Chapter Alignment of Verbal Person Marking

by Anna Siewierska

1. Defining the values

Languages differ not only in regard to the number and nature of the arguments displaying verbal person marking (see Chapter 102), but also in the alignment of the person markers. The term alignment may be intuitively understood as reflecting how the two arguments of the transitive verb, the agentive argument (A) and the more patient-like argument (P), align with the sole argument of the intransitive verb, the S. In Map 100A the alignments of verbal person markers are represented by the following six values:

Values of Map 100A. Alignment of Verbal Person Marking
Go to map
 ValueRepresentation
Neutral alignment (no verbal person marking) 84
Accusative alignment 212
Ergative alignment 19
Active alignment 26
Hierarchical alignment 11
Split alignment 28
Total: 380

Neutral alignment corresponds to absence of verbal person marking.

By accusative alignment is understood a common treatment of the S and A and a different treatment of the P, as in (1) from Tawala (Oceanic; Papua New Guinea), where the S and A markers are prefixes and the Pmarkers are suffixes.

(1) Tawala (Ezard 1997: 289, 116) 

a.

i-bowi-ye-ya

 

3sg.A-deny-tr-3sg.P

 

‘He denied him.’

b.

apo

i-na-nae

 

fut

3sg.S-pot-go

 

‘He will go.’

Observe that the form of the third person singular marker is i-  for both the A in (1a) and the S in (1b), but –ya  for the P in (1a).

In ergative alignment, by contrast, it is the S and P that are treated alike, while the A is treated in another way. This is illustrated in (2) from Konjo (Western Malayo-Polynesian; South Sulawesi).

(2) Konjo (Friberg 1996: 141, 140) 

a.

na-peppe’-i

Amir

asung-ku

 

3.A-hit-3.P 

Amir 

dog-1 

 

‘Amir hit my dog.’ 

b.

a'-lampa-i

Amir

 
 

intr-go-3.S

Amir

 
 

‘Amir goes.’ 

Note that the 3rd person markers of the P in (2a) and of the S in (2b) are the same and are distinct from that of the A in (2a).

In active alignment there are two patterns of identification of the S: sometimes it is treated like the A and sometimes like the P, depending on a range of semantic factors such as eventhood, performance/effect, instigation, control and significant affectedness (Mithun 1991). A language manifesting active alignment of verbal person forms is Koasati (Muskogean; Louisiana).

(3) Koasati (Kimball 1991: 189, 204, 120, 118) 

a.

okolcá

hóihca-li-halpí:s

 

well

dig-1sg.A-ability

 

‘I can dig a well.’ 

b.

tálwa-li-mp

 

sing-1sg.S-hearsay

 

‘(He says) I sing.’

c.

ca-pa:-batápli-t

 

1sg.P-loc-hit-pst

 

‘He hit me on the back.’

d.

ca-o:w-ílli-laho-V

 

1sg.S-loc-die-irr-phr.term

 

‘I will drown.’

Observe that the form of the 1st person singular S marker in (3b) is the same as that of the A marker in (3a), while the form of the S in (3d) is the same as that of the P in (3c). 

In tripartite alignment each of the three arguments S, A and P is treated differently. Tripartite verbal person marking is rare, and when it occurs it is manifested only for part of the person paradigm. For instance, in Yukulta (Tangkic; Queensland, Australia) it is exhibited only by second person singular and all first person forms.

(4) Yukulta (Keen 1983: 239, 237, 215) 

a.

waranaŋkuḷu-ka-ti

 
 

go.neg.des-1sg.S-pres

 
 

‘I’m trying to go.’

b.

ṭalmata-ŋa-nti

ŋita

 

chop.ind-1sg.A-fut

wood

 

‘I’ll chop the wood.’

c.

tʸinkaka-nki

ŋata

 

follow.imp-1sg.P

me

 

‘Follow me.’ 

Finally, in hierarchical alignment the treatment of the A and P is dependent on their relative ranking on the referential and/or ontological hierarchies. Whichever argument is the higher ranking receives special treatment, the details of which vary from language to language. The following example is from Plains Cree (Algonquian ; Canada), in which person marking is determined by the hierarchy 2 > 1 > 3.

(5) Plains Cree (Wolfart and Carroll 1981: 24) 

a.

ki-wāpam-i-n

 

2.A-see-dir-1

 

‘You see me.’

b.

ki-wāpam-iti-n

 

2.P-see-inv-1.A

 

‘I see you.’

c.

ki-wāpam-ikw-ak

 

2.P-see-inv-3pl.A

 

‘They see you.’

We see in (5) that the second person is always marked by a prefix (ki- ) irrespective of whether its referent is in A function as in (5a) or in P function as in (5b-c). In Cree, as in many other languages which have hierarchical verbal person marking, if the higher ranking argument is a P rather than an A, a special inverse marker occurs on the verb, iti  in (5b) and ikw  in (5c). As we see in (5a), Cree also has a special marker for when the A outranks the P, called a direct marker. This is rather unusual.

In the examples of the various types of alignments given above, all three constituents, the S, A and P, were overtly marked. However, this need not be the case. For instance, accusative alignment may involve overt marking of the S and A and no marking of the P, or, alternatively, overt marking of the P and no marking of the S and A. The same in principle holds for the other types of alignment. 

While in many languages the verbal person markers always manifest one type of alignment, in others several alignments may be observed. Such splits in alignment may involve two or more non-neutral alignments, or a non-neutral alignment and neutral alignment. Since the term neutral alignment  is interpreted here as meaning absence of verbal person marking, these two types of splits have been treated differently. Languages displaying splits involving two or more non-neutral alignments - for instance, accusative and ergative, or active and accusative, or ergative and tripartite - have all been classified in terms of value 6, split alignment. Internal splits in alignment involving combinations of neutral and non-neutral alignment, on the other hand, have been grouped under the non-neutral alignment (thus ignoring the neutral alignment).

Both types of splits in alignment, those involving non-neutral alignments and those involving a non-neutral and a neutral alignment, may be conditioned by a range of factors. One of the factors which commonly determines alignment is person. Quite often the alignment of the first and second person differs from that of the third. For example, in many languages with active alignment, this alignment is confined to the first and second person while the third person displays neutral alignment. This is in fact the case in, mentioned above, Koasati as well as in Amuesha (Arawakan; Peru), Tlingit (Na-Dene; British Columbia) and Wichita (Caddoan; Oklahoma). The converse situation, neutral alignment of the first and second person and non-neutral of the third, is also attested, but only rarely. In Trumai (isolate; Upper Xingu, Brazil) there is ergative alignment in the third person, with S and P marked by means of a verbal enclitic e(n), but no marking of either first or second person. Compare (6a-b) with (6c-d).

(6) Trumai (Guirardello 1999a: 95, 99, 29) 

a.

iyi

waţkan-e

   
 

pcl

cry-3sg.S

   
 

‘She cried.’ 

b.

hai-ts

ka-in

iyi

midoxos-e

 

I-erg

pst-foc

prt

call-3sg.P

 

‘I called him.’

c.

ha

pita

ka-in

 
 

I

go.out

pst-foc

 
 

‘I went out.’ 

d.

ka’natl-ek

ha

midoxos

 
 

that-erg

     
 

‘That one called me.’ 

A similar split of first and second person as opposed to third person involving neutral and non-neutral alignment, though this time accusative rather than ergative, is also found in a restricted way in English. Note that the –s  marking of the verb in (S)he come-s  and (S)he like-s apples  is a manifestation of accusative verbal alignment restricted to the third person.

Person splits involving two non-neutral alignments are also to be found. One case in point is, mentioned above, Yukulta, in which the first person singular and nonsingular and the second person singular display tripartite alignment, contrasting with the accusative alignment of the second person non-singular and third person. 

Another common factor influencing the alignment of verbal person marking is tense and aspect. For instance, in Itzaj (Mayan; Guatemala) the verbal person markers align ergatively in the completive aspect and in dependent clauses, but accusatively in the non-completive aspect. Still other factors which may influence alignment include mood, polarity, humanness, animacy, definiteness, word order and main vs. subordinate clause status. Typically such factores distinguish between non-neutral and neutral alignment. Thus, for example, Estonian, Manambu (Sepik; Papua New Guinea), Sentani (Sentani family; Papua, Indonesia) and Tariana (North Arawakan; Brazil) have non-neutral alignment in positive clauses but neutral in negative ones. In Ika (Chibchan; northeastern Colombia), Koegu (Surmic, Nilo-Saharan; Ethiopia) and Siuslaw (Siuslawan; Oregon) the normal non-neutral alignment contrasts with neutral alignment in imperatives. And in the following languages there is neutral alignment in relative clauses but non-neutral in main clauses: Barasano (Tucanoan; Colombia and Brazil), Gimira (Omotic, Afro-Asiatic; Ethiopia), West Greenlandic (Eskimo), Kobon (Trans-New Guinea; Papua New Guinea), Limbu (Kiranti, Tibeto-Burman; Nepal), and Turkish (Cristofaro 2003). As these examples suggest, the neutral alignment tends to occur in non-declaratives, negatives or subordinate clauses, the non-neutral in main, positive declarative clauses. Occasionally, however, the reverse is the case. For instance, in Nivkh (isolate; Sakhalin Island, v) accusative alignment is confined to imperatives. The classification of the alignments of verbal person markers depicted in Map 100A is based only on the alignments found in main, positive, declarative clauses.

2. Geographical distribution

The most common alignment of verbal person markers is accusative. It occurs in around 55% of the languages in the sample. It is not only the dominant alignment overall but also in every major geographical area. The next most common alignment is neutral, occurring in just over one fifth of the languages in the sample. Neutral alignment is found mainly in West Africa, the Caucasus, and South and Southeast Asia. The frequency of each of the other alignment types in the sample is under ten per cent, with split alignment being the most common (8%) and hierarchical the least (3%). 

The vast majority of the split alignments involve ergative/accusative splits. Such splits are not characteristic of any particular area or areas. They are, however, more common in areas with languages manifesting some form of (unsplit) ergative alignment, be it of verbal person forms or of free pronouns (see Chapters 98-99). Ergative alignment of verbal person forms is attested mainly in South America, Southeast Asia (Taiwan, Sulawesi and the Philippines) and Mesoamerica (the Mayan languages). It is also found in Eurasia in several Caucasian languages and in Basque, as well as in the north of North America, where the Eskimo-Aleut languages are spoken. Ergative alignment of verbal person markers is absent in New Guinea, Australia and the Pacific.  One does, however, encounter sporadic instances of split accusative/ergative alignment in these areas. The area virtually devoid of ergativity is Africa. Turning to active alignment, most instances of active alignment come from the Americas. Active alignment also occurs in Southeast Asia and New Guinea. It is not attested in Australia or Africa nor, apart from the isolate language Ket, in Eurasia. Hierarchical alignment, like active, features primarily in the languages of the Americas. The only other area in which it is in evidence, more than just sporadically, is among the Tibetan languages of India and Nepal.

The above data on the distribution of the alignment of verbal person forms correspond closely to the data presented in Nichols (1992) based on a smaller sample of 174 languages.