by Swanton

684 (1921).

685

686 (6) Reduplication in Tunica (Swanton 1921:5-6)

687

Base Gloss Reduplication Gloss

miːliː ‘red miːliːmiːliːta ‘many red things’

meːliː ‘black’ meːliːmeːliːta ‘many black things’

toːluː ‘round’ toːluːtoːluːta ‘many round things’

koːra ‘to drink’ koːkoːra ‘drink (pl.)’

36

688 As the forms from Tunica indicate, the shape of the reduplicant can vary. In these data,

689 the first three forms thus display reduplication of the entire base, while the last form

690 undergoes reduplication of the first CV sequence.

691 A salient, though not universal feature, of reduplication is that the reduplicant

692 adheres to a defined template rather than merely copying a constituent of the . This

693 property is evident in the Creek data in (7), which illustrate reduplication in the

694 distributive of intransitive verbs (Martin 2011).

695

696 (7) CV reduplication in Creek (Martin 2011:203-204)

697

Base Gloss Reduplication Gloss

(a) likátʃw-iː ‘dirty’ likatʃliw-íː (two or more)

hopánk-iː ‘broken’ hopanhok-íː (two or more)

(b) apoːk-itá ‘(three or more) apoːpok-itá (in several places)

to sit, live’

tóːsk-iː ‘mangy’ toːstok-íː (two or more)

(c) hátk-iː ‘white’ háthak-íː (two or more)

tolk-itá ‘to fall over’ toltok-íta (two or more)

698

699 The reduplicant is characteristically a copy of the first CV sequence of the root and falls

700 to the left of the final consonant of the root (7a). The fixed nature of the reduplicative

701 template as CV is demonstrated by the fact that a long vowel in the first syllable of the 37

702 root surfaces as short in the reduplicant (7b) and a coda consonant is not copied (7c). The

703 CV reduplicant template employed in Creek is one of many attested in languages

704 throughout the world. Others include a single heavy syllable (either CVV or CVC), a

705 long vowel (CVV), a disyllabic sequence, and an entire root.

706 A relatively rare type of reduplication involves copying of a single segment. The

707 data in (8) and (9) illustrate consonant reduplication in Jakalktek (Day 1973) and vowel

708 reduplication in Nuxalk (Nater 1984:109), respectively.

709

710 (8) Consonant reduplication in Jakaltek (Day 1973:45)

Base Gloss Reduplication Gloss

piʦ’-a ‘squeeze s.th. ʧa piʦ’p-e ‘you squeeze s.th.

gently’ gently several times’

onom juk-e ‘noise and/or ʧa jukj-e ‘you shake s.th.’

motion of leaves

on shaken trees’

711

712 (9) Vowel reduplication in Nuxalk (Nater 1984:109)

713

Base Gloss Reduplication Gloss

t’ixɬala ‘robin’ ʔit’ixɬala-j (diminutive)

k’n̩ ts ‘sperm whale’ ʔn̩ k’n̩ ts (diminutive)

714 38

715 Reduplication can also involve two copies or “triplication”, as in the data in (10)

716 from Stát’imcets (Shaw 2005, van Eijk 1997) combining diminutive and

717 plural/collectivity reduplication (with accompanying of the sonorant in the

718 second set of examples).

719

720 (10) Multiple reduplication in Stát’imcets (van Eijk 1997:62, Shaw 2005:177)

721

s-ˈqaχaʔ Nom-Root ‘dog’

s-ˈqə-qχaʔ Nom-DIM-Root ‘puppy’

s-qəχ-ˈqə-qχaʔ Nom-PL-DIM -Root ‘puppies’

s-ˈjaqʧaʔ Nom-Root ‘woman’

s-ˈjʔə-jʔqʧaʔ Nom-DIM-Root +[CG] ‘girl’

s-ˈjəq-jʔə-ˈjʔqʧaʔ Nom-PL-DIM-Root +[CG] ‘girls’

722

723 It is also possible for reduplication to involve copying in conjunction with an

724 invariant segment. Koasati displays this variety of “fixed segmentism” reduplication in

725 the construction termed “punctual reduplication” by Kimball (1991), in which the first

726 consonant of the root plus the vowel [oː] is inserted before the final syllable of the root.

727

728

729 39

730 (11) Fixed segmentism in Koasati punctual reduplication (Kimball 1991:325-6)

731

tahaspin tahastoːpin ‘to be light in weight’

lapatkin lapatloːkin ‘to be narrow’

talasban talastoːban ‘to be thin’

ɬimihkon ɬimihɬoːkin ‘to be smooth’

ʧofoknan ʧofokʧoːnan ‘to be angled’

732

733 As stated earlier, a salient characteristic of reduplication is its adherence to a

734 template that is not dependent on the shape of the base. Although this feature is nearly

735 universal, there are rare instances of reduplication varying as a function of the shape of

736 the base. One such exceptional case involves syllable reduplication in the habitual aspect

737 in the Uto-Aztecan language Hiaki (Haugen 2003; 2014). In Hiaki, the shape of the

738 reduplicant varies between CV and CVC to match the shape of the first syllable of the

739 root (12).

740

741 (12) Syllable reduplication in Hiaki (Haugen 2014:511)

742

Template Base Reduplication Gloss

CV vu.sa vu.vu.sa ‘awaken’

ʧi.ke ʧi.ʧi.ke ‘comb one’s hair’

he.wi.te he.he.wi.te ‘agree’ 40

CVC vamse vam.vam.se ‘hurry’

hit.ta hit.hit.ta ‘make a fire’

ʔat.bʷa ʔat.ʔat.bʷa ‘laugh at’

bʷal.ko.te bʷal.bʷal.ko.te ‘soften, smooth’

743

744 Reduplication is not the only type of templatic morphological operation attested

745 cross-linguistically. Another variety of templatic morphology involves truncation of

746 material from the base to adhere to a fixed shape, as in hypocoristic formation in

747 Japanese (Poser 1990), in which the abbreviated name adheres to a bimoraic template

748 consisting of two light syllables, e.g. takatɯgɯ → taka-ʨaɴ, taɺoː → taɺo-ʨaɴ, or a

749 single heavy syllable, e.g. joːko → joː-ʨaɴ.

750 Although the template in most cases of truncation is enforced on the preserved

751 material (as in Japanese), there is another type of truncation, in which the template

752 instead holds of the deleted (or subtracted) material. Subtractive morphology is illustrated

753 by the Koasati plurals in (13), which are formed by deleting the final rime of the root

754 (either a short vowel + coda consonant or a long vowel) and adding the suffixes –ka or –li

755 followed by –n (Kimball 1983; 1991, Martin 1988).

756

757

758

759 41

760 (13) Subtractive morphology in Koasati (Martin 1988:230-1)

761

Singular Plural Gloss

lataf-kan lat-kan ‘to kick something’

lasap-lin lap-lin ‘to lick something’

misip-lin mis-lin ‘to wink’

tipas-lin tip-lin ‘to pick something off’

atakaː-lin atak-lin ‘to hang something’

albitiː-lin albit-lin ‘to place on top of’

aʧokʧanaː-kan aʧokʧan-kan ‘to quarrel with someone’

762

763 In summary, Native American languages have contributed substantially to typological

764 knowledge of prosodic morphology, including the range of variation in the shape of

765 templates and the types of phenomena employing those templates.

766

767 5.4. Phonetics and phonology in language endangerment and revitalization

768

769 The decline in language usage in many Native American communities has spawned new

770 areas of engagement that transcend the traditional boundaries of scholarly research.

771 Already at the time of IJAL’s inception, Boas (1917a) had recognized the profound

772 impact that extensive contact with European languages had exerted on the phonetic

773 characteristics of Native American languages, citing as an example the reduced strength 42

774 of the glottalized consonants found in languages of the Pacific Coast. The influence of

775 language contact has become increasingly more pervasive over the last century, spurring

776 linguists and community members to investigate the ways in which languages are shaped

777 by other languages, the differences between the speech of native speakers and that of

778 semi-speakers of a language, and the mechanism of language acquisition for second

779 language speakers of Native American languages. On a practical level, understanding of

780 these issues informs methodologies for effectively teaching second language learners. An

781 important thread of this work, both theoretical and applied, is the inextricable relationship

782 it assumes between the language itself and its speakers (or learners) for whom language is

783 an essential marker of cultural identity.

784 Rankin’s (1978) paper on the phonemes of , which had at the time of his

785 work recently lost its last native speaker, made clear the complex nature of phonological

786 change associated with language desuetude in the face of language contact (see also Cook

787 1989 on Tsuut’ina and Dëne Sųłiné, the latter of which is also the subject of a

788 longitudinal acquisition study in Cook 2006). Although certain shifts he reports likely

789 reflect the influence of English, e.g. the deglottalization of glottalized fricatives and

790 stops, the change of retroflex fricatives to palato-alveolars and ultimately clusters of

791 alveolar plus /r/, other developments were not plausibly attributed to English.

792 For example, for semi-speakers whose parents had been fluent and who still could

793 produce many words (between 150 and 300) and some short sentences, the contrast

794 between aspirated and unaspirated stops was neutralized to the unaspirated series even in

795 pre-tonic contexts where English has aspiration. Rankin (1978) suggests that this shift is 43

796 attributed to the less marked status of unaspirated stops relative to aspirated stops, as

797 reflected in cross-linguistic frequency patterns and ultimately attributed to phonetic

798 factors such as ease of articulation and perceptual salience. He offers a similar

799 explanation for the sporadic preservation of glottalization in stops for the oldest

800 generation of non-fluent speakers (represented by a woman in her 70s in Rankin’s study),

801 even though they completely lost glottalization in fricatives, which is considerably rarer

802 cross-linguistically. As English continued to make inroads, Quapaw’s phonology became

803 increasingly English-like, even at the expense of increasing markedness. Thus, the

804 youngest generation of speakers in his study had completely adopted the allophonic

805 distribution of aspiration found in English.

806 Although Rankin’s study clearly demonstrates the profound nature of

807 phonological change among speakers not using a language regularly, it leaves open the

808 relative contribution of external factors such as language contact as opposed to internal

809 pressures such as markedness. Much of the uncertainty arising in the interpretation of

810 Rankin’s results stems from the inherent elusiveness of the notion of markedness, which

811 in the case of the Quapaw data could be attributed to frequency or phonetic naturalness.

812 For example, a statistical predominance of unaspirated stops in Quapaw relative to their

813 aspirated counterparts could have led to their adoption by speakers neutralizing the two

814 series. Alternatively, the preservation of the unaspirated stops might have been attributed

815 to their greater phonetic “naturalness”. Investigation of the source of the aspiration

816 patterns in Quapaw would also have been aided by a phonetic study to determine whether

817 voice-onset-time was in fact influenced on a low level by stress as in English. 44

818 Haynes (2010) attempts to tease apart the many factors, both linguistic and

819 cultural, driving language change in a detailed production and perception study of

820 phonetic characteristics of Numu (Oregon Northern Paiute) among four populations

821 living in or near the Warm Springs, Oregon tribal community. The first group included

822 fluent speakers of Numu from Warm Springs. The second group consisted of members of

823 the Warm Springs community who were not fluent but had experienced direct exposure

824 to Numu through family members or classes. The third population included Warm Spring

825 community members with only ambient exposure to Numu, though some had exposure to

826 another (genetically unrelated) Native American language spoken in the Warm Springs

827 community. The fourth group consisted of residents of the nearby town of Madras who

828 reported no exposure to Numu.

829 Haynes (2010) examines a number of phonological and phonetic characteristics of

830 Numu as reproduced by non-fluent speakers mimicking a list of words uttered by fluent

831 speakers in order to examine the extent to which populations with differing degrees of

832 exposure to Numu are able to replicate features absent from English and to determine the

833 source of discrepancies between the three groups in their performance. Her results

834 suggest a number of linguistic and social factors at work.

835 As expected, there was considerable variation among non-fluent speakers even

836 those belonging to the same broad categories adopted by Haynes. Furthermore, intertoken

837 variation was also observed even within speakers, a result that mirrors data from

838 Rankin’s (1978) study. Also in keeping with Rankin (1978), Haynes (2010) finds that

839 many sounds and contrasts that are not found in English are either replaced or 45

840 phonetically altered in a direction that is predictable from English influence. In some

841 cases, exposure to Numu confers some advantage in the ability to replicate native speaker

842 pronunciation. For example, the initial affricate /ts/ in Numu is preserved in gradient

843 fashion in proportion to a speaker’s level of exposure to Numu. Similarly, in a

844 comparison of fortis and lenis stops, which differ from each other along a number of

845 dimensions including voice-onset-time, duration, and burst intensity, Haynes (2010) finds

846 that non-fluent speakers exaggerate VOT differences but reduce durational differences

847 between the two stop series in proportion to their degree of exposure to Numu, where the

848 group with the most exposure to Numu is most accurate in its phonetic reproduction of

849 the fortis vs. lenis contrast. Only the non-fluent group with the greatest exposure to Numu

850 is able to reproduce the burst intensity distinction between the two sets of stops.

851 In other cases, there is no clear advantage associated with increased exposure to

852 Numu. For example, the uvularization of /k/ before the vowels /a/ and /ɔ/ is not reliably

853 produced by non-fluent speakers regardless of their level of experience to Numu. All

854 non-fluent groups also decrease the duration ratio between intervocalic geminate and

855 singleton nasals relative to the fluent speakers, primarily due to lengthening of the

856 singleton series, a result that Haynes (2010) speculates is due to the lack of a gemination

857 contrast in English. Similarly, influence of English likely contributes to the compacted

858 vowel space of Numu relative to that of non-fluent speakers, especially in the central and

859 back space, where Numu contrasts /ɨ/, /u/, /ɔ/, and /a/.

860 Haynes (2010) also observes evidence of hypercorrection in her data. Non-fluent

861 speakers from the two Warm Springs groups but not the Madras group thus often apply 46

862 final vowel devoicing even in contexts where fluent speakers do not, which Haynes

863 suggests is attributed to devoicing being identified as a distinguishing characteristic of

864 Numu that is being overapplied by non-native speakers. Another interesting finding is the

865 spontaneous production of ejectives by participants belonging to both Warm Springs

866 groups even though Numu lacks ejectives. Haynes (2010) hypothesizes that the ejective

867 productions are due to the occurrence of ejectives in two other Native American

868 languages spoken in Warm Springs, Kiksht and Ichishkin, which has lead non-fluent

869 Numu speakers to associates ejectives with Native American languages, even those like

870 Numu that lack them. She suggests that this type of overgeneralization potentially offers

871 an explanation for the common linguistic phenomenon of areal spreading of features,

872 including ejectives, an areal feature of Northwest Native American languages.

873 In a companion perception study in Haynes (2010), tokens produced by non-

874 fluent speakers and containing non-native features of Numu were played to two native

875 speaker listeners to ascertain which properties contribute most to a perceived non-native

876 accent. There was considerable variation between the two speakers in their tolerance of

877 non-native features and their sensitivity to different features as markers of a non-native

878 accent. Some of the more robust findings, however, were that deviations from the native

879 devoicing patterns triggered relatively strong non-native judgments, as did non-native

880 productions of the fortis vs. lenis contrast. Interestingly, however, differences in VOT, on

881 which non-native speakers relied in producing the contrast, did not strongly induce non-

882 native judgments in the perception task. Ejectives only exerted an effect on judgments for

883 one of the two listeners. Social factors such as the individual and the gender of the 47

884 speaker (males perceived as less native-like) also impacted perceptions of native

885 pronunciation.

886 Haynes’ (2010) results highlight the considerable complexity of the mapping

887 between speech characteristics of native speakers of Native American languages and

888 those with varying degrees of proficiency. Besides the expected influence of a socially

889 dominant language like English the cultural association of particular properties with a

890 language, whether actually part of that language or not, may exert an impact on the

891 production of non-native speakers. Her results also indicate that properties that appear to

892 be most divergent between native and non-native speakers in production may not

893 necessarily be those that listeners identify as critical to perceived linguistic proficiency on

894 the part of the listener. As Haynes (p. 128) points out, this result suggests that “Numu

895 [language] (and other endangered languages in similar social contexts) may continue to

896 be an authentic marker of Numu culture, even if it incorporates some of the features of

897 second language learner speech”.

898

899 6. Prospects for the future

900

901 The scope and volume of recent studies attest to the vitality of the research program

902 devoted to the phonetics and phonology of Native American languages. This is

903 demonstrated not only by work published in IJAL, but also by articles appearing in

904 diverse journals and by recent edited books focused on phonetic and phonological topics

905 (e.g. Hargus and Rice 2005, Goodwin Gómez and Van der Voort 2014, Avelino et al. 48

906 2016). Concomitant with the expanded range of research, there has been a geographic

907 burgeoning in the form of vastly increased work on the phonology and phonetics of

908 Central and South American languages. This trend is amply reflected in the pages of

909 IJAL over the last decade (e.g. Hintz 2006, Malone 2006; 2010, Stenzel 2007, Elías Ulloa

910 2009, Maddieson et al. 2009, Everett 2011, Dicanio 2012, Clopper and Tonhauser 2013,

911 Michael et al. 2013, Noyer 2013, Crowhurst and Trechter 2014, Caballero and Carroll

912 2015) and reflects tremendous progress from the state of affairs at the time Boas

913 (1917a:1) wrote in his introduction to the inaugural issue of IJAL “it is not saying too

914 much if we claim that for most of the native languages of Central and South America the

915 field is practically terra incognita”.

916

917 REFERENCES

918

919 ARCHANGELI, DIANA. 1984. Underspecification in Yawelmani Phonology and

920 Morphology. Ph.D. Dissertation, MIT (Published in 1988 by Garland Publishing,

921 New York).

922 AVELINO, HERIBERTO, MATT COLER, AND W. LEO WETZELS (eds.). 2016. Laryngeal

923 Features in the Languages of the Americas (Brill’s Studies in the Indigenous

924 Languages of the Americas) Boston: Brill.

925 BAGEMIHL, BRUCE. 1991. Syllable structure in Bella Coola. 22:589–

926 646. 49

927 BECK, DAVID AND DAVID BENNETT. 2007. Extending the prosodic hierarchy: Evidence

928 from Lushootseed narrative. Northwest 1:1–34.

929 BECKMAN, MARY, JAN EDWARDS, AND JANET FLETCHER. 1992. Prosodic structure and

930 tempo in a sonority model of articulatory dynamics, in G. Docherty and D. R. Ladd

931 (eds.), Papers in Laboratory Phonology II: Gesture, Segment, Prosody, pp. 68–86.

932 Cambridge: Cambridge University Press.

933 BEREZ, ANDREA. 2011. Prosody as a genre-distinguishing feature in Ahtna: a quantitative

934 approach. 18:210–236.

935 BIRD, SONYA. 2011. The nature of laryngealization in St’át’imcets laryngealized

936 resonants. International Journal of American Linguistics 77:159–184.

937 BIRD, SONYA, MARION CALDECOTT, FIONA CAMPBELL, BRYAN GICK AND PATRICIA

938 SHAW. 2008. Oral-laryngeal timing in glottalized resonants.

939 36:492–507.

940 BOAS, FRANZ. 1911. Handbook of American Indian Languages, vol. 1. Washington:

941 Smithsonian Institution.

942 BOAS, FRANZ. 1917a. Introductory. International Journal of American Linguistics 1:1–8.

943 _____. 1917b. El Dialecto Mexicano de Pochutla, Oaxaca. International Journal of

944 American Linguistics 1:9-44.

945 BUCKLEY, EUGENE. 1994. Persistent and cumulative extrametricality in Kashaya. Natural

946 Language and Linguistic Theory 12:423–464.

947 _____. 2013. Kashaya extrametricality and formal symmetry. Supplemental 50

948 Proceedings of the 2013 Meeting on Phonology.

949 CABALLERO, GABRIELA AND LUCIEN CAROLL. 2015. Tone and stress in Choguita

950 Rarámuri (Tarahumara) word prosody. International Journal of American

951 Linguistics 81:457–493.

952 CAMPBELL, ERIC AND ANTHONY WOODBURY. 2010. The comparative tonology of

953 Chatino: A prolegomenon. Paper presented at the Society for the Study of the

954 Indigenous Languages of the Americas (SSILA), winter meeting. Baltimore,

955 Maryland, January 7-10.

956 CHAFE, WALLACE. 1976. The Caddoan, Iroquoian, and . The Hague:

957 Mouton.

958 CHO, TAEHONG AND PETER LADEFOGED. 1999. Variation and universals in VOT:

959 evidence from 18 languages. Journal of Phonetics 27:207–229.

960 CHO, TAEHONG, ALICE TAFF, MOSES DIRKS, AND PETER LADEFOGED. 1997. Some

961 phonetic structures of Aleut. UCLA Working Papers in Phonetics 95:68–90.

962 CLARK, MARY. 1983. On the distribution of contour tones. West Coast Conference on

963 Formal Linguistics 2:44–55.

964 CLOPPER, CYNTHIA G. AND JUDITH TONHAUSER. 2013. The prosody of focus in

965 Paraguayan Guaraní. International Journal of American Linguistics 79:219–251.

966 COOK, EUNG-DO. 1989. Is phonology going haywire in dying languages? Phonological

967 variations in Chipewyan and Sarcee. 18:235–55.

968 _____. 2006. The patterns of consonantal acquisition and change in Chipewyan (Dëne

969 Sųłiné). International Journal of American Linguistics 72:236–263. 51

970 CROWHURST, MEGAN J., AND LEV MICHAEL. 2005. Iterative footing and prominence-

971 driven stress in Nanti (Kampa). Language 81:47–95.

972 CROWHURST, MEGAN AND SARA TRECHTER. 2014. Vowel-rhotic metathesis in Guarayu.

973 International Journal of American Linguistics 80:127–173.

974 CROWHURST, MEGAN J., AND AMADOR TEODOCIO OLIVARES. 2014. Beyond the Iambic-

975 Trochaic Law: the joint influence of duration and intensity on the perception of

976 rhythmic speech. Phonology 31:51–94.

977 CRUZ, HILARIA, AND ANTHONY C. WOODBURY. 2006. La fonología y tonología

978 comparativa del Chatino: un informe de campo en Zacatepec. Proceedings of the

979 Conference on Indigenous Languages of Latin America II. Austin: AILLA.

980 DART, SARAH (1991). Articulatory and Acoustic Properties of Apical and Laminal

981 Articulations (UCLA Working Papers in Phonetics 79). Los Angeles: UCLA

982 Department of Linguistics.

983 _____. (1993). Phonetic properties of O'odham stop and fricative contrasts. International

984 Journal of American Linguistics 59:16–37.

985 DAY, CHRISTOPHER. 1973. The Jacaltec Language. Indiana University Publications,

986 Language Science Monographs 12). Bloomington: Indiana University Press.

987 DE JONG, KENNETH, AND JOYCE MCDONOUGH. 1993. Tone and tonogenesis in Navajo.

988 UCLA Working Papers in Phonetics 84:165–182.

989 DICANIO, CHRISTIAN T. 2012. The phonetics of fortis and lenis consonants in Itunyoso

990 Trique. International Journal of American Linguistics 78:239–272.

991 EDWARDS, JAN, MARY BECKMAN, AND JANET FLETCHER. 1991. The articulatory 52

992 kinematics of

993 ELÍAS-ULLOA, JOSÉ. 2009. The distribution of laryngeal segments in Capanahua.

994 International Journal of American Linguistics 75:159–206.

995 ESLING, JOHN H., KATHERINE E. FRASER, AND JIMMY G. HARRIS. 2005. Glottal stop,

996 glottalized resonants, and pharyngeal: a reinterpretation with evidence from a

997 laryngscopic study of Nuuchahnulth (Nootka). Journal of Phonetics 33:383–410.

998 EVERETT, CALEB. 2011. Variable velic movement in Karitiâna. International Journal of

999 American Linguistics 77:33–58.

1000 EVERETT, DANIEL. 1986. Pirahã, in Desmond Derbyshire and Geoffrey Pullum (eds.),

1001 Handbook of Amazonian Languages 1, 200–325. Berlin: Mouton de Gruyter.

1002 EVERETT, DANIEL, AND KEREN EVERETT. 1984. On the relevance of syllable onsets to

1003 stress placement. Linguistic Inquiry 15:705–711.

1004 FITZGERALD, COLLEEN M. 1998. The meter of Tohono O’odham songs. International

1005 Journal of American Linguistics 64:1–36.

1006 FLEMMING, EDWARD, PETER LADEFOGED, AND SARAH THOMASON. 1994. Phonetic

1007 structures of Montana Salish. UCLA Working Papers in Phonetics 87:1–33.

1008 GAFOS, ADAMANTIOS. 1999. The Articulatory Basis of Locality in Phonology. New York:

1009 Garland.

1010 GERFEN, CHIP, AND KIRK BAKER. 2005. The production and perception of laryngealized

1011 vowels in Coatzospan Mixtec. Journal of Phonetics 33:311–334. 53

1012 GICK, BRYAN, HEATHER BLISS, KARIN MICHELSON, AND BOSKO RADANOV. 2012.

1013 Articulation with acoustics: “Soundless” vowels in Oneida and Blackfoot. Journal of

1014 Phonetics 40:46–53.

1015 GODDARD, PLINY EARLE. 1907. The Phonology of the Hupa Language, Part I, The

1016 Individual Sounds. Berkeley, California: University of California Press.

1017 _____. 1912. Elements of the Kato Language. California University of California Press.

1018 _____. 1928. Pitch Accent in Hupa. Berkeley, California: University of California

1019 Press.

1020 GÓMEZ, GALE GOODWIN AND HEIN VAN DER VOORT (eds.). 2014. Reduplication in

1021 Indigenous Languages of South America. Boston: Brill.

1022 GORDON, MATTHEW. 1996. The phonetic structures of Hupa. UCLA Working Papers in

1023 Phonetics 93:164–187.

1024 _____. 2001. A typology of contour tone restrictions. Studies in Language 25:405–444.

1025 _____. 2004. A phonetic and phonological study of word-level stress in Chickasaw,

1026 International Journal of American Linguistics 70:1–32.

1027 _____. 2005a. An autosegmental/metrical model of Chickasaw intonation, in S-A. Jun

1028 (ed.), Prosodic typology—the phonology of intonation and phrasing, pp. 301–330.

1029 Oxford University Press.

1030 _____. 2005b. A perceptually-driven account of onset-sensitive stress. Natural

1031 Language and Linguistic Theory 23:595–653. 54

1032 GORDON, MATTHEW, AND PETER LADEFOGED. 2001. Phonation types: a cross-linguistic

1033 overview, Journal of Phonetics 29:383–406.

1034 GORDON, MATTHEW, AND ED LUNA. 2004. An intergenerational investigation of Hupa

1035 stress, Berkeley Linguistics Society 30:105–117.

1036 GORDON, MATTHEW, JACK MARTIN, AND LINDA LANGLEY. 2015. Some phonetic

1037 structures of Koasati. International Journal of American Linguistics 81, 83–118.

1038 GORDON, MATTHEW, AND PAMELA MUNRO. 2007. A phonetic study of final vowel

1039 lengthening in Chickasaw, International Journal of American Linguistics 73:293–

1040 330.

1041 GORDON, MATTHEW, PAMELA MUNRO, AND PETER LADEFOGED. 1997. The phonetic

1042 structures of Chickasaw. UCLA Working Papers in Phonetics 95:41–67.

1043 GORDON, MATTHEW, AND FRANÇOISE ROSE. 2006. Émérillon stress: a phonetic and

1044 phonological study. Anthropological Linguistics 48:132–168.

1045 GOUSKOVA, MARIA. 2003. Deriving Economy: Syncope in Optimality Theory. Ph.D.

1046 Dissertation, University of Massachusetts, Amherst. (Available online on the

1047 Rutgers Optimality Archive, ROA 610, http://roa.rutgers.edu)

1048 _____. 2007. The reduplicative template in Tonkawa. Phonology 24:367–396.

1049 GUREVICH, NAOMI. 2000. Reduplication in Southern Paiute and Correspondence Theory.

1050 West Coast Conference on Formal Linguistics 19:167–177.

1051 HAAS, MARY. 1977. Tonal accent in Creek, in Larry Hyman (ed.), Studies in Stress and

1052 Accent (Southern California Occasional Papers in Linguistics 4), pp. 195–208. Los 55

1053 Angeles: University of Southern California. [Reprinted in Sturtevant, William C.

1054 (ed.), 1987, A Creek source book. New York: Garland Publishing.]

1055 HALLE, MORRIS, AND VERGNAUD, JEAN-ROGER. 1987. An Essay on Stress. Cambridge,

1056 Mass.: MIT Press.

1057 HANSSON, GUNNAR ÓLAFUR. 2001. The phonologization of production constraints:

1058 Evidence from consonant harmony. Chicago Linguistic Society 37:187–200.

1059 HARGUS, SHARON, AND VIRGINIA BEAVERT. 2006. Word-initial clusters and minimality in

1060 Yakima Sahaptin. Phonology 23:21–58.

1061 HARGUS, SHARON AND KEREN RICE (eds.). 2005. Athabaskan Prosody. Philadelphia: John

1062 Benjamins Publishing Company.

1063 HAUGEN, JASON. 2003. Allomorphy in Yaqui reduplication, in Studies in Uto-Aztecan,

1064 MIT Working Papers on Endangered and Less Familiar Languages 5. Cambridge,

1065 MA: MITWPL, 75-103.

1066 _____. 2014. Hiaki (Yaqui) reduplication redux: Morphological and prosodic targets in

1067 syllabic reduplication. International Journal of American Linguistics 80:507–532.

1068 HAYES, BRUCE. 1987. A revised parametric metrical theory. Northeastern Linguistics

1069 Society 17:274–289.

1070 _____. 1989. Compensatory lengthening in moraic phonology. Linguistic Inquiry

1071 20:253-306.

1072 _____. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago: University

1073 of Chicago Press. 56

1074 HAYNES, ERIN. 2010. Phonetic and Phonological Acquisition in Endangered Languages

1075 Learned by Adults: A Case Study of Numu (Oregon Northern Paiute). Ph.D. thesis,

1076 University of California, Berkeley.

1077 HICKERSON, NANCY. 1958a. An acoustic analysis of Shawnee: I. International Journal of

1078 American Linguistics 24: 20–29.

1079 _____. 1958b. An acoustic analysis of Shawnee: II. International Journal of American

1080 Linguistics 24: 130–141.

1081 _____. 1959a. An acoustic analysis of Shawnee: III. International Journal of American

1082 Linguistics 25: 22–31.

1083 _____. 1959b. An acoustic analysis of Shawnee: IV. International Journal of American

1084 Linguistics 25: 97–104.

1085 HINTON, LEANNE. 1984. Havasupai Songs: A Linguistic Perspective. Tübingen: Gunter

1086 Narr Verlag.

1087 _____. 1990. Song metrics. Linguistics Society 16, Parasession on Native Languages.

1088 Berkeley: Department of Linguistics, University of California.

1089 HINTZ, DIANE. 2006. Stress in South Conchucos Quechua: a phonetic and phonological

1090 study. International Journal of American Linguistics 72:477–521.

1091 HOCKEY, BETH ANN, AND ZSUZSANNA FAGYAL. 1999. Phonemic length and pre-boundary

1092 lengthening: An experimental investigation on the use of durational cues in

1093 Hungarian. Proceedings of the Fourteenth International Congress of Phonetic

1094 Sciences, pp. 313–16. 57

1095 HOGAN, JOHN T. 1976. An analysis of the temporal features of ejective consonants.

1096 33:275–284.

1097 HOIJER, HARRY. 1933. Tonkawa: An Indian language of Texas. New York: Columbia

1098 University.

1099 HOWE, DARIN, AND DOUGLAS PULLEYBLANK. 2001. Patterns and timing of glottalisation.

1100 Phonology 18:45-80.

1101 HUDU, FUSHEINI. 2008. The low vowel and retraction in St’át’imcets: An ultrasound

1102 investigation. SKY Journal of Linguistics 21:67–81.

1103 HYMAN, LARRY. 1988. Syllable structure constraints on tonal contours. Linguistique

1104 Africaine 1:49-60.

1105 JOHNSON, KEITH, AND JACK MARTIN. 2001. Acoustic vowel reduction in Creek: Effects of

1106 distinctive length and position in the word. Phonetica 58:81–102.

1107 KENSTOWICZ, MICHAEL, AND CHARLES KISSEBERTH. 1979. Generative Phonology:

1108 Description and Theory. New York: Academic Press.

1109 KIMBALL, GEOFFREY. 1983. Verb pluralization in Koasati. 1982 Mid-America Linguistics

1110 Conference Papers: 401–411.

1111 KINGSTON, JOHN. 2005. The phonetics of Athabaskan tonogenesis, in Sharon Hargus and

1112 Keren Rice (eds.), Athabaskan Prosody, pp. 137–184. Amsterdam: John Benjamins.

1113 KLATT, DENNIS H. 1975. Vowel lengthening is syntactically determined in a connected

1114 discourse. Journal of Phonetics 3:129–40.

1115 LADEFOGED, PETER, AND DANIEL EVERETT. 1996. Phonetic structures of Banawá, an

1116 endangered language. UCLA Working Papers in Phonetics 93:47–64. 58

1117 LINDAU, MONA (1984). Phonetic differences in glottalic consonants. Journal of

1118 Phonetics, 12:147–155.

1119 LOVICK, OLGA, AND SIRI TUTTLE. 2012. The Prosody of Dena’ina narrative discourse.

1120 International Journal of American Linguistics 78:293–334.

1121 MACEACHERN, MARGARET, BARBARA KERN, AND PETER LADEFOGED. 1996. Wari’

1122 phonetic structures. UCLA Working Papers in Phonetics 93:65–93.

1123 MADDIESON, IAN. 1984. Patterns of Sounds. Cambridge: Cambridge University Press.

1124 MADDIESON, IAN, HERIBERTO AVELINO, AND LORETTA O’CONNOR. 2009. The phonetic

1125 structures of Oaxaca Chontal. International Journal of American Linguistics 75:69–

1126 101.

1127 MADDIESON, IAN, NICOLA BESSEL, AND CAROLINE SMITH. 1996. A preliminary report on

1128 the sounds of Tlingit. UCLA Working Papers in Phonetics 93:125–148.

1129 MADDIESON, IAN, AND KRISTIN PRECODA. 1990. Updating UPSID. UCLA Working

1130 Papers in Phonetics 74:104–111.

1131 MALONE, TERRY. 2006. Tone and Syllable Structure in Chimila. International Journal of

1132 American Linguistics 72:1–58.

1133 _____. 2010. Nasality and nasal prosody in Chimila. International Journal of American

1134 Linguistics 76:1–41.

1135 MARTIN, JACK. 1988. Subtractive morphology as dissociation. West Coast Conference on

1136 Formal Linguistics 7:229–240.

1137 _____. 2011. A grammar of Creek (Muskogee). Lincoln: University of Nebraska Press. 59

1138 _____. 2013. The geminating grade in Koasati and Muskogean. Paper presented at the

1139 annual SSILA meeting, Boston, Mass.

1140 MARTIN, JACK, AND KEITH JOHNSON. 2002. An acoustic study of "tonal accent" in Creek.

1141 International Journal of American Linguistics 68:28–50.

1142 MCCARTHY, JOHN. 2007. Consonant harmony via correspondence: evidence from

1143 Chumash. Linguistics Department Faculty Publication Series. Paper 31. (available at

1144 http://scholarworks.umass.edu/linguist_faculty_pubs/31)

1145 MCCARTHY, JOHN, AND PRINCE, ALAN. 1986. Prosodic Morphology. Ms. University of

1146 Massachusetts, Amherst & Rutgers University. (Published 1996. Linguistics

1147 Department Faculty Publication Series. Paper 13.

1148 http://scholarworks.umass.edu/linguist_faculty_pubs/13)

1149 _____. 1995. Faithfulness and reduplicative identity, in J. Beckman, S. Urbanczyk, and

1150 L. W. Dickey (eds.), University of Massachusetts Occasional Papers in Linguistics

1151 18: Papers in Optimality Theory. Amherst, MA: Graduate Linguistics Students

1152 Association, 249–384. (Available online on the Rutgers Optimality Archive, ROA-

1153 60, www.roa.rutgers.edu)

1154 MCDONOUGH, JOYCE, AND MARTHA AUSTIN-GARRISON. 1994. Vowel enhancement and

1155 dispersion in the vowel space of Western Navajo: a study of Navajo speakers. UCLA

1156 Working Papers in Phonetics 87:93–104.

1157 MCDONOUGH, JOYCE, AND PETER LADEFOGED. 1993. Navajo stops. UCLA Working

1158 Papers in Phonetics 84:151–164. 60

1159 MCDONOUGH, JOYCE, PETER LADEFOGED, AND HELEN GEORGE. 1993. Navajo vowels and

1160 phonetic universal tendencies. UCLA Working Papers in Phonetics 84:143–150.

1161 MCDOWELL, RAMONA. 2004. Retraction in Montana Salish Lateral Consonants: An

1162 Ultrasonic Study. M.A. thesis, University of .

1163 MICHAEL, LEV, STEPHANIE FARMER, GREGORY FINLEY, CHRISTINE BEIER, AND KARIAN

1164 SULLÓN ACOSTA. 2013. A sketch of Muniche segmental and prosodic phonology.

1165 International Journal of American Linguistics 79:307–347.

1166 NAMDARAN, NAHAL. 2006. Retraction in St’at’imcets:́ An Ultrasonic Investigation. M.A.

1167 thesis, University of British Columbia.

1168 MUNRO, PAMELA. 1996. The Chickasaw sound system. Ms., University of California, Los

1169 Angeles.

1170 MUNRO, PAMELA, AND CHARLES ULRICH. 1984. Structure-preservation and Western

1171 Muskogean rhythmic lengthening. West Coast Conference on Formal Linguistics

1172 3:191-202.

1173 MUNRO, PAMELA, AND CATHERINE WILLMOND. 1994. Chickasaw: An Analytical

1174 Dictionary. Norman: University of Oklahoma Press.

1175 _____. 2008. Let’s Speak Chickasaw: Chikashshanompa’ Kilanompoli’. Norman,

1176 Oklahoma: University of Oklahoma Press.

1177 NAGANO-MADSEN, YASUKO. 1992. Mora and Prosodic Coordination: A Phonetic Study

1178 of Japanese, Eskimo, and Yoruba. Lund: Lund University Press 61

1179 NATER, H. F. 1984. The Bella Coola Language, National Museum of Man Mercury

1180 Series, Canadian Ethnology Service Paper No. 92. Ottawa: National Museums of

1181 .

1182 NEWMAN, STANLEY. 1944. of California. New York: Viking Fund.

1183 NOSKE, ROLAND. 1993. A Theory of Syllabification and Segmental Alternation.With

1184 Studies on the Phonology of French, German, Tonkawa and Yawelmani. Tübingen:

1185 Niemeyer.

1186 NOYER, ROLF. 2013. A generative phonology of San Mateo Huave. International Journal

1187 of American Linguistics 79:1–60.

1188 OSWALT, ROBERT L. 1988. The floating accent of Kashaya, in William Shipley (ed.), In

1189 Honor of , pp. 611-621. Mouton de Gruyter, Berlin.

1190 PAYNE, DORIS. 1990. Asheninca stress patterns. In Doris Payne (ed.), Amazonian

1191 Linguistics, pp. 185–212. Austin: University of Texas Press.

1192 PENG, SHU-HUI. 1997. Production and perception of Taiwanese tones in different tonal

1193 and phrasal contexts. Journal of Phonetics 25:371–400.

1194 PHELPS, ELAINE. 1975. Iteration and disjunctive domains in phonology. Linguistic

1195 Analysis 1:137–172.

1196 PLAUCHÉ, MADELAINE C., ROSEMARY BEAM DE AZCONA, RUNGPAT ROENGPITYA, AND

1197 WILLIAM F. WEIGEL. 1998. Glottalized sonorants: a phonetic universal? Berkeley

1198 Linguistics Society 24:381–389.

1199 POSER, WILLIAM. 1990. Evidence for foot structure in Japanese. Language 66:78–105. 62

1200 _____. 2004. On the status of Chumash sibilant harmony. Ms., University of

1201 Pennsylvania, Philadelphia (Available online at

1202 http://www.ai.mit.edu/projects/dm/featgeom/poser-chumashnew.pdf).

1203 POTTER, BRIAN, MATTHEW GORDON, JOHN DAWSON, WILLEM DE REUSE, AND PETER

1204 LADEFOGED. 2000. Phonetic structures of Western Apache. UCLA Working Papers

1205 in Phonetics 98:10–38.

1206 PRIETO, PILAR, JAN VAN SANTEN, AND JULIA HIRSCHBERG. 1995. Tonal alignment

1207 patterns in Spanish. Journal of Phonetics 23:429–451.

1208 PRINCE, ALAN, AND SMOLENSKY, PAUL. 1993. Optimality Theory: Constraint Interaction

1209 in . Ms. Rutgers University and University of Colorado at

1210 Boulder. (Published in 2004 Blackwell Publishers, Malden Mass.)

1211 RAIMY, ERIC. 2000. The Phonology and Morphology of Reduplication. Berlin: Mouton de

1212 Gruyter.

1213 RANKIN, ROBERT. 1978. The unmarking of Quapaw: A case study in language death.

1214 Kansas Working Papers in Linguistics 3:45–52.

1215 RUSSELL, KEVIN. 2008. Sandhi in Plains Cree. Journal of Phonetics 36:450–464.

1216 SAPIR, EDWARD. 1912. The Takelma Language of Southwestern Oregon. Washington:

1217 Government Printing Office.

1218 _____. 1930. Southern Paiute: a Shoshonean Language. Proceedings of the American

1219 Society of Arts and Sciences 65.

1220 SAPIR, EDWARD, AND VICTOR GOLLA. 2001. Hupa texts, with notes and lexicon. In Victor 63

1221 Golla and Sean O’Neill (eds.), Northwest California Linguistics [The Collected

1222 Works of Edward Sapir XIV], pp. 19–1011. New York: Mouton.

1223 SHAW, PATRICIA A. 2005. Non-adjacency in reduplication, in Bernhard Hurch (ed.),

1224 Studies on reduplication, pp. 161–210. New York: Mouton de Gruyter.

1225 SHAW, PATRICIA A., FIONA CAMPBELL, FLORA EHRHARDT, AND PATRICIA MCKAY. 2005.

1226 Patterns and timing of resonant glottalization in Nɬeʔkepmxcin. Proceedings of the

1227 40th International Conference on Salish and Neighbouring Languages, UBC

1228 Working Papers in Linguistics 16:210–224.

1229 SILVERMAN, DANIEL, BARBARA BLANKENSHIP, PAUL KIRK, AND PETER LADEFOGED.

1230 1994. Phonetic stuctures in Jalapa Mazatec. UCLA Working Papers in Phonetics

1231 87:113–130.

1232 SILVERMAN, KIM, AND JANET PIERREHUMBERT. 1990. The timing of prenuclear high

1233 accents in English, in John Kingston and Mary Beckman (eds.), Papers in

1234 Laboratory Phonology 1: Between the Grammar and Physics of Speech, pp. 72–106.

1235 Cambridge: Cambridge University Press.

1236 STENZEL, KRISTINE. 2007. Glottalization and other suprasegmental features in Wanano.

1237 International Journal of American Linguistics 73:331–366.

1238 STERIADE, DONCA. 1999. Phonetics in phonology: the case of laryngeal neutralization, in

1239 Matthew K. Gordon (ed.), UCLA Working Papers in Linguistics, vol. 2, October

1240 1999, Papers in Phonology 3:25–146.

1241 SWANTON, JOHN. 1921. The Tunica language. International Journal of American

1242 Linguistics 2:1–39. 64

1243 TOPINTZI, NINA. 2004. Moraic onsets and WSP partition in Pirahã. CamLing 2004:211–

1244 218.

1245 _____. 2010. Onsets: Suprasegmental and Prosodic Behaviour. Cambridge:

1246 Cambridge University Press.

1247 TUTTLE, SIRI. 2000. Unaspirated coronal stops in Jicarilla Apache. UCLA Working

1248 Papers in Phonetics 98:39–53.

1249 UCHIHARA, HIROTO. 2013. Tone and Accent in Oklahoma Cherokee. Ph.D. dissertation,

1250 The University at Buffalo, State University of New York.

1251 UMEDA, NORIKO. 1975. Vowel duration in American English. Journal of the Acoustical

1252 Society of America 58:434–45.

1253 VAN EIJK, JAN P. 1997. The Language: Phonology, Morphology, and Syntax.

1254 Vancouver: University of British Columbia Press.

1255 VILLARD, STÉPHANIE. 2009. Grammatical Sketch of Zacatepec Chatino. M.A. thesis,

1256 University.

1257 WIGHTMAN, COLIN, STEFANIE SHATTUCK-HUFNAGEL, MARIE OSTENDORF, AND PATTI

1258 PRICE. 1992. Segmental durations in the vicinity of prosodic phrase boundaries.

1259 Journal of the Acoustical Society of America 92:1707–17.

1260 WOO, NANCY. 1969. Prosody and Phonology. Ph.D. dissertation, MIT.

1261 WRIGHT, RICHARD. 1996. Tone and Accent in Oklahoma Cherokee. In Pamela Munro

1262 (ed.), Cherokee papers from UCLA. (University of California Occasional Papers in

1263 Linguistics 16). Los Angeles: U. of California, Los Angeles, pp. 11–22.

1264 WRIGHT, RICHARD, SHARON HARGUS, AND KATHARINE DAVIS. 2002. On the 65

1265 categorization of ejectives: data from Witsuwit’en. Journal of the International

1266 Phonetic Association 32:43–77.

1267 XU, YI. 2001. Fundamental frequency peak delay in Mandarin. Phonetica 58:26–52.

1268 ZHANG, JIE. 2002. The Effects of Duration and Sonority on Contour Tone Distribution:

1269 Typological Survey and Formal Analysis. New York: Routledge.

1 Thank you to David Beck, Donna Gerdts, and an anonymous reviewer for their

numerous thoughtful and helpful comments on earlier versions of this paper.