Map 104A represents the order of A and P person markers on the verb relative to each other, where A stands for the person marker of the agentive argument of a transitive verb and P for the person marker of the patient argument of the transitive verb. Five values are represented:
|Go to map
|A and P do not or do not both occur on the verb
|A precedes P
|P precedes A
|Both orders of A and P occur
|A and P are fused
The first value covers languages that exhibit no verbal person marking, have marking of only one of the transitive arguments, or allow both arguments to be marked but not at the same time. This last possibility is sometimes found in languages which display hierarchical alignment of verbal person forms (see chapter 100), as is the case in Guajajara (Tupi-Guarani; north-eastern Brazil). As shown in (1), in Guajajara the transitive verb may agree with either the A as in (1a) or the P as in (1b), depending on which is higher on the hierarchy: 1 > 2 > 3, but it never agrees with both A and P.
‘I saw the monkey.’
‘He cut me with a knife.’
The form of the actual agreement markers is sensitive to whether the A outranks the P or vice versa. Thus the form of the first person marker in (1a), where the A is higher on the hierarchy than the P, is different from that in (1b), where it is the P that is higher than the A.
The second value, A before P, subsumes three types of languages: those in which both the A and P person markers are prefixes, as in Swahili (see 2), those in which they both are suffixes, as in Amharic (Semitic; Ethiopia) (see 3), and those in which the two markers occur on opposite sides of the stem, as in Barbareño Chumash (Chumashan; California) (see 4).
‘I saw him.’
‘He told you.’
‘I see you.’
Analogously, the reverse order, P before A, covers the same three types of markers: A and P prefixes, as in Retuarã (Tucanoan; Colombia) (see 5), A and P suffixes, as in Lillooet (Salishan; British Colombia) (see 6), and a P prefix and A suffix, as in Tauya (Madang, Trans-New Guinea; Papua New Guinea) (see 7).
‘He ate it.’
‘He helped me.’
‘She/he saw them.’
In the vast majority of languages which have been assigned value 2 or 3, the A and P person markers exhibit a unique location relative to the verbal stem: both are prefixes, or both are suffixes, or one is a prefix and the other a suffix. There are, however, languages in which the A and P exhibit the same order relative to each other, but vary with respect to their location vis-à-vis the stem. For instance, in Coeur d'Alene (Salishan; northern Idaho), one set of A and P forms is used in the completive and stative aspects, and another in the progressive. The first set of forms are suffixes, the second are prefixes. As shown in (8), the order of the A and P forms remains the same: the P forms precede the A forms.
‘I saw thee.’
‘I am seeing thee.’
Such languages have been classifed under values 2 or 3, depending on whether the order is A > P or P > A, respectively.
Turning to the fourth value, both orders of the A and P, an example of a language displaying both orders of the A and P relative to each other is Ika (Chibchan; northeastern Colombia). In Ika the P forms and the second person singular and first and second person plural A forms are prefixes while the first singular A marker in the past and irrealis is a suffix. The third person forms are zero. In the case of prefixal forms the A precedes the P (9a), but when the A is first person, the order is P before A (9b).
‘Did you pay us?’
‘I saw you.’
In some languages the existence of alternative orders of the A and P relative to each other is dependent on the position of the referents of the markers on the person and/or animacy hierarchies. This is not uncommon among the languages of Australia (e.g. Bininj Gun-Wok, Yukulta, Yulparija). Such hierarchically based ordering of the A and P is also found in Yimas (Lower Sepik-Ramu; Papua New Guinea), in which two hierarchies are actually involved, a person hierarchy (1 > 2 > 3) and a role hierarchy. The latter ranks the P higher than the A in the case of the first and second person, and the A higher than the P in the case of the third person. The higher ranking participant is placed closer to the verb stem. Thus when both participants are third person or the A is first or second person and the P third, we have PA order as in (10).
‘He saw them.’
‘I saw them.’
But when the A is third person and the P first or second, or the A is second person and the P first, we have AP order, as in (11).
‘They saw me.’
‘You saw me.’
A conflict between the two hierarchies which arises when there is a first person A and a second person P is resolved by means of a portmanteau morpheme mpan-/kampan, as illustrated in (12).
-see . . .
‘I saw you.’
In languages in which both the A and P forms occur on the same side of the verb, it is not always possible to determine which part signals the A and which part the P because the two are fused. Consider, for instance, the verbal paradigm in (13), involving the verb /ill/ - ‘to see’ in the present tense in Jaqaru (Aymaran; Peru).
I see you
you see me
you see us
she sees me
she sees us
she see you
you see him
I see him
we see him
she see him
A closer look at the above forms reveals that the second person is always associated with –ma if it is a P and with –ta if it is an A. But beyond this, no further definite segmentation appears to be possible.
As indicated on the map, among the 193 languages which have verbal person markers for both the A and P, there is a preference for placing the A before the P. This preference, however, is not manifested in all areas of the world. The placement of A before P is particularly frequent in Southeast Asia and the Pacific. It is also very common in Africa. Interestingly, all the African languages in the sample that display the opposite order (i.e. PA) are Nilo-Saharan. AP order also prevails over PA in Australia, South America and to a lesser extent in Eurasia.
In contrast to South America, North America exhibits a preference for PA rather then AP order. AP order is not only disfavoured relative to PA but also more geographically restricted. We see that PA languages are distributed all over the continent, while most of the AP languages are in the western and central parts of the continent. In Mesoamerica, on the other hand, PA order is just as common as AP. This is also the case in New Guinea.
Fused A and P verbal person markers are found primarily in the Americas, particularly in North America. They are also characteristic of the languages of the far north of Australia. Fusion of A and P markers is uncommon in Africa and Eurasia.
Languages which have both orders of the A and P markers relative to each other are common only in Australia. In fact, in Australia such languages prevail over those displaying a single order, be it AP, fused or PA. The only other area in which both orders are encountered with some regularity is North America.
The order of the A and P relative to each other is not independent of whether the two are prefixes, suffixes or occur on opposite sides of the verb. AP order is particularly favoured when the two person markers are on opposite sides of the verb; of the 84 languages which have the two person markers on opposite sides of the verb and display a unique order of the A and P, in 75% the A precedes the P. This tends to be the case in SVO and some verb-initial languages. This explains the frequency of AP order in Southeast Asia, the Pacific and Africa. Among the 36 languages in which the A and P are both prefixes, AP and PA orders are very evenly distributed: 19 languages have AP order and 17 PA. This is also more or less the case in the 34 languages in which the A and P are both suffixes: 15 have AP order and 19 have PA order. Fused markers occur somewhat more frequently in prefixal position (12 languages) than in suffixal position (8 languages). This also holds for the languages in the sample manifesting alternative orderings of the A and P; 13 (65%) are prefixing.
One of the explanations that have been advanced for the ordering of person affixes relative to each other is the principle of relevance (cf. Bybee 1985, Bybee et al. 1990), which defines a preference for placing affixes with a greater semantic effect on the head closer to the head than those exerting a lesser effect. The effect of the principle of relevance on the ordering of the A and P is taken to be dependent on the alignment of verbal person forms (cf. Bittner and Hale 1996). The expectation is that in accusative alignment the P should be positioned closer to the verbal stem than the A, and vice versa in ergative alignment. The predicted ordering patterns are thus the ones shown in (14).
Among the 70 languages in which the A and P are either both prefixes or both suffixes and which have a unique ordering of the two relative to each other, only 50 display either accusative or ergative alignment as the sole alignment. The alignment is accusative in 43 languages and ergative in 7. The P is placed closer to the verbal stem than the A in 63% of the languages with accusative verbal alignment, and the A is placed closer to the stem than the P in 71% of the languages with ergative alignment. In all, the predicted ordering occurs in 64% of the relevant languages. Another factor which is seen to influence affix order is the degree of grammaticalization of the morphemes in question. Diachronically older forms, i.e. forms that have undergone more development, are expected to occur closer to the stem than younger forms (see e.g. Bybee et al. 1991: 33). Due to lack of diachronic data, the effects of this factor are much more difficult to test.