Map 103A represents the distribution of third person zeroes among verbal person markers of the sole argument of an intransitive clause (i.e. of the S argument). I have chosen to depict the person marking of the S rather than of either of the transitive arguments, as the verbal person markers of the latter are sometimes fused with each other (see Chapter 104), which makes it difficult to decide which one has a zero realization. The zero realization of third person S markers is captured by means of six values:
| Go to map | ||
|---|---|---|
| Value | Representation | |
 |
No person marking of the S | 96 |
 |
No zero realization of third person S forms | 181 |
 |
Zero realization of some third person singular forms | 21 |
 |
Zero realization of all third person singular S forms | 45 |
 |
Zero realization of all third person forms/No third person S forms | 36 |
 |
Zero realization only of third person non-singular | 1 |
| Total: | 380 | |
Most of the languages subsumed under the first value have no verbal person marking at all. Some may, however, evince verbal person marking of just the P, as is the case in Palikur (Arawakan; French Guiana) and Yapese (Austronesian; Caroline Islands) (see Chapter 102, value 3).
Given that person markers typically combine person distinctions with those of language’s number and, less often, gender, the size of the inventory of third person S forms depends on how many number and/or gender distinctions it exhibits in the third person. Thus Kilivila (Western Oceanic; Trobriand Islands, Papua New Guinea), which has a simple singular/plural opposition in the third person S, possesses just two forms. Tanimbili (Remote Oceanic; Utupua Island) with its singular/dual/plural opposition has three forms (this language is not included in the sample). And Larike (Central Malayo-Polynesian; Ambon Island), which exhibits a singular/dual/trial/plural contrast plus a gender opposition in the singular and plural, has six basic forms:
(1) Kilivila (Senft 1986: 46-47)
sg
du
pl
1incl
ta-
ta- -s
1excl
a-
ka-
ka- -si
2
ku-
ku- -si
3
i-/e-
i- - si
(2) Tanimbili (Tryon 1994: 628)
sg
du
pl
1incl
si
misu
1excl
nyi-
me-
misu
2
nu-
mwa
muku-
3
i-
ŋgi(li)-
ŋgu-
(3) Larike (Laidig 1993: 321)
sg
du
tri
pl
1incl
itua-
itidi-
ite-
1excl
au-
arua-
aridu-
ami-
2
a-/ai-
irua-
iridu-
imi-
3.hum
ma-/mei-
matua-
matidu-
mati-
3sg.nonhum
i-
iri-
Languages may also have separate verbal person paradigms dependent on tense, aspect, mood and even polarity (e.g. Salinan). Needless to say, this too may result in a greater number of third person forms. For instance, in Amele (Trans-New Guinea; Madang, Papua New Guinea) there are eight classes of person markers, used in different tenses, aspects and modes. As shown in (4), though in the dual and plural the second and third person are non-distinct, there are still 12 different third person forms.
(4) Amele (Roberts 1987: 277-278)
1
2
3
4
5
6
7
8
1sg
-ig
-ig
-ig
-ig
-min
-m
-em
-em
2sg
-g
-g
-g
-g
-m
-m
-em
-em
3sg
-i
-Ø
-igi
-i
-b
-b
-n
-Ø
1du
-w
-w
-w
-w
-hul
-h
-h
-h
2/3du
-si
-si
-was
-was
-bil
-b
-sin
-sin
1pl
-q
-q
-q
-q
-mun
-m
-m
-m
2/3pl
-eig
-eig
-qap
-w
-bil
-b
-ein
-ein
The second value in Map 103A represents languages in which all the third person forms that the language distinguishes are overtly realized, as is the case in Kilivila, Tanimbili and Larike, though not Amele.
As Amele illustrates, zero realization is primarily associated with the third person singular as opposed to the non-singular categories. Languages seen as having a zero for the third person singular fall into two subtypes. To the first subtype, represented by value 3, belong languages in which only some realizations of the third person singular are zero while other realizations are not. Amele is a case in point. Lango (Nilotic, Nilo-Saharan; Uganda), is another. Lango has two paradigms of S (and A) person markers, an “A” set used in the perfective, habitual and subjunctive and a “B” set used in the progressive. As shown in (5), the third person singular has a zero realization only in the habitual.
(5) Lango (Noonan 1992: 91)
sg
pl
Set A
Set B
Set A
Set B
1
ǎ-
â-
ǒ-
ô-
2
ǐ-
î-
ǐ-wùnú
î-wunu
3
ò-(perf,subj)
à-
ɪ̌
ɪ̌-
Ø-(hab)
To the second subtype, represented by value four, belong languages in which the third person singular is always zero, such as Chepang (Kiranti, Tibeto-Burman; Nepal).
(6) Chepang (Caughley 1982: 54-55)
sg
du
pl
1incl
-ŋə-cə
-ŋ-sə
1excl
-ŋa
-təyh-cə
-təyh-ʔi
2sg
-naŋ
-naŋ-jə
-naŋ-sə
3sg
Ø
-cə
-ʔi/sə
The fifth value represents languages which have overt verbal person forms for the first and second person but not for the third person (neither singular nor plural), as is the case in So (Kuliak, Nilo-Saharan; Uganda).
(7) So (Carlin 1993: 79)
sg
pl
1incl
-(i)ine
1excl
-(ɪ)sa
-(i)ise
2sg
-(ɪ)ba
-(i)ide
3sg
Ø-
Ø-
Another way of describing such languages is to say that they exhibit no third person forms at all. The description in terms of zero realization is chosen here because we are comparing these languages with others showing values 2 to 6.
The final, sixth value covers languages which have zero marking of the third person solely in some non-singular number. The only instance of such zero marking that I am aware of (apart from languages such as English or Trumai which have overt person marking only in the third person singular) involves gender distinctions as well: this is Barasano (Tucanoan; Brazil and Colombia), where there is a zero marker for third person plural inanimates.
(8) Barasano (Jones and Jones 1991: 73-74)
1sg
-ha
1pl
-ha
2sg
-ha
2pl
-ha
3sg.m
-bõ
3pl.anim
-bã
3sg.f
-bĩ
3pl.inan
-Ø
3sg.inan
-ha
Map 103A illustrates that overt realization of third person S markers is the cross-linguistic norm, as nearly two-thirds of the languages in the sample display such marking. What we also see, however, is that there are considerable areal differences with regard to the presence of third person zeroes. In Southeast Asia, the Pacific and New Guinea, overt third person forms are clearly favoured over zeroes. This may also be observed in Africa and somewhat less clearly in Eurasia. In Australia and the Americas, on the other hand, there are slightly more languages with some form of zero marking in the third person than languages with only overt third person forms. Within the Americas, zero marking is more common in North America than in South America, and is particularly frequent in Mesoamerica.
Significant areal differences may also be observed in regard to the type of zero marking displayed. In North America complete lack of third person markers is encountered more frequently than a zero just in the third singular or just a third person zero allomorph. In Australia and Eurasia, the converse is the case. And in South America the instances of zero marking are more or less evenly distributed over three types of zero marking, namely absence of third person forms, zero in third singular and only a zero allomorph.
Several lines of explanations have been advanced for the existence of third person zeroes. Some scholars treat the issue as an instance of loss and/or reanalysis of previously overt markers, others as a case of failure of the third person markers to develop.
The most widely accepted explanation of the first kind attributes the zero marking of third person verbal forms to the principle of economy. The principle of economy favours one of the exponents of a paradigm being non-overt. That this should be the third person rather than the second or the first is seen to follow from the higher frequency of third person verbal forms in discourse than of second or third person forms. The effect of frequency on the form of linguistic expressions has long been recognized and is captured in what is commonly referred to as Zipf's law, i.e. “high frequency is the cause of small magnitude” (Zipf 1935: 29). In other words, it is the tendency for speakers to shorten the linguistic expressions used most commonly that motivates the existence of zero realizations of the third person. For some discussion of this position, see Haiman (1985) and Croft (1990).
An alternative explanation for third person zeroes has been developed by Koch (1995), who argues that there is a strong tendency in languages to reinterpret third person verbal forms as part of the stem or as tense markers. This tendency he claims is due to the pressure of iconicity, i.e. the preference for morphological structure to mirror cognitive structure. Adopting the view, most strongly articulated by Benveniste (1971), that the third person is cognitively a “non-person”, and therefore unmarked vis-à-vis the first and second person, he argues that it should therefore be also unmarked morphologically.
In contrast to the above two explanations, Ariel (2000) attributes third person zeroes to the fact that third person forms, unlike first and second person ones, simply did not develop. In accessibility theory, which constitutes the context of her explanation, the coding of discourse referents is seen to reflect speakers' assumptions as to the degree of accessibility of the referents in the memory store of the addressee. The claim is that the higher the accessibility of the referent, the more attenuated its encoding. Verbal person markers are considered to be high-accessibility coding devices, second only to zero and reflexives. Thus, according to accessibility theory they should be used as the sole means of referent encoding, only in the case of highly accessible referents. According to Ariel, first and second person pronouns meet this criterion, but third person pronouns do not. In other words, the referents of third person pronouns tend to be insufficiently accessible to warrant phonological reduction, cliticization and affixation. Consequently, third person verbal markers, in contrast to first and second person ones, tend not to arise. A critical assessment of this view is provided in Siewierska (2004), based in part on the fact that it is primarily the third person singular that is non-overt, rather than the third person in general, as the accessibility explanation would lead one to expect.