Coronal obstruents

The dental and alveolar plosives are distinguished by both visual and acoustic characteristics as determined by spectrographic evidence and discussion with our speaker about the location and contact of articulators. The dental stop is laminal, made with the tip of the tongue against the upper teeth and the blade of the tongue resting against the alveolar ridge. The tongue tip may be seen interdentally during the articulation of the dental stop, especially when word-final and unreleased (e.g. maat /maːt̪/ [maːt̪] ‘body’). The alveolar stop is apical, made with tongue tip contact against the center of the alveolar ridge. Discussing the difference in tongue placement with our speaker, the tongue tip was seen curved upward during the articulation of the alveolar stop, such that the body of the tongue is concave in shape. The apico-alveolar stop may vary in its anterio-posterior placement on the alveolar ridge, sometimes having a characteristically retroflex sound. The speaker recorded for this Illustration consistently contrasts a lamino-dental stop with an apico-alveolar stop, suggesting that both place of articulation and primary articulator play a role in defining the anterior stop contrast. To simplify the transcription of the contrast in the consonant chart above, the lamino-dental stop is represented with the dental diacritic as /t̪/, and the apico-alveolar stop is represented without diacritic as /t/.Footnote ²

Lamino-dental and apico-alveolar stops are acoustically distinguishable by formant transitions and burst spectra. Pre-vocalically, the laminodental stop has a noisier burst spectrum than the apico-alveolar stop, and the second formant transition into the following vowel enters at a lower frequency (Figure 2). A noisier burst spectrum is consistent with a noisier laminal release, while the brevity of release and louder energy in the mid-frequency range (2–6 kHz) of the apico-alveolar spectrum are consistent with the relative quickness of an apical gesture (Figure 3) (Hamann Reference Hamann2003). Post-vocalically, formant transitions also distinguish these stops (Figure 4). While the second formant transition out of the vowel is similar for the lamino-dental and apico-alveolar stop, the third and fourth formants are lower preceding the apico-alveolar than the lamino-dental. Lowered third and fourth formants are consistent with an apical closure produced behind the alveolar ridge, and indistinguishable second formants are consistent with the acoustic complexities of place and surface area of contact: laminality raises F2, but dentality lowers F2; apicality lowers F2, but a sublingual cavity raises F2 (Hamann Reference Hamann2003).

Figure 2 Spectrograms showing pre-vocalic /t̪/ in /t̪a’pin/ ‘heat’ (left) vs. /t/ in /tapʃ/ ‘flower’ (right).

Figure 3 Spectral slices taken from the durational mid-point of the plosive release show differences in pre-vocalic /t̪/ vs. /t/.

Figure 4 Spectrograms showing post-vocalic /t̪/ in /maːt̪/ ‘body’ vs. /t/ in /mat/ ‘earth’.

In addition to the dental-alveolar place contrast, this language variety also contrasts plain and palatalized alveolar stops in word-final position. In Figure 5, we see F1 decrease and F2 increase into the alveolar closure, suggesting pre-palatalization. While palatalized laterals are reconstructed for Proto-Yuman (Langdon Reference Epstein and Langdon1996), palatalized coronal stops are proposed to have developed from alveolo-palatal affricates in post-tonic position (Miller Reference Miller2016b). Consistent with this observation, we have documented palatalized laterals word-initially and inter-vocalically, but palatalized coronal plosives only word-finally, leading us to posit them as marginal in the phonological system given their restricted distribution.

Figure 5 Spectrogram showing post-vocalic /t^j/ in the word /mat^j/ ‘you’.

As seen in Figures 4 and 5, the speaker in this recording does not release word-final stops. This pattern is innovative among Kumiai varieties, which are described as having clearly released stops with variation between aspirated and non-aspirated allophones in word-final position (Langdon Reference Langdon1970). To the best of our knowledge, word-final unreleased stops have not been reported in any other Kumiai variety. Though a systematic investigation controlling for prosodic contexts has not yet been undertaken, final unreleased stops are documented in a variety of utterance contexts, suggesting that the effect pertains to the word level.

A contrast between a postalveolar affricate and a postalveolar fricative is documented in several dialects of Kumiai (e.g. Jamul Tiipay, Miller Reference Miller2001), including Ja’a Kumiai (Miller Reference Miller2016b). According to Miller (Reference Miller2016b), however, some speakers appear to have neutralized the contrast in this variety. This is the case for our consultant, who exhibits free variation between a postalveolar fricative and a postalveolar affricate, except in post-consonantal environments, where only the fricative production is attested. Figure 6 exemplifies the free variation between the fricative and affricate realizations, with the word for ‘one’ realized as both [t͡ʃin] and [ʃin].

Figure 6 Spectrograms showing variable production of /ʃin/ ‘one’ with /ʃ/ vs. /t͡ʃ/.

The postalveolar fricative in Ja’a Kumiai is reported to be variably realized as laminal postalveolar [ʃ] or retroflex [ʂ] (Miller Reference Miller2016b). (Miller (Reference Miller2001, Reference Miller2016a, Reference Millerb) represents this fricative with the non-IPA symbol [ṣ̌], and describes a third ‘hybrid’ phonetic realization between the laminal and apical productions (Miller Reference Miller2016b). This fricative is posited to derive historically from Proto-Southern-Kumeyaay */ʂ/ (Miller Reference Miller2016b), and is described for the Mesa Grande variety by Langdon (Reference Langdon1970: 30) as an ‘alveolar or post-alveolar [fricative], pronounced with great tenseness, with the apex almost touching the alveolar ridge’. Miller (Reference Miller2016b) proposes that changes underway in Ja’a Kumiai involve the neutralization of */t͡ʃ/ and */ʂ/ in pre-tonic syllables, with /ʂ/ realized variably as [ʃ] or [ʂ]. Under this account, the neutralization involves both fortition of */ʂ/ and lenition of */t͡ʃ/. While there is no evidence that our consultant has a contrast between [ʃ] and [t͡ʃ], we include them both in the phonemic chart above since they are reported to be contrastive for at least some speakers of Ja’a Kumiai (Miller Reference Miller2016b).

Post-alveolar affricates are both acoustically and morphophonologically distinct from stop–fricative clusters in the speech of our speaker. This distinction is exemplified in the spectrograms in Figure 7 with the near-minimal pair /t͡ʃou/ ‘to build’, which contains a word-initial affricate, and /tʃok/ ‘to clean’, which contains a word-initial consonant cluster. As shown in these spectrograms, the affricate in /t͡ʃou/ is characterized acoustically by a shorter duration of the fricative and a shorter closing interval before the onset of the fricative than the stop–fricative cluster in /tʃok/. Phonologically, the affricate is differentiated from the stop–fricative cluster in contexts where the exponence of a derivational category, nominalizing /aʔ-/Footnote ³ is prefixed to plosive-initial bases (e.g. [a’nak] /aʔ-’nak/ ‘chair,’ derived from /nak/ ‘to sit’) and infixed to bases that have a word-initial consonant cluster ([xa’tup] /x-aʔ-’tup/ ‘trampoline,’ derived from /xtup/ ‘jump’). In this morphological context, word-initial affricates are treated as a unit by the nominalizing prefix (e.g. [a’t͡ʃau] /aʔ-’t͡ʃau/ ‘brick’ (lit.: ‘something to build’) but the plosive–fricative cluster is split through an infixation process (e.g. [t̪a’ʃok] /t-aʔ-’ʃok/ ‘cleaning rag’).

Figure 7 Spectrograms showing /tʃ/ in /tʃok/ ‘to clean’ vs. /t͡ʃ/ in /t͡ʃou/ ‘to build’.

Dorsal obstruents

The dorsal plosive of our Ja’a Kumiai speaker varies gradiently from velar to backed velar,Footnote ⁴ diverging from varieties of Kumiai in which a systematic distribution of distinctly velar and uvular plosives is found. In Jamul Tiipay, Miller (Reference Miller2001) reports a pattern of complementary distribution in uvular and velar plosives where the uvular plosive surfaces in stressed syllables following a non-front vowel. Similarly, Gil Burgoin (Reference Gil Burgoin2016) reports complementary distribution of these allophones in nearby San José de la Zorra Kumiai, where the uvular alternant surfaces post-tonically, often word-finally, and the velar allophone surfaces elsewhere. In Ja’a Kumiai, however, dorsal stops with velar and backed velar place occur in the same contexts. For example, /kur’ʔak/ [k̙uɾ’ʔak] ‘elderly man’ and /ku’tu/ [k̙u’tu] ‘to kick’ share acoustic properties of sounds articulated further back in the oral cavity than canonical velars, but are found in the same phonological context (pretonic followed by [u]) as plosives with canonical velar qualities such as small F3–F2, /ku’nil^j/ [ku’nil^j] ‘blacken’ and /ku’l^jak/ [ku’l^jak] ‘light weight’. Typical variation in the backness of the dorsal stop can be seen acoustically in the comparatively lower burst frequency, more turbulent release, and greater F3–F2 of /ku’tu/ in Figure 8 (left).

Figure 8 Spectrograms for /ku’tu/ ‘kick’ (left) and /ku’l^jak/ ‘light weight’ (right) show variation in /k/.

The dorsal fricative shows a similarly variable distribution as the dorsal stop, with slight variation in the backness of velar place. In addition, the dorsal fricative exhibits allophonic variation in terms of degree of frication, with a voiceless velar fricative [x], an approximant [x], and a glotal fricative [h] realization. Moderate frication is attested in word-initial contexts such as /xa ku’nil^j/ [xa ku’nil^j] ‘coffee’ and weaker constriction may be attested in obstruent clusters such as /t͡ʃxui/ [t͡ʃxui] ‘to annoint oneself with incense’ evidenced by its comparatively weaker energy in Figure 9 (center). Even weaker frication, lower energy, and broader spectral range is attested in words such as /pax/ [pah] ‘to return’ (Figure 9) and /xpʃu/ [hpʃu] ‘green’. The variable realization of the dorsal fricative is similar to varieties in which stress is reported to be a conditioning factor. In Jamul Tiipay the dorsal fricative is reported to be realized as a voiceless velar approximant [̥ɰ] post tonically, alternating between [x]~[̥ɰ] elsewhere (Miller Reference Miller2001). In San José de la Zorra Kumiai the fricative is reported to have a variable realization as velar [x], uvular [χ], or glottal [h] in all contexts (Gil Burgoin Reference Gil Burgoin2016: 52).

Figure 9 Spectrograms for /xa ku’nil^j/ ‘coffee’ (left), /t͡ʃxui/ ‘to annoint oneself with incense’ (middle), and /pax/ ‘return’ (right) show variation in /x/.

The glottal stop in Ja’a Kumiai may be produced canonically with complete occlusion of the airstream (see /t͡ʃaʔ’jou/ [t͡ʃaʔ’jou] ‘song’), as glottalization of an adjacent vowel (see /xaʔ’nak/ [χa̰’nak̙] ‘to put on a necklace’), and as glottalization of an adjacent sonorant (see /taʔ’nap/ [ta’n̰ap] ‘braid’). In these three examples, the glottal stop is part of the exponence of the nominalizing prefix /aʔ-/, which infixes with certain stems as described above. Glottalization may also be greatly reduced so that the only trace of its realization is a lowering of f0, e.g. /t͡ʃaʔ’jou/ ‘song’ is variably realized as [t͡ʃaʔ’jou] or [t͡ʃa’jou]. Figure 10 and Figure 11 demonstrate these three types of glottal stop realizations.

Figure 10 Spectrograms for /t͡ʃaʔ’jou/ ‘song’ show variable presence of /ʔ/.

Figure 11 Spectrograms showing variable realization of /ʔ/ as creaky voice on a vowel as [a̰] in /xaʔ’nak/ ‘necklace’ (left) and on a nasal as [n̰] in /taʔ’nap/ ‘braid’ (right).

Sonorants

The trill surfaces with three articulations, all of which surface in predictable contexts: a tap [ɾ], a trill [r], or with frication []. These variants may be devoiced in word-initial and word-final contexts. The trill is articulated as a tap [ɾ] intervocalically (e.g. /sa’rap/ [sa’ɾap] ‘five’) and in consonant clusters when flanked by consonants (e.g. /xprʃa/ [xpɾʃa] ‘sycamore’). Word-initially, this sound can be produced as a partially devoiced trill [r̥] (e.g. /rap/ [r̥ap] ‘pain’, /’rʔaːk/ [r̥ʔaːk̙] ‘elderly men’). The trill can also be realized with varying degrees of frication, voicing, and retroflexion in word-medial rime contexts (e.g. /ku’rʔak/ [k̙u’ʔak̙] ‘elderly man’ and /ka’sark/ ‘left’ [ka’saʃk]) but varies freely with a tap articulation (e.g. /t͡ʃir’kwi loi/ [t͡ʃɪɾ’kwi loi] ‘ant that lives in oak trees’ and /per’wi/ [pəɾ’wi] ‘dove’). Word-finally, this sound is generally devoiced (e.g. /aɬ^j’mar/ [aɬ^j’mar] ‘to light a fire’, /tiɲor/ [t̪iɲo] ‘to color’). Illustration of tap, trill, and fricative variants of /r/ can be seen in Figure 12.

Figure 12 Spectrograms showing realization of /r/ as tap in /sa’rap/ ‘five’ (left), trill in /rap/ ‘pain’ (center), and fricative in /ti’ɲor/ ‘color’ (right).

Additionally, in a pattern attested across the Delta-California branch of Yuman (Golla Reference Golla2011), the lateral approximant and voiceless lateral fricative exhibit a marginal contrast between plain and palatalized forms, with only a few documented examples of the palatalized kinds (see /xa’ɬ^jak/ ‘duck’ and /t͡ʃa’mel^j/ ‘older brother’ in the list of examples at the start of the section) which are attested in free variation in unstressed syllables for other speakers of Ja’a Kumiai (Miller Reference Miller2016b).

Monophthongs

Each of the five contrastive vowels in Ja’a Kumiai contrast in length in stressed environments. Long vowels are most commonly found in forms derived by length ablaut, a common derivational process in Ja’a Kumiai frequently exhibited in plural and nominalized forms, e.g. /ki’nus/ [ki’nus] ‘beautiful one’ vs. /ki’nuːs/ [ki’nuːs] ‘beautiful ones’. Long vowels in stressed syllables are slightly less centralized than their short vowel counterparts (see Figure 14), and the length contrast is not preserved in unstressed syllables of Ja’a Kumiai (Miller Reference Miller2016b), though it remains preserved in the unstressed syllables of related varieties, as in, for example, Mesa Grande (Langdon Reference Langdon1970). Mid vowels, /o/ and /e/, are innovations of Ja’a Kumiai, and it has been speculated that they have arisen due to contact with Spanish (Miller Reference Miller2016b). Related varieties (Jamul Tiipay and Los Conejos) have no mid vowels, and Mesa Grande and Campo have only back mid-vowel /o/ (Langdon Reference Langdon1970, Epstein & Langdon Reference Epstein and Langdon1996, Miller Reference Miller2001, Miller & Langdon Reference Miller and Langdon2008).

Unstressed vowels in word-initial position consistently preserve their quality (see the examples in the list of word-initial vowels above), though elsewhere unstressed vowels are frequently realized as [ə] or [ɪ], as /per’wi/ [pəɾ’wi] ‘dove’ and /si’ɲau nʔor/ [sɪ’ɲau nʔor] ‘acorn’ demonstrate, respectively. Excrescent [ə] or [ɪ] may also appear between consonants in complex clusters. These excrescent vowels are brief (~20 ms), appear unpredictably both within and across tokens, vary freely in quality when they are expressed, and do not participate in phonological processes which pertain to vowels. Yumanists refer to these excrescent, intrusive vowels as inorganic, in contrast to organic, phonemic vowels. Figure 13 illustrates the variable presence of an excrescent [ə] in the consonant cluster of /xpʃiu/ ‘green’ ([xpʃiu] or [xəpʃiu]).

Figure 13 Spectrograms for /xpʃiu/ ‘green’ showing variable presence of excrescent vowel [ə].

Diphthongs

Ja’a Kumiai has a rich set of closing diphthongs /ai ei oi ui au ou iu/, exemplified as follows:

As seen in Figure 14 articulation of final target /u/ and initial targets /i/ and /o/ are more central in diphthongs than in single-target monophthongs. The next section addresses the distribution of diphthongs across syllable types and provides detail about other basic aspects of syllable structure.

Figure 14 This plot approximates the vowel space of Ja’a Kumiai using F1 and F2. Shaded elipses indicate 68% confidence intervals on the tokens gathered for long and non-long monphthongs. Arrows indicate the directionality and endpoints of diphthong trajectories. The legend at bottom left shows diphthong–arrow color correspondences.

Broad phonemic transcription

txa ɲak ɲa ɲu:p ta’niu xwakɬ uwiu mʔei sper xan ‖ ɲa’pom kwamp ax’kei ma:t t̪u’pit tu’ju ‖ ɲa’pu:m u’wei mʔei kwamp t̪u’pit puɲu’t̪it ɲip ksper xan ‖ txa ɲak psui sper war ‖ t̪u’pit pu u’t̪ip xui ɲa psui sper xan kwampt kwam ʃa’win xan t̪u’pit ‖ txa ɲa:k an’mak ɲa ‖ ɲat ɲa’pu:m sper xan kwam t̪u’pit u’t̪ip ʃot ‖ ɲa’pu:m txat ɲak u’jau ɲat^j mar

Orthographic transcription

Ttja ñak ña ñuup ttaniu jwaklj uwiu m’ei sper jan, ñapom kwamp ajkei maatt tupitt ttuju. Ñapuum uwei m’ei kwamp tupitt puñutitt ñip ksper jan. ttja ñak psui sper war, tupitt pu utip jui ña psui sper jan kwampt kwam chawin jan tupitt. Ttja ñaak anmak ña, ñat ñapuum sper jan kwam tupitt utip chot. Ñapuum ttjatt ñak ujau ñath mar.

English translation

The North Wind and the Sun are fighting to see which of the two of them is stronger, when suddenly an elderly man wrapped in a thick cape passes by. They decide that whichever makes the man remove his cape is the strongest. The North Wind begins to blow with great force, but after blowing forcefully, the elderly man clings to his cape. The North Wind stops, [and] the Sun. The Sun shines brightly, and immediately the old man removes his cape. So the North Wind recognizes that the Sun is stronger.

Spanish translation

El viento del norte y el sol pelean para decidir cual era más fuerte, cuando de repente pasó un an anciano envuelto en una capa. Deciden en que aquel que primero pudiera hacer que el anciano se quitara la capa sería el más fuerte. Entonces el viento del norte comenzó a soplar con mucha fuerza, pero después de soplar mucho, el anciano se aferra a su capa. El viento del norte paró, [y] el sol. El sol brilla brillantemente, y el anciano se quita la capa inmediatamente. Entonces el viento del norte reconoce que el sol es más fuerte.