This chapter examines the order of negative morpheme and lexical verb for negative morphemes that code simple clausal negation, like English not in (1).
The mouse did not eat the cheese.
Not considered are negative constructions involving more than simple clausal negation, like (2a), or clauses with nonverbal predicates, like (2b) to (2d), in languages where one or more of these is treated differently from clauses with verbal predicates, negative imperative (prohibitive) sentences, like (2e), or constituent negation, where a non-verbal constituent is the focus of the negation, as in (2f).
The mouse never eats cheese.
John is not honest.
John is not a doctor.
John is not at his house.
Don't take those bananas.
It is not John who took the bananas.
See Chapter 112 for discussion of different types of negative morphemes and Chapters 113 and 114 for other typological features of clausal negation. This chapter distinguishes negative affixes from negative words, but among negative words does not distinguish negative auxiliary verbs from negative particles (see Chapter 112). Maps that make such a distinction can be obtained by combining Map 112A with maps in this chapter, though I have such data for only a subset of the languages in this chapter. Chapter 144 further discusses the position of negative words relative to the subject and object as well as to the verb. Chapter 71 examines prohibitive sentences.
Map 143A shows all the languages in the sample, while Maps 143B, 143C, and 143D show only the languages with double and triple negation. Maps 143E, 143F, and 143G rearrange the data on the first four maps and show languages with preverbal negative morphemes, postverbal negative morphemes, and negative tone respectively.
No distinction is made here between morphemes that only code negation, like English not, and morphemes that code grammatical features in addition to negation. For example, Ma'di (Central Sudanic, Nilo-Saharan; Sudan and Uganda) has two negative words kʊ̄ and kʊ̄rʊ̀, the former for nonpast negation, the latter for past negation, as in (3a) and (3b).
‘I am not opening the door.’
‘I did not open the door.’
More complex cases involve words that also code the person and number of the subject, like the initial word bàn in the Hausa (West Chadic, Afro-Asiatic; Nigeria and Niger) example in (4), which also codes a first person singular subject.
‘I don’t know his name.’
In this case bàn counts as a negative word because it codes negation even though it also codes first person singular. In some cases, one can identify a negative morpheme within such words, as in English, where in a form like didn’t, the n’t is a suffix on the auxiliary verb did. However, since (as discussed just below) we are examining the position of negative morphemes relative to the main verb (rather than auxiliaries), the entire word didn’t counts as a negative word, like bàn in Hausa, so this is not considered a construction involving a negative suffix but rather as one involving a negative word.
In examining the order of negative morpheme and verb, we examine its position relative to the main verb rather than the finite verb; in other words, in clauses where the finite verb is an auxiliary verb, we examine the position of the negative relative to the main verb, ignoring the auxiliary verb. In some languages, this can obscure the fact that the position of the negative morpheme is defined, according to the grammatical rules of the language, relative to the finite verb, which is generally the auxiliary verb if there is one. For example, in Danish, the negative word immediately follows the finite verb. When there is no auxiliary verb, as in (5a), the negative will follow the main verb, but when there is an auxiliary verb, as in (5b), it will precede the main verb, since in such cases it is immediately following the finite auxiliary verb.
‘John did not buy a car yesterday.’
‘He hasn’t bought a car this year.’
Because we ignore the auxiliary verb, Danish is classified in this chapter as a language with NegV/VNeg order, as one with both orders of negative and verb, with neither dominant. The decision to use the order of negative and main verb, even if the main verb is not the finite verb in the clause, is motivated by a number of factors. One is that we want to use a consistent approach in describing word order in both this chapter and the next chapter, Chapter 144, where we examine the position of the negative morpheme relative to the subject and object as well as the verb. In describing the order of subject, object, and verb, it is conventional to base this on the position of the subject and object with respect to the main verb, not the auxiliary verb if there is one. For example, SAuxOV clauses in German are normally considered SOV, based on the position of the main verb, rather than SVO, based on the position of the auxiliary verb, despite the fact that the verb-second order in German means that both SVO and SAuxOV clauses in German are placing the object after the finite verb. In discussing the position of the negative in SOV languages, we want to use SOV in its conventional sense. Furthermore, if we were to classify languages according to the position of the negative relative to the auxiliary verb, we would have to either ignore the main verb altogether or describe the position of the negative relative to both the main verb and the auxiliary. Ignoring the main verb altogether would be as serious as ignoring the auxiliary verb and describing the position of the negative, subject, and object relative to both the main verb and the auxiliary verb would have complicated things in Chapter 144 immensely. Another reason for ignoring auxiliary verbs is that in most languages in which there is little verb morphology, it is generally unclear whether a word coding tense and/or aspect should be considered a verb or not and unclear what should be considered the finite verb. If we have no clear basis for deciding whether a word is an auxiliary verb in a given language or what is the finite verb, it is necessarily unclear how to classify such a language in terms of the position of the negative relative to the finite verb. A final reason is that when the negative itself is an auxiliary verb, we need to code its position relative to the nonfinite main verb; in these cases, we have to classify such languages in terms of the position of the main verb, even though the main verb is not the finite verb. If we want to be consistent, then we must code the negative relative to the main verb rather than the finite verb in all languages.
Before beginning the discussion, it is necessary to explain the notation used in characterizing the types on this map and subsequent maps. Negative affixes are indicated with square brackets and a hyphen: prefixes are denoted by [Neg-V] and negative suffixes by [V-Neg]. Negative words are indicated without square brackets or hyphen. For example, Neg[V-Neg] means a preverbal negative word plus a negative suffix. On later maps which include types with optional negative morphemes, parentheses are used to indicate an optional negative morpheme. For example, [(Neg-)V]Neg means an optional negative prefix and an obligatory postverbal negative word.
|Go to map|
|Type 1 / Type 2||22|
|Type 1 / Type 3||9|
|Type 1 / Type 4||12|
|Type 2 / Type 3||2|
|Type 2 / Type 4||9|
|Type 3 / Type 4||8|
|Type 3 / Negative Infix||1|
|Optional Single Negation||1|
|Obligatory Double Negation||114|
|Optional Double Negation||80|
|Optional Triple Negation with Obligatory Double Negation||5|
|Optional Triple Negation with Optional Double Negation||1|
(The notation “Type X / Type Y” means that the language has two constructions of the types in Types 1 to 5, with neither clearly dominant.)
Type 1, and by far the most frequent type, represents those languages in which the normal expression of clausal negation is a negative word which precedes the verb, as in the example from Kutenai (isolate; western North America) in (6).
(6) Kutenai (own data)
‘I don't smoke.’
Note that in languages of this type, the negative word need not immediately precede the verb. In Yuwaalaraay (Pama-Nyungan; Australia), the negative word normally occurs at the beginning of the sentence, separated from the verb, as in (7).
‘You didn’t cut any firewood.’
Negative clitics, i.e. negative morphemes which attach phonologically to some word but vary as to what word they attach to, are treated here as negative words. For example, in Sidaama (Eastern Cushitic, Afro-Asiatic; Ethiopia), the negative clitic attaches either to the verb, as in (8a), or to another constituent preceding the verb which is the focus of negation, as in (8b).
‘Bule didn’t give food to Dangur.’
‘Bule didn’t give food to Dangur (but to someone else).’
Constituent negation of the sort illustrated in (8b) is excluded from consideration from this chapter and from Chapter 144; however, the fact that the negative morpheme in Sidaama can attach to other constituents shows that di is not a prefix in (8a) but rather a clitic.
‘Koko did not hit Goko.’
‘He did not tell about the women.’
‘I am not eating.’
Type 5 represents languages in which the normal expression of negation is by means of distinctive tone on the verb. While we will see that there are a number of languages that use tone to signal negation, most combine this with some segmental coding or use it as an alternative to segmental coding. The sample used here contains only one language in which tone is the sole coding of negation, namely Engenni (Edoid, Niger-Congo; Nigeria), in which a high tone on the verb signals negation (Thomas 1978: 68). (There is no Engenni example in my source that is clear enough to illustrate this.) It is worth mentioning here the case of Mano (Eastern Mande, Niger-Congo; Guinea and Liberia), in which negation is coded either by a preverbal negative particle or by changing the tone on the subject pronoun. Compare affirmative (12a), with mid tone on the first person singular subject pronoun, with negative (12b), with falling tone on the subject pronoun.
‘I don’t know.’
Because the negative coding is on a preverbal word, the coding in (12b) counts as a preverbal negative word in the sense that there is a preverbal word that codes the person and number of the subject and negation, analogous to the case of bàn in Hausa in (4) above.
The next seven types represent mixed types, languages that have two orders or constructions from Types 1 to 5. In some languages where there are two such constructions, the same morpheme is involved, as in (13) from Eipo (Trans-New Guinea; Papua, Indonesia), where the same word gum either precedes the verb, as in (13a), or follows the verb, as in (13b).
‘... I will not eat.’
‘It is not that you come to wound me.’
More commonly, however, a different morpheme is involved, as in (14) from Aghem (Bantu, Niger-Congo; Cameroon), where negation is signaled in completive aspect by a word kà that precedes the verb as in (14a), while in incompletive aspect, it is signaled by a word 'yɔ̀ that follows the verb, as in (14b).
‘He did not hit the mat (earlier today).’
‘He is not hitting the mat.’
In some languages where there are two options, the choice is semantically or grammatically governed, as is the case with Aghem. In other cases, the choice is not grammatically governed, but determined by extragrammatical factors, without a clear semantic contrast associated with the two options. This is apparently the case with Eipo in (13). To a large extent, one can predict whether or not the choice is semantically or grammatically governed from whether the same morpheme is involved: if different morphemes are involved, as in Aghem, the choice is generally semantically or grammatically conditioned, while if the same morpheme is involved, as in Eipo, the choice is generally conditioned by extragrammatical factors. However, there are exceptions to this generalization, cases where different morphemes are involved but the choice is still apparently not governed semantically or grammatically, as in (15) from Lavukaleve (Solomons East Papuan; Solomon Islands), where Terrill (2003: 465) describes the choice between the postverbal negative word tamu in (15a) and the suffix -la in (15b) as “stylistically” governed.
'We didn't eat.'
'... he didn't eat ...'
The first of the types involving a mixture of Types 1 to 5 is Type 6, which represents languages where negation is expressed by a negative word that either precedes or follows the verb. Both Eipo and Aghem, illustrated above in (13) and (14) respectively, are instances of Type 6.
Type 7 represents languages in which one finds either a negative word preceding the verb or a negative prefix, as illustrated by Maasai (Nilotic, Nilo-Saharan; Kenya and Tanzania), where negation is expressed by a prefix m- in the present and continuous tenses, as in (16a), and by a separate preverbal word eItÚ in the past tense, as in (16b).
‘I did not sing.’ (Tucker and Mpaayei 1955: 67)
‘I do not sing.’ (Tucker and Mpaayei 1955: 67)
‘The men don’t make a dory.’
‘He did not arrive.’
‘I am not carrying water.’
‘I am not going to carry the basket.’
Type 10 represents languages in which the negative morpheme is either a separate word following the verb or a suffix, illustrated above in (15) by Lavukaleve.
‘Kings do not build palaces.’
‘I did not go to Benares’
Type 12 represents languages in which the negative morpheme is either a prefix or an infix. The sole instance of a language of this type in the sample is Dolakha Newar (Tibeto-Burman; Nepal), illustrated in (20). In this language, the negative morpheme is an affix which immediately precedes the last syllable of the verb stem. If the verb stem is monosyllabic, it will appear as a prefix, as in (20a). If it is disyllabic, it will appear as in infix, as in (20b).
‘These days father also doesn’t love me.’
An alternative way of characterizing Dolakha Newar would be to say that the negative affix is always an infix in the sense that there is actually only one construction that defines the position of the affix relative to a particular syllable in the word, making it an infix, and in some instances the position of that infix happens to precede all material in the root to which it is infixed.
Type 13 represents languages in which it is possible to have negative clauses without a negative morpheme. In Wyandot (Iroquoian; Canada), negated clauses involve an optional preverbal negative word tąʔą ~ hąʔą ~ ąʔą, illustrated in (21a), (21c) and (21g) plus one of two verbal prefixes, one which Kopris (2001) characterizes as a negative morpheme, t(e(ʔ))-, illustrated in (21a) and (21b) the other which he calls contrastive, t(i)- illustrated in (21c) and (21d), which must be used instead of the negative prefix when the verb is in factual mood. However, the so-called negative prefix, t(e(ʔ))-, is also used in counterfactual conditionals, as in (21e), so that I treat it as some sort of ‘irrealis’ prefix. The contrastive prefix t- is also used in contexts where English would use the word but, as in (21f). It is also possible to use the negative particle itself, without either the negative or the contrastive prefix, as in (21g).
'He did not mind.'
'Horses don't hatch.'
'They could not find out [what was the matter].'
'[The eagle mother began to worry] because they [her children] could no longer eat.'
'[I wish that] if I were married, I would converse like this forever.'
'[Raccoon then competed with Turtle,] but Turtle arrived there first.
'I will never give him back these two young women.'
There are two ways we could analyse Wyandot. First, since neither the irrealis prefix nor the contrastive are specifically negative, we could say that they do not count as negative morphemes. But that would mean that examples like (21b) and (21d) where the negative particle is not used would lack a negative morpheme. Under that view, Wyandot is an example of a language with optional single negation in which it would be possible to have negative clauses without a negative morpheme. An alternative view is that even though the negative and contrastive prefixes sometimes occur with nonnegative meaning, they still count as negative morphemes, since in some contexts they apparently convey specifically negative sense. I will opt here for the first view on the grounds that in cases where there is no negative particle, it appears to be the case that the specifically negative sense is not due to the meaning of the prefixes but due to the pragmatic context.
Type 14 represents languages which employ obligatory double negation, where there are two morphemes simultaneously coding negation, as in (22) from Izi (Igboid, Niger-Congo; Nigeria), which contains both a preverbal negative word and a negative suffix.
‘The man is not speaking it.’
Note that the two morphemes in a Type 14 language need not occur on different sides of the verb. In Hdi (Chadic, Afro-Asiatic; Cameroon and Nigeria), for example, there are two postverbal negative words, one immediately after the verb, the other in clause-final position, as in (23).
‘Hdi do not like school.’
Such cases are unusual, however; it is far more common for multiple negative morphemes to occur on opposite sides of the verb. Map 143B expands on different subtypes of Type 14, languages with obligatory double negation.
Type 15 represents languages with optional double negation, in which it is possible to have two negative morphemes occurring simultaneously but in which it is also possible to have just one negative morpheme, as in Mupun (West Chadic; Nigeria), illustrated in (24), where the preverbal negative word ba is optional, while the clause-final negative word kas is obligatory.
‘I did not eat the bird meat.’
For the purposes of this chapter, a language is deemed to have optional double negation if it is possible to express negation with a single negative morpheme and if it is also possible to express negation by the use of two negative morphemes. An additional condition is that if the use of two negative morphemes is fairly rare in a given language, then the language is not classified as having optional double negation. Although it is difficult to give a precise definition for what counts as “rarely used”, two kinds of examples are (1) if the source says explicity that it is rarely used; and (2) if it is restricted to subordinate clauses or clauses containing nonindicative verbs.
There are in fact a variety of different ways in which a language can satisfy the criteria for optional double negation described above, some of which are worth discussing here. The simplest type of optional double negation is illustrated by the Biloxi (Siouan; Mississippi) example in (25), where there is an obligatory postverbal negative word ni and an optional negative prefix k-. This is the simplest case: there is one negative morpheme that always appears and a second negative morpheme that only sometimes appears.
‘You did not go.’
‘You did not go.’
The Mupun example in (24) above also illustrates this sort of optional double negation.
A more complex case of optional double negation is represented by Kwoma (Sepik; Papua New Guinea), illustrated in (26), where there is either double negation with a preverbal negative word and a negative suffix on the verb, as in (26a), or single negation, with just a negative suffix on the verb, as in (26b).
‘He didn’t go’ or ‘He isn’t going.’
‘He didn’t go.’
The Kwoma case differs from Mupun and Biloxi in that the negative morpheme that is used when there is single negation, -kasakech in (26b), is different from the negative morpheme that occurs in the same position when there is double negation, -wak in (26a). This illustrates that the term ‘optional’ is being used in an extended sense, to include situations like this one where the choice between the two constructions is determined grammatically and where there is no unique obligatory negative morpheme that occurs in both constructions. Rather, the term is being used broadly simply to refer to any situation where there are two possibilities, one with one negative morpheme and one with two negative morphemes.
Kwoma also differs from Mupun and Biloxi in that the choice between the two options, double negation and single negation, is determined grammatically in Kwoma, but apparently by extragrammatical factors in Mupun and Biloxi. In fact, it appears to be generally the case that whenever there is no unique negative morpheme shared by the two constructions, the choice between the two constructions will be determined grammatically, as in the case of Kwoma, while whenever the same negative morpheme occurs in both constructions, the choice between the two constructions will be determined by extragrammatical factors, as in the case of Biloxi and Mupun. However, there are occasional exceptions to this generalization. For example, in Tariana (Arawakan; Brazil), stative verbs are negated with a negative suffix -kade, as in (27a), while nonstative verbs are negated with both a negative prefix ma- and the negative suffix -kade, as in (27b).
‘I am not angry.’
‘I didn’t drink whiskey.’
In other words, there is a unique morpheme shared by both constructions in Tariana, but the choice between the two constructions is grammatically conditioned.
In the case of Kwoma, although there is no unique negative morpheme involved in the two constructions, there is at least a unique morphosyntactic position which is filled in both constructions, albeit by different morphemes, namely a verbal suffix position, so there is at least an obligatorily filled negative position. A case not conforming to this pattern, however, is provided by Pokot (Nilotic; Kenya and Uganda), in which one construction involves a preverbal negative word, as in (28a), while the other involves the combination of a negative prefix and a negative suffix on the verb, as in (28b).
‘We are not coughing.’
‘I do not take.’
Two other instances of this type are Duna (Trans-New Guinea; Papua New Guinea) and Sahidic Coptic (Egyptian Coptic; Egypt). Duna is like Pokot, except that the construction involving a negative word places the negative word after the verb, as in (29).
‘I didn’t hear.’
‘I do not eat snake.’
All the cases of optional double negation discussed so far have involved two options, one involving two negatives, the other involving a single negative which usually, though not always, occurs in the position of one of the two negatives in the construction with two negatives. But a more complex possibility is provided by Nkem (Bantu; Nigeria), illustrated in (30). In Nkem, there are really three constructions, one of which involves two negative morphemes, the other two of which involve a single negative morpheme. The position of the single negative morpheme in each of these latter two constructions corresponds to the position of one of the two negative morphemes in the construction with two negative morphemes. In (30a), negation is expressed only by a negative suffix -dom on the verb. In (30b), negation is expressed only by a negative word nɨ̂m preceding the verb. And in (30c), there is double negation, with both a negative suffix -lom (a phonological variant of -dom ) and a preverbal negative word kà.
‘I did not see any house.’
‘I will not buy any house.’
‘I have not finished.’
A particularly complex type is provided by Ayutla Mixe (Mixe-Zoque; Mexico), in which there are three negative morphemes, all of which can occur alone and any two of which can occur together, but where apparently all three do not co-occur. Two of the negative morphemes, ni and ka’t, are preverbal words, one of them illustrated in (31a), while the other negative morpheme, ka-, is a negative prefix, illustrated in (31b). Example (31c) illustrates the two negative words co-occurring, while (31d) illustrates one of the negative words co-occurring with the negative prefix.
‘Carlos did not eat tamale.’
‘Carlos does not eat.’
‘Yes, not even your dad could cheer me up...’
‘He did not study...’
Map 143C shows different subtypes of languages with optional double negation.
Types 16 and 17 involve languages with optional triple negation, Type 16 languages being those with obligatory double negation, Type 17 those with optional double negation. In other words, Type 16 languages employ either two or three negatives, while Type 17 languages employ one, two, or three. An example of a Type 16 language is Gunbalang (Gunwinygic; Australia), illustrated in (32). Example (32a) involves double negation: there is both a negative prefix and a negative suffix. But (32b) involves not only the same prefix and suffix but also an additional negative word ngunta preceding the verb.
‘I didn't go.’
'I didn't go down there.'
The sample of languages for this chapter contains only one instance of a Type 17 language, one with optional double and optional triple negation, namely Kwomtari (Kwomtari-Baibai; Papua New Guinea), illustrated in (33). All four options involve a preverbal negative word glei. The first option, shown in (33a), involves this word in combination with both a negative suffix -i on the verb and a postverbal negative word mwa. The second option, shown in (33b), involves this preverbal word in combination with just the negative suffix -i on the verb. The third option, shown in (33c), involves this preverbal word in combination with just the postverbal negative word mwa. And the fourth option involves the preverbal word alone.
‘Now we don’t do it well.’
‘The people did not habitually come and look at Peter.’
‘They didn’t just carry them away and tie them.’
‘This man didn’t love father God.’
Map 143D shows languages with both of these two types of triple negation.
It should be noted that in examining the position of negative morphemes for this chapter, we ignore the fact that the structure of negative clauses is often more complex than simply adding a negative morpheme, as discussed in Chapter 114 and Miestamo (2005). For example, in Hixkaryana (Cariban; Brazil), illustrated in (34), the negative construction also involves the addition of a form of the verb for ‘be’.
‘I went hunting.’
‘I did not go hunting.’
In cases like this, the position of the negative morpheme can really only be understood in the context of the overall negative construction. Simply talking about negative morphemes here misses the point that what we really have here is a negative construction, not just a negative morpheme. Similarly, in Tarao (Kuki-Chin, Tibeto-Burman; India), negation is expressed not only by the use of a negative suffix, but also by the absence of subject prefixes which are used in affirmative clauses, as illustrated in (35), where not only does (35b) employ a negative suffix, but the subject prefix ki- ‘’ that is used in the affirmative sentence in (35a) is absent in the negative sentence in (35b).
‘I eat rice.’
‘I do not eat rice.’
This chapter only examines the position of negative morphemes with respect to the verb, so this aspect of negative constructions is not taken into account here, nor in Chapter 144.
Note that to count as a negative morpheme, a morpheme has to be specifically negative in meaning. In some languages with double negation, the negative construction requires, in addition to a negative morpheme, a second morpheme with irrealis meaning or something similar, but which is not specifically negative in that it also occurs in nonnegative contexts. For example, in Alamblak (Sepik; Papua New Guinea), negative clauses differ from corresponding affirmative clauses in the use of a preverbal negative word and obligatory irrealis suffix on the verb, as in (36a). However, irrealis mood also occurs in a number of nonnegative contexts, such as counterfactual conditionals, where both clauses are marked irrealis, as in (36b).
‘They did not live in the forest.’
‘If you had gone (and you should have), I would have followed you.’
Although the use of irrealis in Alamblak can be considered part of the negative construction, the irrealis morpheme is not treated as a negative morpheme because it is not specifically negative, and thus Alamblak is not treated as a case of double negation.
The case of Wyandot discussed above and illustrated in (21) above is an instance of a language in which it is possible to have negative clauses without a negative morpheme because although the language does have a negative morpheme, that single negative morpheme is optional. However, a more extreme logical possibility that is not clearly attested in the sample is a language in which there is no morpheme that can be considered a negative morpheme. There are many logically possible ways in which this might arise. A language might use a different word order for coding negation, the way a few languages code polar questions (see Chapter 116). There are languages which employ a different word order in negative clauses (see Map 144B), but these all involve negative morphemes as well. Another way a language might lack a negative morpheme would be if it had an affirmative morpheme that is absent in negative clauses (again parallel to what we find in a few languages for polar questions). Or a language might employ two morphemes in negative clauses, neither of which is specifically negative, because both occur in nonnegative clauses, but which only occur together in negative clauses. The closest case of a language of this sort that I am aware of is Igbo (Benue-Congo; Nigeria), which is not in my sample for this chapter because the complete facts are not clear to me from my sources. My description here follows Miestamo’s (2005: 119-120) interpretation of negation in Igbo. There are three ways in which negation is coded in Igbo. First, there is a suffix ‑ghɪ̀, illustrated in (37a), which some sources describe as a negative suffix, but which apparently occurs occasionally in nonnegative clauses, so it is not considered a negative morpheme for the purposes of this chapter. Second there is a prefix a-, illustrated in (37b), that occurs in negative imperfective clauses in which the subject is not a singular pronoun. However, in perfective clauses, this prefix occurs on the verb only if the clause is not negative, illustrated in (37c), where it is glossed here as ‘’ for ‘prefix, so it also cannot be considered a purely negative morpheme. Finally, there are tonal differences between verbs in affirmative and negative clauses, illustrated by the differences between the tones on the two verbs in (37c); however, what is not clear is whether there is a particular tone that can be considered a marker of negation, as opposed to there being tonal differences but where the tones in negative clauses occur in nonnegative clauses in a way that would mean that tone cannot be considered a purely negative morpheme either.
‘He does not want money.’
‘Ada does not want money.’
‘We carried some but they did not.’
If there is not a specific tone that can be considered a marker of negation, then Igbo would count as a language without any morpheme that is specifically negative.
Not all instances of two negative morphemes are counted as double negation. Instances of two negative prefixes or two negative suffixes are not counted as double negation, even if other morphemes can intervene. For example, in Venda (Bantu; South Africa and Zimbabwe), illustrated in (38), there are two negative prefixes, separated by a subject noun class prefix; but this is not treated as double negation.
‘The dog did not bite the child.’
On the other hand, separate negative words, as opposed to affixes, are treated as double negation if they occur in different syntactic positions, where words can occur between them, as in (39) from Bafut (Bantu; Cameroon), where one negative word precedes the subject and the other follows.
‘Ngwa has not built a house.’
Another example is presented by Breton (Celtic; France), where two negative words bracket the inflected verb. When there is no auxiliary, these words bracket the main verb, as in (40a); however, when there is an auxiliary verb, they bracket the auxiliary verb, as in (40b) (where the first one cliticizes onto the auxiliary verb).
‘The children do not collect flowers by the stream.’
‘Alan didn’t go into the cave ...’
It is the example in (40b) that illustrates the point in question. Namely, the two negative words in (40b) cannot be adjacent to each other because the auxiliary verb must occur between them. The fact that they cannot occur adjacent to each other means that this is counted as double negation, so Breton is considered a language with obligatory, not optional, double negation.
In contrast, in one of the negative constructions in Doyayo (Adamawa, Niger-Congo; Cameroon), negation is coded both by the preverbal negative word taa¹² and by high tone on the preceding subject pronoun hi¹, as illustrated in (41). Each of these alone would count as a preverbal negative word, but they both precede the verb without the possibility of a phrase occurring between them, so this is not counted as double negation. (However, this construction does involve double negation for a different reason; namely, negation is also coded by the tone on the incompletive suffix -ko³ ). If a word could be inserted between hi and taa, this would count as triple negation.
‘They will not come.’
As noted above, the most frequent type by far is Type 1, with preverbal negative words. Not only is Type 1 by far the most frequent, but when one adds in Types 6, 7, and 8, in which one of the options involves preverbal negative words, as well as the subset of languages of Types 13 to 16 with double or triple negation that involve preverbal negative words (more specifically Types 1, 2, 7, 8, and 10 to 15 on Map 143B, Types 1 to 4, 11 to 14, 16 to 20, and 22 on Map 143C, and Types 1 and 3 through 6 on Map 143D; see also Map 143E below), these make up a majority of the languages on the map. Although we can observe this preference for preverbal negative words over postverbal negative words, we do not find any significant difference between the frequency of negative prefixes and negative suffixes. However, this can probably be attributed to the general crosslinguistic suffixing preference, which competes with the preference for preverbal negative morphemes, thus resulting in roughly equal numbers of the two types of languages with negative affixes.
Languages with obligatory double negation (Type 14) make up approximately 9% of the languages in the sample, while those with optional double negation (Type 15) make up about 6% of the sample, so that approximately 15% of languages have optional or obligatory double negation.
Languages with preverbal negative words are the dominant type and they are found across the globe. Languages with postverbal negative words show a much more clearly defined geographical distribution, a point which is discussed in more detail in the discussion of Map 144D in the next chapter, since the primary source of the geographical pattern is SVONeg languages. The clearest area in which languages with postverbal negative words are found is central Africa, with a second smaller area around New Guinea. Languages with negative prefixes are especially common in Africa, more specifically centered around east Africa, most of these languages being Bantu languages. Negative prefixes are also common in an area around northeast India, primarily among Tibeto-Burman languages. The largest concentration of languages with negative suffixes is in the northern half of South America. Languages with double negation (either obligatory or optional) are notable for their low frequency in Eurasia.
|Go to map|
|Negative tone & VNeg||1|
|Negative tone & [Neg-V]||2|
|Type 1 / Type 2||1|
|Type 1 / Type 3||1|
|Type 1 / Type 5||1|
|Type 1 / Type 7||1|
|Type 1 / Type 9||1|
|Type 2 / Type 4||1|
|Obligatory double negation with optional triple negation||5|
Double negation involves constructions with two negative morphemes. However, as discussed in more detail above, two prefixes or two suffixes are not counted as double negation and two negative words are not treated as double negation if they obligatorily occur adjacent to each other. If material can occur between two negative words, then these words are counted as being an instance of double negation.
Map 143B examines more deeply different subtypes of languages with obligatory double negation. Types 1 through 4 are ones in which the two negative morphemes are either separate words or affixes on opposite sides of the verb. Types 5 and 6 represent languages with obligatory double negation where one of the two negative morphemes is tonal. Types 7 to 9 represent languages where the two negative morphemes occur on the same side of the verb. Types 10 to 15 represent mixed subtypes, subtypes that involve two of the types in Types 1 to 9. Type 16 represents all languages with obligatory double negation with optional triple negation; their subtypes are shown on Map 143D.
For each of the two negative morphemes, there are the following two parameters: (1) is the morpheme a separate word or an affix? (2) does it precede or follow the verb or verb stem? The first four types on Map 143B represent the four logical possibilities among languages in which one negative morpheme precedes the verb and the other follows, varying by these two parameters as to whether the morpheme that precedes the verb or verb stem is a separate word or a prefix and whether the morpheme that follows the verb is a separate word or a suffix.
Type 1 represents languages where both negative morphemes are separate words, on opposite sides of the verb. Example (42) from Abun (West Papuan; Papua, Indonesia) illustrates a language of this type.
‘They did not give money to me.’
Type 2 represents languages with a preverbal negative word and a negative suffix on the verb. Example (22) above from Izi illustrates a language of this type, as does Baré (Arawakan; Brazil and Venezuela), illustrated in (43).
‘My son did not go away to spend his holiday.’
Type 3 represents languages where the negative morpheme preceding the verb or verb stem is an prefix while the one following is a separate word. Example (44) from Bwe Karen (Karen, Tibeto-Burman; Myanmar) illustrates this type.
‘These plants, I don’t know their names.’
‘The sun is not shining now.’
The remaining ten types are less frequent. Type 5 represents one language with double negation which combines a separate negative word following the verb with a distinctive negative tone on the verb. This language, Baule (Kwa, Niger-Congo; Côte d'Ivoire), is illustrated in (46); (46a) illustrates a negative present tense sentence, while (46b) illustrates an affirmative present tense sentence and (46c) an affirmative future tense sentence.
‘he is not looking’
‘he is looking’
‘he will look’
Note that the tone on the verb nian ‘look’ is different in all three examples, with a distinctive high tone in (46a) marking the verb as negative in combination with the postverbal negative word mán. The tone on the subject pronoun o also differs in (46a) and (46b), so Baule might be treated as a language with triple negation. Two alternative approaches are (1) to view the high tone on the subject pronoun as the spreading of high tone from the verb so that there is really only one high tone spread over the subject pronoun and the verb; or (2) the tone on the subject pronouns seems to be high for negative and future clauses and low for past and present so it could be analysed as irrealis versus realis, so the high tone on the subject pronoun is not specifically negative. I assume that one of these is the correct view and therefore regard Baule as having just double negation, not triple negation.
Type 6 represents languages in which the double negation involves the combination of a negative prefix and negative tone on the verb. In Degema (Benue-Congo; Nigeria), negation is coded by a combination of prefixes and tone on the verb, as in (47), where (47a) shows an affirmative clause and (47b) the corresponding negative clause. Observe the different subject prefixes and the different tones on the verbs.
‘He is running.’
‘He is not running.’
Types 7 to 9 represent languages with obligatory double negation where the two negative morphemes occur on the same side of the verb or verb stem. There are four logically possible ways this might arise, differing along two parameters in a way similar to the way Types 1 to 4 differ: (1) do they both precede or do they both follow the verb or verb stem? and (2) are they both separate words? or is one of them an affix? Only three of these four logical possibilities are attested, namely Types 7 to 9; there is no instance in the sample of a language with a negative suffix and a negative word following the verb.
Type 7 represents languages in which the negative morphemes are both separate words that precede the verb. There is only one language of this type in the sample, Bafut, illustrated above in (39).
‘We did not eat it.’
Type 9 represents languages with two negative words following the verb. There are two languages of this sort in the sample, Hdi, illustrated above in (23), and Zande (Ubangi, Niger-Congo; Democratic Republic of Congo, Central African Republic, and Sudan), illustrated in (49).
‘I am not doing work.’
The next six types represent mixed types, that is languages with two different negative constructions, neither of which is clearly dominant, but each of which falls into one of the above nine types.
Type 10 represents languages that combine Types 1 and 2. In other words, they have one construction with a negative word preceding and a negative word following and a second construction with a negative word preceding and a negative suffix. There is one language of this type, Awa Pit (Barbacoan; Colombia and Ecuador), illustrated in (50). There is a negative word shi preceding the verb in both (50a) and (50b), but there is a postverbal separate word ki in (50a) and a suffix -ma in (50b).
‘She is not standing.’
‘Santos did not go.’
Type 11 contains a single language, Luvale (Bantu; Angola), that combines Types 1 and 3. It has two constructions containing a negative word (actually a clitic) following the verb but the two constructions differ in that one has a negative prefix, illustrated in (51a), while the other has a negative word preceding the verb, illustrated in (51b).
‘They did not find those things they sought.’
‘He does not come to school.’
Type 12 represents languages which have a negative word following the verb and either a negative word preceding the verb or negative tone on the verb. The sole instance of a language of this type is Kisi (Southern Atlantic, Niger-Congo; Guinea, Liberia, and Sierra Leone), illustrated in (52). In Kisi, the word order is SVO in clauses without an auxiliary verb, and when such clauses are negative, there is a clause-final negative word plus distinct negative tone on the verb, as in (52a). When there is an auxiliary verb, the word order is SAuxOV, and when such clauses are negative, the same clause-final negative word occurs, in addition to either negative tone on the auxiliary verb, as in (52b), or a distinct negative auxiliary verb, as in (52c). These last two cases both count as NegVNeg; the distinct negative tone on the auxiliary verb still counts as a negative word, since the entire auxiliary verb in this case counts as a negative word, albeit one that codes other information as well; as discussed above (and in Chapter 112), morphemes coding negation in addition to other information are not distinguished from those coding only negation.
‘Nobody loves that man.’
‘She is not sharpening the machete.’
‘He was not striking the people.’
Type 13 represents languages in which there are two negative words, where either they occur on opposite sides of the verb or both precede the verb. The sole instance of a language of this type is Breton, illustrated above in (40).
Type 14 represents languages which have either a negative word preceding the verb and a negative word following the verb or two negative words both following the verb (but not necessarily adjacent). The sole instance in the sample of a language of this type is Miya (West Chadic, Afro-Asiatic; Nigeria), illustrated in (53), where (53a) illustrates a negative word preceding the verb and one following the verb, while (53b) illustrates two negative words following the verb, but separated, in this instance, by both the object and the subject.
‘Nduya will not plant millet.’
‘Nduya did not plant millet.’
Type 15 represents languages in which both constructions employ a negative suffix, where one employs a negative word preceding the verb and the other employs a negative prefix. The sole instance of this type in the sample is Tirmaga (Surmic, Nilo-Saharan; Ethiopia), illustrated in (54), where the construction in (54a) is used for the imperfective aspect, the one in (54b) for the perfective.
‘And the women had not danced yet.’
‘Now, I don’t want beer.’
Type 16 represents languages with obligatory double negation but with the option of triple negation. These languages are expanded upon below on Map 143D.
The first four types on Map 143B, those involving negative words and negative affixes on opposite sides of the verb or verb stem, are the most common types. It is clear that double negation typically involves morphemes on opposite sides of the verb. The sample contains 36 languages with negative words on opposite sides of the verb (Type 1), but only 4 (Types 7 and 9) where the two words occur on the same side of the verb. Furthermore, if the language has a preverbal negative word and a negative affix, then that affix is more likely to be a suffix (Type 2; 28 languages) than a prefix (Type 8; 2 languages). And if the language has a postverbal negative word and a negative affix, then that affix is more likely to be a prefix (Type 3; 9 languages) than a suffix (unattested).
Among the first four types, one type is distinctly less common than the other three, namely Type 3 involving a negative prefix and a postverbal negative word. It is not surprising that this type is less common than Types 1 and 2, since these last two types involve preverbal negative words, and as shown on Map 143A, these are much more common than languages with postverbal negative words or negative prefixes or suffixes. It is less obvious why Type 3 is less common than Type 4, languages in which there is a negative prefix and a negative suffix. On Map 143A, we saw that in languages with a single negative morpheme, it is much more common for this negative to be a separate word preceding the verb, while the other three basic types (a separate word following the verb, a prefix, and a suffix) are all about equally common. That might lead us to expect that in languages with double negation, those involving a separate word preceding the verb would be more common than those that do not involve such, where the negative morpheme preceding the verb is a prefix. But that predicts that Types 1 and 2 on Map 143B should be more common than Types 3 and 4, but we find Type 4 ([Neg-V-Neg]) as common as Types 1 and 2.
One way of thinking about this is the following. The lower frequency of Type 3 ([Neg-V]Neg) relative to Type 4 ([Neg-V-Neg]) amounts to an implicational generalization of the form ‘If a language has double negation and one of the negative morphemes is a prefix on the verb, then the other one will probably be a suffix rather than a separate word’. Significantly, the reverse of this (‘If a language has double negation and one of the negative morphemes is a suffix, then the other one is likely to be a prefix’) is not true. In other words, negative prefixes predict negative suffixes while negative suffixes do not predict negative prefixes. But this would seem to be simply reflecting the crosslinguistic suffixing preference: the first implicational generalization above correctly predicts suffixes, while the second one, which is not true, predicts prefixes. Put another way, if a language has a negative prefix, then because of the overall suffixing preference, that means it is likely to be a language in which grammatical morphemes are more often affixes rather than separate words, so that if it has a negative prefix, then the postverbal negative morpheme is more likely to be a negative suffix than a negative word.
|Go to map|
|NegV plus optional change to verb stem||1|
|NegV or NegativeTone&VNeg||1|
|Optional double negation with optional triple negation||1|
Map 143C shows languages with optional double negation, languages in which there is sometimes a single negative morpheme and sometimes two negative morphemes. As discussed above (and illustrated by Kwoma in (26) and Pokot in (28)), the notion of optional double negation is used broadly here to include any situation in which there is at least one option involving one negative and and at least one other involving two negative morphemes, even if the morphemes involved in the first case are not one of those involved in the second case.
Types 1 to 17 are types where there are two options, one involving a single negative morpheme, the other involving two negative morphemes, as in Kwoma, illustrated above in (26). Types 18 to 22 involve three options, two involving a single negative morpheme and a third involving two negative morphemes, as in Nkem, illustrated above in (30). Type 23 involves four options, one involving a single negative morpheme, two involving two negative morphemes, the third involving three negative morphemes.
On the preceding map, Map 143B, there were four basic types of obligatory double negation involving negative morphemes on opposite sides of the verb, varying for whether each of the two morphemes was a separate word or an affix. For each of these types, there are two corresponding types involving optional double negation, differing in which negative is optional. For example, given the obligatory double negation type with two negative words, one preceding and one following the verb, there are two corresponding optional double negation types, one where it is the preverbal word that is optional and one where it is the postverbal word that is optional. The first eight types on Map 143C are the eight types corresponding to the first four types on Map 143B.
Types 1 and 2 involve a preverbal and a postverbal negative word, differing as to which is optional. Type 1 represents languages where the obligatory negative word is preverbal. An example of such a language is Hausa (West Chadic, Afro-Asiatic; Nigeria, Niger), illustrated in (55), where (55a) illustrates the possibility with two negative words, (55b) the possibility with only a preverbal negative word.
‘I don’t know his name.’
‘I am not going with you.’
‘We don’t see a jinn.’
‘I won’t be able to sleep.’
Types 3 and 4 involve languages with a preverbal negative word and a negative suffix on the verb. Type 3 represents languages where it is the preverbal negative word that is obligatory, as in Temein (Eastern Sudanic, Nilo-Saharan; Sudan), illustrated in (57).
‘I don’t drink.’
‘People do not dance today.’
‘I am not going to walk.’
Types 5 and 6 involve languages with a negative prefix and a postverbal negative word. Type 5 represents languages where it is the negative prefix that is obligatory, as in Adzera (Oceanic; Papua New Guinea), illustrated in (59).
‘I am not able to help you.’
Type 6 represents languages with an obligatory postverbal negative word and an optional negative prefix. Biloxi, illustrated in (25) above, is the sole language of this type in the sample.
Types 7 and 8 involve languages with a negative prefix and a negative suffix. Type 7 represents languages where it is the negative prefix that is obligatory, as in Ao (Kuki-Chin, Tibeto-Burman; India), illustrated in (60).
‘From that day onwards, [aforementioned] Leopard Cat did not fear Rooster.’
‘And then, night comes and [she] doesn’t return.’
Type 8 represents languages with an obligatory negative suffix and an optional negative prefix. Tariana, illustrated in (27) above, is an instance of a language of this sort.
The preceding types are ones in which the two negative morphemes occur on opposite sides of the verb. Types 9 to 12 are ones in which the two negative morphemes occur on the same side of the verb. Type 9, illustrated in (61) from Lamang (Chadic, Afro-Asiatic; Nigeria), involves languages which are V(Neg)Neg, with two postverbal negative words, an optional one immediately following the verb and an obligatory one in clause-final position. Both negative words in Lamang are apparently clitics. Example (61a) illustrates the clause-final negative clitic, following an object. Example (61b) illustrates the optional negative suffix -b that immediately follows the verb (stem). Its status as a clitic or suffix is somewhat unclear. The fact that it precedes the subject pronominal morphemes which otherwise behave like verbal suffixes might suggest that it is a suffix. However, it can occur on other constituents when those constituents are in focus, as in (61c), suggesting that it is a clitic, the analysis assumed here.
‘They did not reach the water’
‘I am not sitting.’
‘It is not the tongue that I keep cutting.’
Type 10 represents a type of optional double negation where both negatives follow the verb stem, with an obligatory negative suffix and an optional negative word following the verb. There are two instances of languages of this type, Nivkh (isolate; Sakhalin Island, Russia) (Panfilov 1962: 156-157) and Maba (Maban, Nilo-Saharan; Chad). Example (62) illustrates the option with double negation for Maba.
‘I am not tired.’
Note that the case of Type 10 with double negation corresponds to an unattested type of obligatory double negation, [V-Neg]Neg. In other words, if the postverbal negative word in (62) were obligatory, Maba would be an instance of this unattested type.
Type 11 represents languages in which there are either one or two negative words preceding the verb. The sole instance of a language of this type is Rapanui (Oceanic; Easter Island), illustrated in (63), where (63a) illustrates the single negative, (63b) the double negative.
‘I did not go up.’
‘I am not going up.’
Note that the two preverbal negative words in (63b) count separately since there is an intervening subject.
Type 12 represents languages with an obligatory preverbal negative word and an optional negative prefix. The sole instance of a language of this type is Oneida (Iroquoian; New York State and southern Ontario), illustrated in (64), where (64a) illustrates the double negation, (64b) single negation.
‘I don’t know.’
‘He didn’t pass (e.g. a grade at school).’
The negative prefix that occurs in (64a) is replaced by a “contrastive” prefix in (64b) in certain aspects and this contrastive prefix is not specific to negative clauses and hence does not count as a negative morpheme.
Type 13 represents languages that code negation with a preverbal negative word and in some cases changes to the verb stem. The sole instance of a language of this type is Tamashek (Berber; Mali), illustrated in (65). In the future, negation is coded only by a preverbal negative word, as in (65a). In the perfective and imperfective, there is not only the preverbal negative word, but there are also changes to the verb stem, typically changes in the vowel. Example (65b) illustrates this possibility; compare the negative imperfective form of the verb stem of the verb ‘to stand’, bəddəd, in (65b) with the positive imperfective form of the stem of the same verb, which is bɑ́ddæd.
‘I will not see.’
‘He is not standing.’
(The difference between the two negative words in (65) is phonologically conditioned by the next word.)
Type 14 represents a more unusual type of optional double negation, consisting of one language, Pokot, illustrated above in (28). In Pokot, the position of the one negative morpheme when there is just one is not the same as the position of either of the two negative morphemes when there are two of them, in that the negative morpheme when there is just one is a preverbal negative word, whereas the two negative morphemes are a negative prefix and a negative suffix when there are two of them.
Type 15 is similar to Type 14, with one construction involving a negative word and the other involving both a prefix and suffix on the verb, except that in Type 15 the negative word follows the verb. The sole language of this type is Duna, illustrated in (29) above.
Type 16 is in a sense the opposite of Types 14 and 15: it represents languages where one option is a negative prefix alone, the other option a negative word preceding the verb plus a negative word following the verb. The sole instance of a language of this sort is Sahidic Coptic (Reintges 2004: 338, 346, 371).
Type 17 represents languages which either have a negative word preceding the verb or have negative tone on the verb combined with a postverbal negative word. The sole language of this type is Ewondo (Bantu, Niger-Congo; Cameroon), illustrated in (66), where (66a) illustrates an affirmative present verb form and (66b) the corresponding negative with different tone on the verb and the postverbal negative word ki. The example in (66c) illustrates the negative past, where an auxiliary verb is used; here the negative word immediately follows the auxiliary and precedes the main verb so this construction is classified as a preverbal negative word.
‘I am eating’
‘I am not eating’
‘I didn’t eat’
The remaining types represent instances of optional double negation where there are more than two options for negative clauses, illustrated by Nkem and Ayutla Mixe in (30) and (31) above. In Types 18 to 22, there are three options, while in Type 23 there are four options. In Type 18, there is either a preverbal negative word or a negative suffix or both. The examples in (30) above illustrate this type for Nkem, as do the examples in (67) for Guhu-Samane (Trans-New Guinea; Papua New Guinea): (67a) illustrates the preverbal negative word bamu, used in the future; (67b) illustrates a negative suffix, used in the present; and (67c) illustrates both a preverbal negative word te and a negative suffix used in the past, both different from the negative morphemes in (67a) and (67b).
‘You will not look.’
‘He does not go.’
‘I did not go.’
In Type 19, there is either a preverbal negative word or a postverbal negative word or both. Two languages, Tacana (Tacanan; Bolivia) (Ottaviano and Ottaviano 1967: 144) and Welsh Romani (Indic, Indo-European; Wales), are languages of this type, the latter illustrated in (68).
‘I do not want to sell him.’
‘I will not kill you.’
‘The stick will not beat the dog.’
In Type 20, represented only by Domari (Indic, Indo-European; Israel, Iran, and Jordan), there is either a preverbal negative word, as in (69a), a negative suffix, as in (69b), or a negative prefix plus the negative suffix, as in (69c).
‘They did not see.’
‘I do not bring.’
‘I do not bring’
Type 21 represents languages where there is a negative prefix, a postverbal negative word, or both a prefix and a suffix. The sole instance of a language of this type is Tshangla (Tibeto-Burman; Bhutan and India), illustrated in (70), where (70a) illustrates the negative prefix used alone, (70b) and (70c) illustrate the postverbal word, and (70d) illustrates the negative prefix used in combination with a negative suffix.
‘I am not going to fight with you.’
‘At one time, you had not been knowing the mercy of god, ...’
‘I haven’t been taking my medicines at all.’
‘He didn’t get the gold.’
These three options in Tshangla arise in the following way. All three options involve a negative prefix ma- ~ man-, which occurs on the main verb if there is no auxiliary, as in (70a) and (70d) or on the auxiliary if there is one, as in (70b) and (70c). When it occurs on the main verb, the verb must bear a suffix, but sometimes this suffix is not specific to negative constructions, as in (70a) and sometimes it is specific to negative constructions, as in (70d) (in which case we have double negation). When the negative prefix occurs on the auxiliary verb, as in (70b) and (70c), this auxiliary counts as a negative word by the criteria assumed here. Now the morpheme I am treating as a negative suffix in (70d) is actually the same morpheme as the root of the auxiliary verb in (70b). This morpheme is more generally a past tense copula verb. As a copula or auxiliary verb, it is not specifically negative. However, it apparently occurs as a suffix only in negative constructions, so I am treating it in those cases as a negative morpheme, though one could just as easily treat it as a morpheme not specific to negative constructions, in which case, Tshangla would have only double negation.
Type 22 is represented only by Ayutla Mixe, illustrated above in (31). In this language, there is a preverbal negative word or a negative prefix or both.
The last type, Type 23, involves four options where there is optional double negation plus optional triple negation. The one language of this type, Kwomtari, is illustrated above in (33). This type is also shown on Map 143D.
The most frequent type on this map is Type 2, with an obligatory postverbal negative word and an optional preverbal negative word. This type is almost twice as frequent as the next more frequent type, Type 1, with an obligatory preverbal negative word and an optional postverbal negative word. It is somewhat surprising that Type 2 is the more frequent type, since among languages with a single negative word, languages with preverbal negative words (524 languages) are over three times as common as languages with postverbal negative words (171 languages). However, when we look at the distribution of the two types on the map, we see that many of the Type 2 languages are concentrated in a few areas. In fact, 13 of the 20 Type 2 languages are spoken in Africa. And three of the others are Yuman languages in northwestern Mexico and adjacent United States. While 3 of the Type 1 languages are Torricelli languages of Papua New Guinea, the Type 1 languages overall have a broader geographical distribution.
There is also a curious difference between the frequency of the different types with optional double negation and the corresponding types with obligatory double negation. If we pool Types 1 and 2 on Map 143C (NegV(Neg) and (Neg)VNeg), and likewise for the next three pairs of types, we can compare the frequencies of languages requiring obligatory double negation with those allowing optional double negation, given in the following table:
The primary difference between the types with obligatory and optional double negation is that among languages with optional negation, the first type, NegVNeg, is twice as common (21 languages) as the next most frequent type (11 languages), but not significantly greater among languages with obligatory double negation (36 languages versus 28). The only possible explanation I am aware of is that reflects the same areal phenomenon mentioned in the preceding paragraph, an accident of the high number of (Neg)VNeg languages in Africa.
|Go to map|
|[V-Neg]Neg with optional negative infix or prefix||1|
|(i) Neg[V-Neg] or (ii) NegVNeg with negative tone||1|
The sample contains six languages with optional triple negation, each of a unique type. Types 1 to 5 have obligatory double negation with optional triple negation. Type 6 has optional double negation with optional triple negation.
Type 1, represented by Gunbalang, is illustrated above in (32); it involves an obligatory negative prefix and negative suffix on the verb plus an optional negative word preceding the verb.
Type 2, represented by Tabla (Sentani; Papua, Indonesia), involves an obligatory negative suffix on the verb and obligatory negative word following the verb plus an optional negative infix or prefix, as illustrated in (71). The example in (71a) illustrates the possibility without the optional negative infix. The examples in (71b) and (71c) illustrate the possibility with the negative infix; when the verb stem begins with a vowel, as in (71c), the infix appears as a prefix.
‘not fell (trees)’
‘not put in’
Forms with the infix are apparently somewhat irregular. With some verbs whose stems begin with consonants, the first vowel is replaced by /i/, with others the /i/ occurs as a prefix, and with others there is no possibilility of an infix or prefix. Note that the case without the optional infix or prefix is [V-Neg]Neg, an unattested type of double negation without triple negation.
Type 3, represented by Noni (Bantu, Niger-Congo; Cameroon), is illustrated in (72). There are two options with two negatives, either a clause-initial negative word plus a clause-final particle, as in (72a), or a negative word immediately preceding the verb plus the clause-final particle, as in (72b). The second option, with three negatives, illustrated in (72c), involves one negative in clause-initial position, a second one in immediately preverbal position (in the sense that neither the subject nor the object intervene between the negative and the verb), and a third one in clause-final position.
‘They will not hit (later today).’
‘They didn’t sing and hit the children.’
‘They will not hit (later today)’
Type 4, represented by Doyayo (Adamawa, Niger-Congo; Cameroon), is illustrated in (73). In the first option, illustrated in (73a) (repeated from (41) above), there is negative tone on the subject pronoun hi¹, a negative auxiliary verb taa¹² preceding the verb, plus negative tone on the incompletive suffix ko³, which counts here as a negative suffix. As explained in the discussion of this example above under (41), the first two of these count as a single negation rather than as double negation since they occur adjacent to each other without the possibility of intervening material. In the second option, illustrated in (73b), there is first negative tone on the subject pronoun, which counts here as a negative word preceding the verb; second, negative tone on the verb; and third, a clause-final negative word gɛɛ²³.
‘They will not come.’
‘I didn’t see him.’
If we had chosen to define double negation differently such that adjacent negative words were counted separately, then Doyayo would have counted as a language with obligatory triple negation since (73a) involves two preverbal negative words as well as negative tone on the suffix on the verb and (73b) already counts as triple negation by the criteria assumed here. However, none of the languages in the sample have obligatory triple negation by the criteria assumed here.
Type 5 represents languages with an obligatory preverbal negative word or prefix, an obligatory postverbal negative word, and an optional negative suffix. The sample contains only one language of this type, Kongo (Bantu, Niger-Congo; Democratic Republic of Congo), illustrated in (74). Example (74a) illustrates the possibility of a preverbal negative word ka, a negative suffix -a: on the verb, and a clause-final negative word kó. The preverbal negative word in some instances optionally fuses with the subject prefix, as in (74b), so that in these instances we have a negative prefix, a negative suffix, and a clause-final negative word. The negative suffix is not used, however, in relative completive aspect, as in (74c).
‘I will not eat chicken.’
‘You are not coming.’
‘We have not yet eaten.’
Kongo is perhaps the closest case to a language with obligatory triple negation (assuming the criteria used here) since the possibility of double negation is restricted to only one of a number of aspects.
Type 6 involves the only type with optional double negation and optional triple negation. That is, it is like the preceding Types 1 to 5 in having optional triple negation, but differs from them in that double negation is also optional for Type 6 languages. It represents languages with an obligatory preverbal word and two optional morphemes, a negative suffix and a postverbal negative word, where it is possible to have neither of the optional morphemes, just one or the other of them, or both. The sole instance of a language of this type, Kwomtari, is illustrated above in (33).
Maps 143E to 143G show the same languages that have been shown on the preceding four maps, but the data is rearranged. Map 143E shows languages that allow preverbal negative morphemes, either separate words or prefixes, and includes types shown on each of the preceding four maps. It includes languages where the sole possibility involves a preverbal negative, languages that allow an alternation at least one of the possibilities of which involves a preverbal negative, and languages with double negation in which one of the negatives is a preverbal negative.
|Go to map|
|Preverbal negative word||681|
|Both preverbal negative word and negative prefix||23|
|No preverbal negative morpheme||390|
Type 1 represents languages in which at least one possibility involves a preverbal negative word, but does not include languages that also have negative prefixes. This includes languages where this is the sole possibility (Type 1 on Map 143A), languages where there are two possibilities, at least one of which involves a preverbal negative word (Types 6 to 8 on Map 143A), languages with obligatory double negation where at least one possibility involves a preverbal negative word (Types 1, 2, 7, 8, and 10 to 14 on Map 143B), languages with optional double negation where at least one possibility involves a preverbal negative word but where a negative prefix is not possible (Types 1 to 4, 9 to 11, 13, and 17 to 19 on Map 143C), languages with optional single negation where the optional negative morpheme is a preverbal negative word (Type 13 on Map 143A), and languages with triple negation where at least one possibility involves a preverbal negative word (Types 3, 4 and 6 on Map 143D). Type 2 represents languages in which at least one possibility involves a negative prefix but where a preverbal negative word is not possible. Type 3 represents languages in which both a preverbal negative word and a negative prefix exist as possibilities. This includes languages with single negation where there is an alternation between preverbal negative words and negative prefix (Type 7 on Map 143A, Types 14, 16, 20, and 22 on Map 143C) and languages where preverbal negative words and negative prefixes can co-occur (Type 8 on Map 143B, Types 12 and 21 on Map 143C, and Type 1 and 5 on Map 143D) . Type 4 represents languages where there is no possibility of expressing negation with a preverbal negative morpheme, where only postverbal negative morphemes or tone are used.
|Go to map|
|Postverbal negative word||288|
|Both postverbal negative word and negative suffix||18|
|No postverbal negative morpheme||711|
|Go to map|
|Negative stem change||1|
|No negative tone, infix or stem change||1314|
Map 143G shows three types of minor morphological means of coding negation. Type 1 includes all languages that employ tone on the verb as at least part of the way to code negation. Type 2 does the same for infixes and Type 3 the same for changes to the verb stem. Type 4 represents languages which do not represent negation by any of these three morphological means.