Evidentiality in Uzbek and Kazakh

napravi 
(e) napravil 
Macedonian 
napravi 
napravil 
Turkish 
yaptı 
yapmış 
Under this analysis, the simple [+
CONFIRMATIVE
]
past is actually the more marked of the two 
forms, as it bears confirmative meaning. This confirmative meaning is evident in that the only 
place the simple past may not occur is in an antiaffirmative context: 
(5)
Se somnevam deka toj go *napravi/napravil toa. (Macedonian) 
I doubt that he did it. 
(Friedman 1978, 109)
(6)
Inanmıyorum ki o adam bunu *yaptı/yapmış. (Turkish) 
I don’t believe that he did it. 
(Friedman 1978, 110)
A further feature characteristic of the expression of evidentiality in Balkan languages is that the 
double marking, or copularization of the perfect results in the combination of these morphemes 
expressing pure non-confirmativity. In Turkish, for example, the form yapmış in Table 7 is 
ambiguous with regard to confirmativity and may express hearsay, doubt, or admirativity, or may 
simply be employed as a perfect. When speakers wish to disambiguate the confirmative status of 

---

24 
such forms, they have the option of doubly marking the perfect -mIş, resulting in the verb form 
yapmışmış. These forms are marked as non-confirmative ([-
CONFIRMATIVE
])
and therefore 
always express non-firsthand information source or admirativity. 
While evidentiality, or evidential-like phenomena are usually considered a defining part 
of the Balkan Sprachbund, similar expressions of evidentiality and related meanings are found 
throughout much of Eurasia. Haarmann, for example, proposed a Eurasian isogloss characterized 
by an “indirect experience-form” that encompassed Turkic, Mongolic, Tungusic, Uralic, and 
some Caucasian and Paleo-Siberian languages (1970). Friedman (1979) provided a more 
detailed and theoretical account, focusing on Balkan Slavic, Albanian, Azerbaijani, Turkish, 
Georgian, Tajik, and Avar. Comrie (2000, 1-2) also noted this areal phenomenon, noting that 
evidentiality in Eurasia (encompassing Turkic, Tungusic, Iranian, eastern Uralic, Balkan, and 
Caucasian languages) lacks some of the characteristics associated with evidential systems 
elsewhere, such as fine-tuned distinctions based on source of information. He further notes that 
language contact may have played a role in spreading evidentiality among these languages. And 
Aikhenvald (2004, 290) identifies a larger area within Eurasia characterized by a small evidential 
system located within a “largish ‘evidentiality belt’” spreading across the Balkans, the Caucasus, 
and Central Asia into Siberia. She and others have suggested that the influence of the Turkic 
langauges is responsible for the spread of this feature. Kehayov (2008) broke down this larger 
evidentiality belt into four regions: the Balkans, the eastern Baltic, the Caucasus, and the Volga-
Kama area, while acknowledging that more work is needed on Siberia and Central Asia. 
It is difficult to precisely define the borders of the Eurasian evidentiality belt, in part 
because languages at the periphery often display only some of the characteristic features and 
because many languages within this belt are poorly documented, especially in regard to 

---

25 
evidentiality. In Western Eurasia, the borders of the belt are roughly co-terminous with regions 
where Turkic languages are or have been spoken; in the east, poor documentation of evidentiality 
in Mongolic and Tungusic leave it unclear whether it is only Turkic that has spread evidentiality, 
or whether it is an Altaic feature. 
A number of South Asian languages have grammaticalized evidentiality (Bashir 2006); 
this may represent the influence of Turkic on these languages, or it may be an independent 
development, possibly influenced by Tibetan. While evidentiality in Tibetan has been well-
studied, the type of evidentiality found in Tibetan and in a number of neighboring languages is 
quite different from that of the rest of this belt. In Tibetan, the expression of evidentiality (and 
related notions, such as admirativity) is closely tied to the expression of person and volitionality 
(DeLancey 2001). Turkic languages spoken in this Tibetan region, such as Western Yugur, are 
no longer typified by the characteristics described above for the Eurasian belt but base their 
expression of evidentiality in the Tibetan formula (Roos 2000). 
Although Kehayov includes the Baltics in his version of the Eurasian belt, the expression 
of evidentiality in these languages also differs from the rest of Eurasia. In the three Baltic 
languages considered by Kehayov (2008), Estonian, Lithuanian, and Latvian, evidentiality and 
related notions are expressed by either the present participle or by an infinitive, rather than by a 
past-tense-denoting form similar to Turkish -mIş or Uzbek -gan. Furthermore, there exists a 
large distance between the Baltics and the rest of the Eurasian evidentiality belt. This suggests 
that the Baltics should be considered only marginal in this belt. While it is possible (and, in fact, 
quite likely) that the original languages of this gap between the Baltics and the rest of the 
Eurasian belt had some means of expressing evidentiality, this region is now almost exclusively 

---

26 
Russian-speaking. It is important to bear in mind that the spread of Russian has left large gaps in 
what may have once been a larger belt. 
Keeping in mind these many gaps, I present in Figure 1 a rough approximation of the belt 
and its aproximate
borders, with special reference to the languages and regions mentioned in the 
text. 
Figure 1: The Eurasian Evidentiality Belt 
Of particular interest for further study is the northeastern edge. Little has been written on 
evidentiality in the Mongolic or Tungusic languages; so it is difficult to precisely determine the 
eastern edge of the belt. Although, as I show below, the status of Altaic as a valid family is 

---

27 
dubious, the further study of evidentiality in Mongolic and Tungusic would greatly enrich our 
knowledge of Altaic as a Sprachbund.