by Leon Stassen
This map shows the distribution of the various options in the encoding of predicative adjectives, i.e. items which predicate a property of a subject. The basic distinction is between those languages in which predicative adjectives are encoded in a way that is parallel to predicative verbs, and those languages in which the encoding of predicative adjectives and verbs is different. An example of this latter, nonverbal, encoding of predicative adjectives is English. As shown in the sentences in (1), English shows third person agreement in the present tense of its predicative verbs, but this option is not available for predicative adjectives: *John tall-s is not acceptable in the language. On the other hand, Bororo (Macro-Gê; Mato Grosso, Brazil) offers an instance of verbal encoding of predicative adjectives; as the sentences in (2) demonstrate, the encoding of predicative verbs and of predicative “property words” is identical.
John is tall.
‘I am/was tall.’
In addition to the two basic options of verbal and nonverbal encoding, the map shows a third encoding type for predicative adjectives. In this mixed type, both a verbal and a nonverbal encoding of predicative adjectives is available. An example is from Luo (Nilotic; Kenya). Sentence (3b) shows that, in this language, the item be’r ‘good’ can have verbal encoding, while sentence (3c) indicates that it may also have nonverbal encoding.
‘I am calling.’
‘I am good.’
‘I am good.’
In accordance with the above, the following values are shown on the map:
|Go to map|
|Predicative adjectives have verbal encoding||151|
|Predicative adjectives have nonverbal encoding||132|
|Predicative adjectives have mixed encoding||103|
In order to be able to decide whether a given predicative adjective construction is a case of verbal or non-verbal encoding, we need a set of cross-linguistically applicable criteria. For this map, the following three criteria have been employed (see Wetzer 1996; Stassen 1997). The first of these is the Agreement Criterion:
(4) The Agreement Criterion
If a language has (person/number/gender) agreement on predicative verbs, then predicative adjectives in that language will be rated as verbal if they show this agreement marking as well. If they do not, they will be rated as nonverbal.
The application of this criterion to English and Bororo enables us to decide that English has nonverbal adjectives, whereas Bororo has verbal adjectives (see 1-2). Another case in which this criterion is decisive is presented by Tiwi (Australian; Bathurst Island). As the sentences in (5) demonstrate, predicative adjective encoding in Tiwi must be rated as nonverbal, since the agreement marking that is required by verbs is lacking on predicative adjectives.
3-fall . .
‘The stringy bark is good.’
A second criterion which – alone, or in alignment with the Agreement Criterion – can decide on verbal or nonverbal status of predicative adjectives is the Copula Criterion:
(6) The Copula Criterion
If predicative adjectives are marked by the presence of a supportive item (a copula), then their encoding must be rated as nonverbal.
Application of this criterion yields – in addition to the Agreement Criterion – the conclusion that English predicative adjectives are nonverbal, since they require the copular item be. A case in which the Copula Criterion is the sole decisive factor is presented by Irish. In this language, predicative verbs do not show any agreement marking, so that the Agreement Criterion is not applicable here. Nonetheless, we can decide upon nonverbal status for Irish predicative adjectives, by virtue of the Copula Criterion.
‘He is ill.’
For languages that lack both copulas and agreement marking on verbs, the two criteria presented above are not applicable. For this case, we need an additional third criterion, the Negation Criterion.
(8) The Negation Criterion
If predicative verbs and adjectives show different negation, then the encoding of predicative adjectives must be rated as nonverbal.
The application of this criterion can be illustrated by Gumbaynggir (Pama-Nyungan; New South Wales, Australia) and Tagalog (Austronesian; Philippines). Both languages lack copulas and agreement on verbs. However, Gumbaynggir has different negation strategies for verbs and adjectives (Eades 1979: 332), while the negation strategy in Tagalog is the same for the two predicate categories (see 9a-b). Hence we determine that predicative adjectives are verbal in Tagalog, whereas they are nonverbal in Gumbaynggir.
‘The bus did not come.’
‘The meat is not cheap.’
By applying this set of three criteria it is possible to decide on the verbal or nonverbal status of any case of predicative adjective encoding in a language-independent fashion. It should be stipulated here that the three criteraia are ranked as to their application. That is, the first criterion to be applied is the Agreement Criterion. For those languages to which this criterion cannot be applied (because of a lack of agreement on verbs), the Copula Criterion comes into play. Only in cases where the Copula Criterion still does not provide a decision (because the language in question has a zero-copula for predicate nominals) is the Negation Criterion brought in as a final arbiter on the status of predicative property items.
Mixed encoding of predicative adjectives comes in two types. The example from Luo, presented in (3), illustrates a case of switching: the same lexical item (in this case, the item be’r ‘good’) can “switch” between verbal and nonverbal encoding. In the typical instances of switching, the double option of encoding affects only a subset of the predicative adjectives, namely, those for which both a “permanent/inherent” and a “temporary/accidental” interpretation is possible. In such a case, the nonverbal encoding commonly indicates that the property is inherent and/or permanently applicable to the subject. Thus, sentence (3c) (nonverbal) must be interpreted as “I am a good person”, whereas sentence (3b) (verbal) indicates that the property of being good is only temporary and/or accidental. A second example of this switching phenomenon is presented by Maori (Polynesian; New Zealand). As the examples in (10) illustrate, the item pai ‘good’ can be encoded either verbally (sentence 10b, on a par with sentence 10a) or nonverbally. In this latter case, the predicative adjective takes a zero-copula, as do predicate nominals in Maori (see sentences 10c and 10d).
‘The girl runs.’
‘This house is good.’
‘The talk is good.’ (lit. ‘The talk (is) a good one.’)
‘This bird is a kiwi.’
The second form in which mixed encoding of predicative adjectives manifests itself is that of split encoding. In this case, all predicative adjectives have only a single encoding option, but the set of property words is split into a subset with verbal encoding and a subset with nonverbal encoding. An example of this situation can be found in Rama (Chibchan; Nicaragua). As can be seen from the sentences in (11), the item angaling ‘hungry’ gets verbal encoding, whereas the item mliima ‘good’ is encoded nonverbally (by virtue of the Agreement Criterion).
(11) Rama (Colette Grinevald, p.c.)
‘You are dreaming.’
‘We children are hungry.’
‘This pineapple is good.’
As was the case with switching, there are indications that split encoding of predicative adjectives is often governed by considerations of permanency. Although in many languages the situation is far from clear, there seems to be a tendency to apply the verbal strategy with less permanent properties such as 'hungry', 'ill', or 'sad', and the nonverbal strategy with time-stable properties like 'female' or 'golden' (see Stassen 1997: 164-179).
For the purposes of this map, the differentiation between the two types of mixed encoding has been ignored. Thus, a case of mixed encoding on the map may either stand for switching or for split encoding.
The map demonstrates that the distinction between verbal and nonverbal encoding of predicative adjectives has clear areal features. Nonverbal encoding appears to be concentrated in two large linguistic areas. The first of these mega-areas comprises all the languages of Europe (with the notable exception of the Northwest Caucasian languages), Central Asia and Siberia (with the exception of Yukaghir), India, the Middle East and northern Africa.
A second mega-area of nonverbal encoding is formed by (most of the) languages of Australia and New Guinea. Especially along the coastlines of both islands there are a number of counterexamples. Also, there is a certain degree of mixed encoding here, but the central highlands of New Guinea and the bulk of the Australian mainland contain almost uniform nonverbal encoding.
Outside of these two main areas, nonverbal encoding does not seem to be particularly strong. Perhaps the best case for a third nonverbal area can be made for the languages of the southern part of Central America and the eastern part of South America, but there is considerable diversity in this area.
Large unbroken concentrations of verbal encoding are encountered in Africa, Asia, and the Americas. Practically all the languages of Africa in and south of the Sahara exhibit (some degree of) verbal encoding of predicative adjectives. A second area of verbal encoding comprises East and Southeast Asia, and the islands of Indonesia, Melanesia and Polynesia. Furthermore, verbal encoding is the near-exclusive the option in North America, and it is prominent in the south and the east of South America as well.
As could be expected, mixed encoding tends to appear in those areas where verbal and nonverbal encoding meet. The most conspicuous area in this respect is sub-Saharan Africa, but mixed encoding can also be found in other borderline regions, such as New Guinea, Central America, and eastern India/Burma.