Sumi (also known by its exonym ‘Sema’) is a Tibeto-Burman language spoken in Nagaland, North-east India.Footnote 1 It is one of the major languages of the state, with an estimated 242,000 speakers living primarily in Zunheboto district, as well as in the major cities of Kohima and Dimapur. Bradley (Reference Bradley and Bradley1997) places Sumi (referred to as Sema), among the ‘Southern Naga’ languages, which include Angami (also known as Tenyidie) and Ao, in contrast to the ‘Northern Naga’ languages such as Konyak and Nocte. Burling (Reference Burling, Thurgood and LaPolla2003) offers a more conservative classification, placing Sumi (referred to as Simi) in an ‘Angami-Pochuri’ group containing Angami, Chakhesang (Chokri and Kheza) and Mao. Four main dialects of Sumi have been identified: the Western dialect, the Eastern dialect, the Chizolimi dialect, and the Central dialect. The Central dialect is the standard dialect used in published works of Sumi (Sreedhar Reference Sreedhar1976: 4–5).
The earliest published work on Sumi can be found in The Linguistic Survey of India edited by Sir George Grierson (see Grierson Reference Grierson1967). In 1916, John Hutton wrote the first sketch grammar of the language, ‘Rudimentary grammar of the Sema Naga language, with vocabulary’, a revised version of which is included in his published anthropological description The Sema Nagas (see Hutton Reference Hutton1968), along with a revised word list. Sreedhar (Reference Sreedhar1980) provides the most comprehensive grammar of Sumi to date, with a significant portion of A Sema Grammar dedicated to the phonology of the language. This grammar incorporates much of the data already presented in his Sema Phonetic Reader, published earlier, in 1976 (Sreedhar Reference Sreedhar1976). Although Hutton had previously identified three contrastives tones, Sreedhar's work represents the first attempt at transcribing tones for all lexical items. However, a number of words were inconsistently transcribed for tone, and many of his transcriptions differ greatly from those offered in the present Illustration, calling to question the reliability of some of his tonal transcriptions.
The orthography employed here is the same as that used in Sumi language publications, most notably the Bible and language textbooks. The creation of this Latin-based orthography is attributed to the missionary Rev. W. F. Dowd and Inashe Sema, who published a primer entitled Mlali in 1909 (Sreedhar Reference Sreedhar1976). This orthography has not been completely standardised, and tones are not always consistently marked – ‘h’ is sometimes added to the end of a syllable to indicate low tone, and a preceding consonant is sometimes doubled to indicate high tone. However, this practice is often applied at a writer's own discretion to words where it is felt that tone needs to be marked to avoid confusion.
This study was based on the speech of three speakers of the Central dialect: a 39-year-old female speaker from the Satakha area of Zunheboto district; a 38-year-old male speaker residing in the town of Zunheboto, and a 27-year-old male speaker who had spent significant portions of his childhood in both the Zunheboto town and Satakha areas. Minor differences were found to exist between the Zunheboto and Satakha dialects. The recorded text was provided by the female speaker.
Consonants
The inventory of consonant phonemes in Sumi is unusual among its neighbouring languages in that it includes a set of uvular stops and a set of velar fricatives. Sumi also lacks a phonemic contrast between alveolar/dental fricatives and postalveolar fricatives. The status of its coronal rhotic (the voiced alveolar approximant) is only marginal, in contrast to neighbouring languages such as Angami (Tenyidie) and Ao, which typically include both a voiced and a voiceless (or aspirated) coronal rhotic.
Stops, fricatives and affricates
Stops show a three-way voice-onset time contrast, with the exception of the uvular stops, which only show a contrast in aspiration. /qh/ is commonly realised as [qχ] with a fricated release, e.g. aqho /àqhò/ [a˩ qχo˩] ‘brain’, aqhi /àqhì/ [a˩ qχi˩] ‘moon’.
The post-alveolar fricatives /ʃ/ and /ʒ/ and post-alveolar affricates /ʧ/ and /ʧh/ are realised as the alveolar fricatives [s] and [z] and alveolar affricates [ts] and [tsh], respectively, before the central vowels /ɨ/ and /a/, e.g. ashi /àʃì/ [a˩ ʃi˩] ‘meat’ and asü /àʃ
/ [a˩ sɨ˩] ‘wood’, aji /àʒì/ [a˩ ʒi˩] ~ [a˩ ʤi˩] ‘blood’ and aza /àʒá/ [a˩ za˥] ‘mother’, akichi /àkìʧí/ [a˩ ki˩ ʧi˥] ‘mouth’ and akütsü /àkʧɨ/ [a˩ kɨ˩ tsɨ˧] ‘head’. [ʒ] is in free variation with [ʤ].Footnote 4
For speakers from the Zunheboto town area, the labial-velar approximant [w] is in complementary distribution with the labio-dental fricative [v], occurring only before the back rounded vowels /u/ and /o/, e.g. awu /àvù/ [a˩ wu˩] ‘chicken’, awo /àvò/ [a˩ wo˩] ‘pig’, but avi /avi/ [a˧ vi˧] (*[a˧ wi˧]) ‘mithun’ and avü /avɨ/ [a˧ vɨ˧] (*[a˧ wɨ˧]) ‘frost’/‘dew’. For speakers of the Satakha dialect, [v] occurs before all vowels, including the back rounded vowels e.g. awu /àvù/ [a˩ vu˩] ‘chicken’, awo /àvò/ [a˩ vo˩] ‘pig’.
Liquids, approximants and nasals
/m/ is realised as a syllabic nasal in a minor syllable (see section on syllable structure below) in word-initial position, where it is typically followed by a stop or lateral, e.g. mla/mlah /
là/ [
˩ la˩] ‘work’, muku /mùkú/ [˩ ku˥] ‘twenty’ ([mu˩ ku˥] only in very careful speech). /m/ is also realised as syllabic nasal when it bears a different tone from a preceding vowel (see section on tone below). The breathy sonorants /mɦnɦlɦ/ are realised with a breathy [ɦ] release, e.g. amhi /àmɦì/ [a˩ mɦi˩] ‘body hair’, anha /ànɦà/ [a˩ nɦa˩] ‘mucus’ (and nha /nɦa/ [nɦa˧] ‘to cover’), alha /àlɦà/ [a˩ lɦa˩] ‘layer’. A more comprehensive acoustic study of these breathy sonorants is presented in Harris (Reference Harris2009).
The alveolar approximant /ɹ/ is a marginal phoneme in the language, which Sreedhar (Reference Sreedhar1976, Reference Sreedhar1980) does not include in the phonemic inventory in his analysis. While it seems to occur only in borrowed words, its inclusion here is motivated by the fact that speakers are able to consistently place the pitch on words like murasü /mùɹáʃ
/ [mu˩ ɹa˥ sɨ˥] ~ [˩ ɹa˥ sɨ˥] ‘snow’/‘hail’ into tonal categories, suggesting that these words have been more ‘nativised’ than other loanwords. It has two allophones: the alveolar approximant [ɹ], which occurs in syllable-initial position, as in murasü; and the alveolar trill [r], which occurs in syllable-final position, e.g. khurshi /khùrʃì/ [khur˩ ʃi˩] ‘horse’.
Consonant palatalisation
Velar obstruents are realised as palatal obstruents before the front vowels /i/ and /e/, e.g. agi /agí/ [a˧ ɟi˥] ‘face’, aki /àkì/ [a˩ ci˩] ‘house’. The front vowels /i/ and /e/ also often trigger palatalisation of other preceding consonants, e.g. akive /àkìvé/ [a˩ ci˩ ve˥] ~ [a˩ ci˩ vje˥] ‘stomach’, aje /àʒè/ [a˩ ʒe˩] ~ [a˩ ʑe˩] ‘name’, ashi /àʃì/ [a˩ ʃi˩] ~ [a˩ ɕi˩] ‘meat’, axone /àxónè/ [a˩ xo˥ nje˩] ‘fermented soya beans’, ali /àlì/ [a˩ lji˩] ‘pot’.
Glottal stop
A glottal stop is sometimes inserted between two vowels at a word or morpheme boundary, such as between possessive prefixes and noun roots. It is often represented orthographically by a hyphen. This glottal stop can also be realised as creaky voice on the second vowel or deleted altogether, e.g. a-a /à-à/ [a˩ ʔa˩] ~ [a
˩] ~ [aa˩] ‘place’ (where a- is a non-relational prefix marker), i-a /ì-à/ [i˩ ʔa˩] ~ [i ˩] ~ [ia˩] ‘my place’.
A glottal stop is also sometimes added before a word-initial vowel when following another word. One of the male speakers often inserts a glottal stop (or a full pause) in this environment, e.g. aqo /àqò/ [ʔa˩ qo˩] ‘pit’, imu /ìmú/ [ʔi˩ mu˥] ‘my older brother’, omla /òlà/ [ʔom˩ la˩] ‘your chest’. In some instances, this is realised as creaky voice on the initial part of the vowel, e.g. agi /agí/ [ ˧ ɟi˥] ‘face’, amili /àmìlí/ [m˩ li˥] ‘tongue’. In contrast, the female speaker usually produces word-initial vowels in this environment with creaky voice, which leads into modal voice roughly at the mid-point of the vowel, e.g. amla /àlà/ [m˩ la˩] ‘chest’, imla /ìlà/ [
m˩ la˩] ‘my chest’, omla /òlà/ [
m˩ la˩] ‘your chest’. This creaky phonation can accompany any of the three phonemic tones, e.g. aqo /àqò/ [ ˩ qo˩] ‘pit’, avü /avɨ/ [ ˧ vɨ˧] ‘frost’/‘dew’, afo /áfò/ [ ˥ fo˩] ‘older sister’.
Word-final glottalisation (or even a full glottal stop) is often produced following a high tone, e.g. amlo /àlò/ [am˩ loʔ˥] (in isolation) ‘heart’. This glottalisation is independent from the prosodic glottal stop, i.e. it occurs even when a word is produced in isolation and is hence not the result of a glottal stop inserted before a following vowel-initial word. This will be discussed later, in the section on tones.
Vowels
There are six monophthong vowel phonemes, which can be divided into two levels of height: high and low, and three levels of backness: front, central and back. The vowel chart gives the approximate location of each vowel phoneme.
The low front and back vowels /e/ and /o/ can also be realised as [ɛ] or [ɔ], though these appear to be in free distribution with [e] and [o]. The high central vowel /ɨ/ is sometimes realised as [ə] in word-medial position, e.g. akütsü /àk
ʧ
/ [a˩ kɨ˩ tsɨ˩] ~ [a˩ kə˩ tsɨ˩] ‘black’.
Vowel sequences and long vowels
A few vowel sequences are permissible in Sumi, but they occur across syllable boundaries, e.g. aghau /a.ɣa.ù/ [a˧ ɣa˧ u˩] ‘bird’. Phonetic long vowels may occur with the loss of an intervocalic glide, e.g. amiyi /à.mì.jì/ [a˩ mi˩ ji˩] ~ [a˩ mii˩] ‘charcoal’. More generally, they occur as a result of morphological concatenation. In such cases, as mentioned previously, a glottal stop may also be inserted at the morpheme boundary to break up the vowel sequence, e.g. a-a /à-à/ [aa˩] ~ [a˩ ʔa˩] ‘place’, a-i /à-ì/ [ai˩] ~ [a˩ ʔi˩] ‘arum lily’, au /a-u/ [au˧] ~ [a˧ ʔu˧] ‘hand’.
Vowel harmony
Vowel harmony typically occurs in certain syllables (see section on syllable structure below) in non-word-final position. The vowel in these syllables is typically a high vowel /i ɨ u/ that displays harmony in terms of backness with the vowel of the following syllable. One common example of this can be seen with the deverbal prefix kV-, where V is one of the high vowels /i ɨ u/:
Note that in these examples the non-relational prefix a- is also added. The tone alternations in the deverbal forms are more complex and discussed in greater depth in Teo (Reference Teo2009).
Word-final vowel deletion
In the Satakha dialect, there is also a tendency to delete word-final high vowels following a sonorant, irrespective of the tone they carry. The final tone is then realised on the word-final sonorant, e.g. kighinoli /kìɣìnólí/ [ki˩ ɣi˩ nol˥] ‘intestines’, kini /kini/ [kin˧] ‘two’, amu /àmú/ [am˩˥] ‘older brother’, pamu /pamú/ [pam˧˥] ‘his older brother’.
Syllable structure
The canonical syllable in Sumi is open and allows all consonant phonemes to occupy the onset position. Onsetless syllables are permitted, though a glottal stop may be inserted in this position. Verbs in Sumi are minimally monosyllabic, e.g. ba /bà/ [ba˩] ‘to defecate’, while nouns are minimally disyllabic and take the non-relational ‘nrl’ prefix a- to fulfil the minimal requirement of disyllabicity, e.g. aba /à-bà/ [a˩ ba˩] ‘dung’ (nrl-dung). However, the noun root itself (-ba) is minimally monosyllabic. It should be noted that for reasons of economy, this morpheme boundary will be marked only in this section.
Monomorphemic verbs and numerals in isolation can be composed of a minor syllable followed by a full syllable, a structure first described as ‘sesquisyllabic’ by Matisoff (Reference Matisoff and Hyman1973). Unlike a full syllable, minor syllables only allow a restricted set of consonants /ptkm/ to occupy onset position and the vowel of the minor syllable typically displays vowel harmony with that of the full syllable. The vowel in a minor syllable is typically shorter than a vowel in a full syllable and may be altogether deleted between a stop and a lateral approximant, e.g. küla /klà/ [kɨ˩ la˩] ~ [kla˩] ‘to marry’, piti /piti/ [pi˧ ti˧] ‘to burn’, kini /kini/ [ki˧ ni˧] ‘two’, küthü /kth/ [kɨ˩ thɨ˥] ‘three’. Minor syllables also do not allow the full range of tonal contrasts (see section on tone below).
Monomorphemic noun roots can also be sesquisyllabic, but are not free morphemes – in citation form they still take the non-relational prefix a- to fulfil the minimal requirement of disyllabicity. In many cases, the vowel of the minor syllable is not produced at all, except in very careful speech, resulting in a disyllabic word e.g. akichi /à-kìʧhí/ [a˩ ki˩ ʧhi˥] ~ [ak˩ ʧi˧] ‘breast’, akütsü /à-kʧ/ [a˩ kɨ˩ tsɨ˧] ~ [ak˩ tsɨ˧] ‘head’. In compounds containing such noun roots, the vowel of the minor syllable is almost always deleted in speech, even in careful speech, e.g. akichizü /à-kìʧhí-ʒ/ [ak˩ ʧhi˧ zɨ˥] ‘milk’ (‘nrl-breast-water’) – [a˩ ki˩ ʧhi˧ zɨ˥] is only produced when the speaker is reading the word out.
A syllabic bilabial nasal can also occupy the minor syllable slot, e.g. mla /la/ [˩ la˧] ‘to foam’, muku /mùkú/ [˩ ku˥] ‘twenty’. In nouns, the nasal is resyllabified following the non-relational a- prefix, as well as after the prefixes i- ‘my’ and o- ‘your; e.g. amlo /à‑lò/ [am˩ lo˥] ‘heart’, amla /à-là/ [am˩ la˩] ‘chest’, imla /ì-là/ [im˩ la˩] ‘my chest’, omla /ò-là/ [om˩ la˩] ‘your chest’. A vowel is sometimes added in the careful pronunciation of some words, e.g. muku ‘twenty’ [mu˩ ku˥], perhaps due to the orthographic convention of transcribing the vowel. There is also an orthographic bias to transcribe the vowel /i/, e.g. amili /à-mìlí/ ‘tongue’, which is typically pronounced [am˩ li˥], though [a˩ mi˩ li˥] is also considered to be acceptable. The nasal in the noun root is realised as syllabic when the preceding prefix bears a different tone, e.g. pamla /pa-là/ [pa˧ ˩ la˩] ‘his chest’, pamili /pa-mìlí/ [pa˧ ˩ li˥] ~ [pa˧ mi˩ li˥] ‘his tongue’.
In contrast to sesquisyllabic verbs and noun roots, the vowel of the first syllable of a ‘true’ disyllabic verb or noun root does not necessarily display this sort of vowel harmony and is never deleted in speech, e.g. asamo /à-ʃàmò/ [a˩ sa˩ mo˩] ‘dream’, athonhe /a-thonɦe/ [a˧ tho˧ nɦje˧] ‘tortoise’, ayeghi /a-jeɣí/ [a˧ je˧ ɣi˥] ‘earth’, heqhi /hèqhi/ [he˩ qχi˧] ‘to kill’.
However, in trisyllabic compound nouns, which take the form (C)VCVCV, i.e. three full syllables, if the vowel in the second syllable is high, there is a tendency to treat the second syllable as a minor syllable. Often, it is deleted and the word is resyllabified into two syllables, e.g. Sümitsa /ʃɨ-mì-ʧà/ [sɨ˧ mi˩ tsa˩] ~ [sɨm˧˩ tsa˩] ‘Sumi language (old name)’ (‘Sü‑person-language’); avudu /à-vù-dú/ [a˩ vu˩ du˥] ~ [av˩ du˥] ‘rooster’ (‘nrl‑chicken-male’) (Satakha dialect).
This deletion occurs even when the high vowel does not display harmony with the following vowel, as in ajikhu /aʒi-khu/ [aʒ˧ khu˧] ~ [a˧ ʒi˧ khu˧] ‘cup’ (‘rice beer-plate’). Interestingly, in attempting to recover the deleted vowel in this word, younger speakers insert a high vowel that displays vowel harmony with the following vowel, resulting in the morphologically opaque form [a˧ ʒu˧ khu˧].
Such deletion does not occur when the second syllable is a non-high vowel, e.g. awoshi /à-vò-ʃì/ [a˩ wo˩ ʃi˩] (*[aw˩ ʃi˩]) or [a˩ vo˩ ʃi˩] (*[av˩ ʃi˩]) (Satakha dialect) ‘pork’ (‘nrl-pig-meat’). Speakers hearing [aw˩ ʃi˩] would instead interpret this as awushi ‘chicken meat’.
Tones
In Sumi, full syllables can take any of the three tones – low, mid and high:
These tones are generally realised as level tones, e.g. apu /àpù/ [a˩ pu˩] ‘father’, apu /apu/ [a˧ pu˧] ‘water scoop’. Any significant pitch movements (rises or falls) across the syllable can be attributed to anticipation of the following tone or to f0 declination across the utterance. Teo (Reference Teo2009) provides a more comprehensive description of the phonetic realisation of these tones.
Minor syllables, including the syllabic nasal, take only low or mid tone.
In these examples, the tonal contrast always occurs on the full syllable. A tonal contrast only appears on a minor syllable when it is preceded by a full syllable such as the prefix a-:
In these cases, the minor syllable still only bears low or mid tone.
Interaction between consonants and tone
Word-final glottalisation (sometimes a full glottal stop) often accompanies a word-final high tone, e.g. aza /àʒá/ [a˩ zaʔ˥] ‘mother’, alekiphe /àlèkiphé/ [a˩ lje˩ ki˧ pheʔ˥] ‘singing’, apu /àpú/ [a˩ puʔ˥] ‘son’, ikujo /ìkùʒó/ [i˩ ku˩ ʒoʔ˥] ‘we two’, lakhi /lakhí/ [la˧ khiʔ˥] ‘one’, küthü /kth/ [kɨ˩ thɨʔ˥] ‘three’. Such glottalisation does not occur with non-final high tones, e.g. ifo /ífò/ [i˥ fo˩] ‘my older sister’, axone /àxónè/ [a˩ xo˥ nje˩] ‘fermented soya beans’, anathi /ànáthi/ [a˩ na˥ thi˧] ‘banana’ (Satakha dialect), khetsünhe /khèʧnɦè/ [khe˩ tsɨ˥ nɦje˩] ‘sun’.
That such word-final glottalisation is not the same as the previously discussed glottal stop, which is inserted at a morpheme boundary, is shown by its occurence even after high tones in utterance-final position, e.g. amlo /àlò/ [am˩ loʔ˥] ‘heart’, pamu /pamú/ [pa˧ muʔ˥] ‘his older brother’, muku [mu˩ kuʔ˥] ‘twenty’. It does not occur with other utterance-final tones, e.g. ana /ànà/ [a˩ na˩] ‘(cooked) rice’, azü /àʒ/ [a˩ zɨ˩] ‘water’, athonhe /athonɦe/ [a˧ tho˧ nɦje˧] ‘tortoise’.
Stress
Stress is not phonemic in Sumi. However, it might be argued that minor syllables in sesquisyllabic structures receive ‘less prominence’ vis-à-vis full syllables. They are often shorter than full syllables and are usually deleted. They also take a reduced set of consonants and vowels in their onset slot and nucleus slots, as well as a reduced tonal inventory.
Transcription of recorded passage
The passage recorded and transcribed here is ‘The North Wind and the Sun’, translated by the female speaker into Sumi from English. A phonemic transcription is provided along with an orthographic version.
Phonemic transcription
àhùumɨlɦɨ ŋò khèʧnɦè
àhùumɨlɦɨ ŋò khèʧnɦè kɨmánokhíùnoakumtoù kelapì | kɨpɨna àke ɣùlòkí lono | àɣʒɨmí lakhíno àphì | àklvɨpùʃɨ ìɣi ∥ tìlénò kɨmánoiʃixàthàlù | kɨmá dòlokhíùnoatɨɣuʃi | àɣʒɨú pelónò paphì xàvelùpìyekenò | tìpaùnò akumtoù ipìní pì ∥ ikehu | àhùumɨlɦɨnopakká kmʧ ʃnomlɦ | ikèmú panopakká ʃnò mlɦ àye | àɣʒɨúnopaphì peʃɨ pakɨʒɨ kìthelùve ∥ ikehu | àʃèkhaloye | àhùumɨlɦɨyetìlénò tàve ∥ enò kùɣonò khèʧnɦènolɨvɨ iʃiyèìpèɣikemʧa àɣʒɨúnopàikhàvì lono | paphì xàve ∥ ikehu | àhùumɨlɦɨyekhèʧnɦèno | kɨmá dòloakumtoù kepìipìmò àye ʃàkepú ʃive ∥
Orthographic transcription
Ahu-u mülhü ngo Khetsünhe
Ahu-u mülhü ngo Khetsünhe kümano khiuno akumtou kela pi küpüna ake ghuloki lono aghüzümi lakhino aphi akülü vüpusü ighi. Tileno kümano ishi xathalu, küma dolo khiuno atüghushi aghüzü-u pelono paphi xave lupiye keno tipauno akumtou ipini pi. Ikehu, Ahu-u mülhüno paküka kümtsü süno mülhü ikemu pano paküka süno mülhü aye aghüzü-uno paphi pesü pakuzü kithe lüve. Ikehu, ashekhaloye Ahu-u mülhüye tileno tave. Eno kughono Khetsünheno lüvü ishi yeipeghi kemtsa aghüzü-uno pai khavi lono paphi xave. Ikehu ahu-u mülhüye Khetsünheno küma dolo akumtou kepi pimo aye sakepu shive.
Acknowledgements
I would like to thank John Hajek, Janet Fletcher, John Esling and the anonymous reviewers for all their feedback and guidance. Special mention also goes to Olga Maxwell who helped with the vowel plot. Certainly, Inotoli Zhimomi, Jekügha Assümi and Canato Jimomi have my deepest gratitude for allowing me to work with them on their language.
