|Ethnicity||975 Hazabee (2005)|
|Native speakers||800 (2000)|
Hadza is a language isolate spoken along the shores of Lake Eyasi in Tanzania by fewer than a thousand Hadza people, the last full-time hunter-gatherers in Africa. Despite the small number of speakers, language use is vigorous, with most children learning it. In the late 20th century Hadza was included in a proposed Khoisan language family, largely on the basis of its use of clicks, but this classification is no longer accepted.
- 1 Classification
- 2 Sounds
- 3 Grammar
- 4 Numbers
- 5 Dead animal names
- 6 Speculations about early human language
- 7 Bibliography
- 8 External links
- 9 References
Hadza is a language isolate (Sands 1998, Starostin 2013). It was once classified by many linguists as a Khoisan language, along with its neighbor Sandawe, primarily because they both have click consonants. However, Hadza has very few proposed cognates with either Sandawe or the other Khoisan languages, and many of the ones that have been proposed appear doubtful. The links with Sandawe, for example, are Cushitic loan words, whereas the links with southern Africa are so few and so short (usually single consonant–vowel syllables) that they are most likely coincidental. A few words link it with Oropom, which may itself be spurious; the numerals itchâme 'one' and pihe 'two' suggest a connection with Kw'adza, an extinct language of hunter-gatherers who may have had recently shifted to Cushitic. (Higher numerals were borrowed in both languages.)
Hadza syllable structure is limited to CV, or CVN if nasal vowels are analyzed as a coda nasal. Vowel-initial syllables do not occur initially, and medially they appear to be equivalent to /hV/.
Hadza is noted for having clicks within morphemes. This distribution is also found in Sandawe and the Nguni Bantu languages, but not in the Khoisan languages of southern Africa. Some of these words are historically derivable from clicks in initial positions (many appear to reflect lexicalized reduplication, for example, and some are due to prefixes), but others are opaque. As in Sandawe, most medial clicks are glottalized, but not all: puche 'a spleen', tanche 'to aim', tacce 'a belt', minca 'to lick one's lips', laqo 'to trip someone', keqhe-na 'slow', penqhenqhe ~ peqeqhe 'to hurry', haqqa-ko 'a stone', shenqe 'to peer over', exekeke 'to listen', naxhi 'to be crowded', khaxxe 'to jump', binxo 'to carry kills under one's belt'.
It is not known if Hadza has lexical tone. It may have a pitch accent system, but the details have not been worked out. There are no known lexical minimal pairs for stress/tone, though the suffix for the habitual mood is tonic.
Hadza has five vowels, [i e a o u]. Long vowels may occur when intervocalic [ɦ] is elided. For example, [kʰaɦa] or [kʰaː], to climb. Nasal vowels, although uncommon, do occur, though not before consonants that have prenasalized homologues. (That is, CṼCV and CVNCV are in complementary distribution.) All vowels are nasalized before glottalized nasal clicks.
|Click||Aspirated||/ᵏǀʰ/ ch||/ᵏǃʰ/ qh||/ᵏǁʰ/ xh|
|Tenuis||/ᵏǀ/ c||/ᵏǃ/ q||/ᵏǁ/ x|
|/ᵑǀ/ nc||/ᵑǃ/ nq||/ᵑǁ/ nx|
|Glottalized nasal1||/ᵑǀˀ/ cc||/ᵑǃˀ/ qq||/ᵑǁˀ/ xx|
|Stop||Aspirated||/pʰ/ ph||/tʰ/ th||/kʰ/ kh||/kʷʰ/ kwh|
|Tenuis||/p/ p||/t/ t||/k/ k||/kʷ/ kw||/ʔ/ –|
|Voiced||/b/ b||/d/ d||/ɡ/ g||/ɡʷ/ gw|
|Aspirated prenasalized||/mpʰ/ mp||/ntʰ/ nt||/ŋkʰ/ nk|
|Voiced prenasalized||/mb/ mb||/nd/ nd||/ŋɡ/ ng||/ŋɡʷ/ ngw|
|Nasal||/m/ m||/n/ n||/ɲ/ ny||/ŋ/ ng’||/ŋʷ/ ng’w|
|Affricate||Aspirated||/t͜sʰ/ tsh||/t͜ʃʰ/ tch||/c͜ʰ/ tlh3|
|Tenuis||/t͜s/ ts||/t͜ʃ/ tc||/c͜/ tl3|
|Voiced||/d͜z/ z||/d͜ʒ/ j|
|Ejective||/t͜sʼ/ zz||/t͜ʃʼ/ jj||/c͜ʼ/ dl3||/k͜xʼ/ gg4||/k͜xʷʼ/ ggw|
|Aspirated prenasalized||/nt͜sʰ/ nts|
|Voiced prenasalized||/nd͜z/ nz||/ɲd͜ʒ/ nj|
|Fricative||/fʷ/ f||/s/ s||/ɬ/ sl||/ʃ/ sh||(/ʜ/ hh)6|
|Approximant||/ɾ ~ l/ l5||/j/ y||/w/ w||/ɦ ~ h/ h7|
- The nasalization of the glottalized nasal clicks is apparent on preceding and sometimes following vowels, but not during the click itself. The labial [ᵑʘˀ] (or [ᵑʘʷ]) is found in a single mimetic word where it alternates with [ᵑǀ].
- The labial ejective /pʼ/ is only found in a few words.
- The palatal affricates may be pronounced with an alveolar onset (/t͜/ etc.), but this is not required. (If the transcriptions do not display properly, they can also be written [c͜ʎ̥˔] etc.)
- The velar ejective /k͜xʼ/ varies between a plosive [kʼ], a central affricate [k͜xʼ], a lateral affricate [k͜ʼ], and a fricative [xʼ]. The other ejective affricates may also appear as ejective fricatives.
- The lateral approximant /l/ is found as a flap [ɾ] between vowels and occasionally elsewhere, especially in rapid speech.
- The voiceless epiglottal fricative [ʜ] is only known from a single word, where it alternates with /kʰ/.
- The glottal /ɦ/ is sometimes voiceless, especially in tonic syllables.
- The prenasalized consonants, /ɲ ŋ ŋʷ d ɡ ɡʷ dʒ/, and perhaps /dz/ (on darker background) seem to be borrowed (Elderkin 1978).
A practical orthography has been devised by Miller and Anyawire (Miller 2008). It is broadly similar to the orthographies of neighboring languages such as Iraqw and Sandawe. The apostrophe, which is ubiquitous in transcription in the anthropological literature but causes problems with literacy, has been eliminated (apart from Swahili 〈ng’〉 for borrowed /ŋ/): Glottal stop is indicated by vowel sequences (that is, /beʔe/ is written 〈bee〉, as in 〈Hazabee〉 /ɦadzabeʔe/ 'the Hadza'), and ejectives and ɡlottalized clicks by gemination (apart from reduced 〈dl〉 instead of *ddl for /cʼ/). The ejectives are based on the voiced consonants, 〈bb zz jj dl gg ggw〉, because these are otherwise found mostly in borrowings and thus not common. Tc /tʃ/ and tch /tʃʰ/ are as in Sandawe, sl /ɬ/ as in Iraqw. (This is ultimately a French convention.) Nasalized vowels are 〈an en in un〉. Long vowels are 〈â〉 or 〈aha〉, because they're usually due to an elided /ɦ/; likewise, apparent vowel sequences are written with an intervening 〈h〉 (or a 〈y〉 or 〈w〉), because a simple sequence would indicate a glottal stop. A tonic syllable may be written with an acute accent, 〈á〉, but is generally not marked.
The following is taken from Miller (2008).
Hadza is a head-marking language in both clauses and noun phrases. Word order is flexible; the default constituent order is VSO, though VOS and fronting to SVO are both very common. The order of determiner, noun, and attributive also varies, though with morphological consequences. There is number and gender agreement on both attributives (for head nouns) and verbs (for subjects).
Reduplication of the initial syllable of a word, usually with tonic accent and a long vowel (VhV), is used to indicate 'just' (meaning either 'merely' or 'solely') and is quite common. It occurs on both nouns and verbs, and reduplication can be used to emphasize other things, such as the habitual suffix -hé- or the pluractional infix 〈kV〉.
Nouns and pronouns
Nouns have grammatical gender (masculine and feminine) and number (singular and plural). They are marked by suffixes as follows:
The feminine plural is used for mixed natural gender, as in Hazabee 'the Hadza'. For many animals, the grammatical singular is transnumeric, as in English: dongoko 'zebra' (either one or a group). The asculine plural may trigger vowel harmony: dongobee 'zebras' (an individuated number), dungubii 'zebra bucks'. A couple kin terms and the diminutive suffix -nakwe take -te in the m.sg., which is otherwise unmarked.
Gender is used metaphorically, with ordinarily feminine words made masculine if they are notably thin, and ordinarily masculine words made feminine if they are notably round. Gender also distinguishes such things as vines (m) and their tubers (f), or berry trees (f) and their berries (m). Mass nouns tend to be grammatically plural, such as atibii 'water' (cf. ati 'rain', atiko 'a spring').
The names reported for dead animals do not follow this pattern. Calling attention to a dead zebra, for example, uses the form hantahii (masculine hantahee, plural (rare) hatahetee and hantahitchii). This is because these forms are not nouns, but imperative verbs; the morphology is clearer in the imperative plural, when addressing more than one person: hantatate, hantahate, hantahetate, hantahitchate (substitute -si for final -te when addressing only men; see below for the verbal object suffixes -ta-, -ha-, -heta-, -hitcha-).
The -pe and -pi forms of nouns (actually -phee and -phii) often seen in the anthropological literature are copular: dongophee 'they are zebras'. The copular suffixes distinguish gender in all persons as well as clusivity in the 1st person. They are:
Forms with high vowels (i, u) tend to raise preceding mid vowels to high, just as -bii does. Because the Hadza h is murmured, the 3.sg copula tends to sound like a -ya or -wa after high and often mid vowels: /oha, eha/ ≈ [owa, eja], and transcriptions with w and y are common in the literature. The 1.ex suffixes begin with a glottal stop, not written here.
Personal and demonstrative pronouns are:
There are some additional 3rd-person pronouns, including some compound forms. Adverbs are formed from the 3rd-person forms by adding locative -na: hamana 'here', beena 'there', naná 'over there', himiggêna 'in/behind there'.
Verbs and adjectives
The copula was covered above. Hadza has several auxiliary verbs: sequential ka- and ya- 'and then', negative akhwa- 'not', and subjunctive i-. Their inflections may be irregular or have different inflectional endings from those of lexical verbs, which are as follows:
|2sg||-tita ~ -hita||-taa||-tee||-tikwi||-V||-ta|
The functions of the anterior and posterior differ between auxiliaries; with lexical verbs, they are non-past and past. The potential and veridical conditionals reflex the degree of certainty that something would have occurred. 1sg.npst -ˆta and a couple other forms lengthen the preceding vowel. The 1.ex forms apart from -ya begin with a glottal stop (not written). The imp.sg is a glottal stop followed by an echo vowel.
Habitual forms take a tonic -hé, which tends to reduce to a long vowel, before these endings. In some verbs, the habitual has become lexicalized (replacing the h of the 3.POST forms with glottal stop), and so an actual habitual takes a second -hé. Various compound tense-aspect-moods occur by doubling up the inflectional endings. There are several additional inflections which have not been worked out.
The inflectional endings are clitics and may occur on an adverb before the verb, leaving a bare verb stem (verb root plus object suffixes).
As is common in the area, there are only a few bare-root adjectives in Hadza, such as pakapaa 'big'. Most attributive forms take a suffix with cross-gender–number marking: -he (m.sg. and f.pl.) or -hi (f.sg. and m.pl.). These agree with the noun they modify. The -hi form tends to trigger vowel harmony, so that, for example, the adjective one- 'sweet' has the following forms:
The -ko/-bee/-bii ending may be replaced by the copula, but the e/i cross-number-gender marking remains.
Demonstratives, adjectives, and other attributives may occur before or after a noun, but nouns only take their gender–number endings when they occur first in the noun phrase: Ondoshibii unîbii 'sweet cordia berries', manako unîko 'tasty meat', but unîbii ondoshi and unîko mana. Similarly, dongoko bôko but bôko dongo 'those zebra'.
Verbs may also be made attributive: dluzîko akwiti 'the woman (akwitiko) who is speaking', from dlozo 'to say'. This attributive form is used with the copula to form the progressive aspect: dlozênee 'I am speaking' (male speaker), dluzîneko 'I am speaking' (female speaker).
Verbs may take up to two object suffixes, for a direct object (DO) and indirect object (IO). These only differ in the 1ex and 3sg. The IO suffixes are also used on nouns to indicate possession (mako-kwa 'my pot', mako-ha-kwa 'it is my pot').
|1.in||-hona ~ -yona|
|3m||-ha ~ -ya ~ -na||-ma||-hitcha|
Two object suffixes are only allowed if the first (the DO) is 3rd person. In such cases the DO reduces to the form of the attributive suffix: -he (m.sg. / f.pl.) or -hi (f.sg. / m.pl.); only context tells which combination of number and gender is intended. 3rd-singular direct objects also reduce to this form in the imperative singular; 3rd-plural change their vowels but do not conflate with the singular: see 'dead zebra' under nouns above for an example of the forms.
The factors governing the word order within noun phrases is not known. Constituent order tends to be SXVO (where X is an auxiliary) for a new or emphasized subject, with the subject moving further back (XSVO, XVSO, and XVOS), or simply not mentioned (XVO) the better it's established. Where context, semantics, and the verbal suffixes do disambiguate them, verb–noun–noun is understood to be VSO.
The Hadza did not count before the introduction of Swahili. Native numerals are itchâme 'one' and píhe 'two'. Sámaka 'three' is a Datooga loan, and bóne 'four', botáno 'five', and ikhúmi 'ten' are Sukuma. Áso 'many' is commonly used instead of botano for 'five'. There is no systematic way to express other numbers without using Swahili.
Dead animal names
Hadza has received some attention for a dozen 'celebratory' (Woodburn) or 'triumphal' (Blench) names for dead animals. These are used to announce a kill. They are (in the imperative singular):
|Animal||Generic name||Triumphal name|
|other large antelope||hephéhee|
The words are somewhat generic: henqehee may be used for any spotted cat, hushuhee (hushuwee) for any running ground bird. 'Lion' and 'eland' are distinguished only by gender. Blench (2008) thinks this may have something to do with the eland being considered magical in the region.
An IO suffix may be used to reference the person who made the kill. Compare hanta 'zebra' with the more mundane verbs, qhasha 'to carry' and kw- 'to give', in the imperative singular and plural (Miller 2009):
|a zebra!||I got a zebra!|
|carry it!||give it to me!|
Speculations about early human language
In 2003 the press widely reported suggestions by Alec Knight and Joanna Mountain of Stanford University that the original human language may have had clicks. The purported evidence for this is genetic: speakers of Juǀ'hoan and Hadza have the most divergent known mitochondrial DNA of any human populations, suggesting that they were the first, or at least among the first, surviving peoples to have split off the family tree. In other words, the three primary genetic divisions of humanity are the Hadza, the Juǀʼhoan and relatives, and everyone else. Because two of the three groups speak languages with clicks, perhaps their common ancestral language, which by implication is the ancestral language for all humankind, had clicks as well.
However, besides the genetic interpretation, this conclusion rests on several unsupported assumptions:
- Both groups have kept their languages intact, without language shift, because they branched off of the rest of humanity;
- Neither borrowed clicks as part of a Sprachbund, as the Bantu languages Zulu, Sotho, and Yeyi did; and
- Neither the ancestors of the Juǀʼhoan nor those of the Hadza developed clicks independently.
There is no evidence that any of these assumptions are true, or even likely. Linguistic opinion is that clicks may well be a relatively late development in human language, that they are no more resistant to change or any more likely to be linguistic relics than other speech sounds, and that they are easily borrowed: at least one Khoisan language, ǁXegwi, is believed to have reborrowed clicks from Bantu languages, which had earlier borrowed them from Khoisan languages, for example. The Knight and Mountain article is the latest in a long line of speculations about the primitive origin of click consonants, which have been largely motivated by the outdated idea that primitive people speak primitive languages, an idea which has no empirical support.
- Sands, Bonny E. (1998) 'The Linguistic Relationship between Hadza and Khoisan' In Schladt, Matthias (ed.) Language, Identity, and Conceptualization among the Khoisan (Quellen zur Khoisan-Forschung Vol. 15), Köln: Rüdiger Köppe, 265–283.
- Bonny Sands, Ian Maddieson, Peter Ladefoged (1993). 'The Phonetic Structures of Hadza', UCLA Working Papers in Phonetics, No. 84: Fieldwork Studies in Targeted Languages.
- A.N. Tucker, M.A. Bryan, and James Woodburn as co-author for Hadza (1977). 'The East African Click Languages: A Phonetic Comparison'. In J.G. Moehlig, Franz Rottland, Bernd Heine, eds, Zur Sprachgeschichte und Ethnohistorie in Afrika. Berlin: Dietrich Diener Verlag.
- Edward Elderkin (1978) Loans in Hadza: internal evidence from consonants. Occasional Papers 3. Dar es Salaam.
- Kirk Miller (2008) 'Hadza Grammar Notes'. 3rd International Symposium on Khoisan Languages and Linguistics, Riezlern.
- ———— (2009) 'Highlights of Hadza fieldwork'. LSA, San Francisco.
- Hadza bibliography
- Science article speculating on the status of clicks in the original human language (PDF file)
- Ladefoged, Peter; Maddieson, Ian; Sands, Bonny (1991). "Hadza". UCLA Phonetics Lab Archive. Hadza wordlist and sound files.
- Sands, Bonny; Maddieson, Ian; Ladefoged, Peter (1993). "The phonetic structures of Hadza". Fieldwork studies of targeted languages. UCLA working papers in phonetics. pp. 67–87.
- Edenmyr, Niklas (2004). "The semantics of Hadza gender assignment: a few notes from the field" (PDF). Africa & Asia (Göteborg: Dept of Oriental and African languages, Göteborg University) (4): 3–19.
- Blench, Roger (7–9 July 2008). "Hadza Animal Names" (PDF). 3rd International Khoisan Workshop. Riezlern.
- Hadza basic lexicon at the Global Lexicostatistical Database
- Hadza reference at Ethnologue (17th ed., 2013)
- Ya and ka take -ˆto, -tikwa, -ˆte, -ˆti in the 3rd-person posterior rather than -(h)amo, -(h)akwa, -(h)ame, -(h)ami, for example. In lexical verbs, those endings are used with habitual -hé- to emphasize it.
- E.g. in Parkvall (2006) Limits of language
- These glosses are crude; these words do not translate into Swahili or English
- Sands & Güldemann, 2009, "What click languages can and can't tell us about language origins". In Botha & Knight, eds, The Cradle of Language, Oxford.