Evidentiality in Uzbek and Kazakh

Visual * Non-Visual Sensory * Inference * Assumption * Hearsay * Quotation 
Under this typology, languages divide the task of expressing these information sources between 
two or more morphemes. What is notable about this typology is the natural correlation between 
information source, personal experience, and speaker confidence. That is, a speaker is more 
likely to confirm events described on the basis of visual information, and less likely to confirm 

---

19 
those based on assumption or hearsay. 
In examining Aikhenvald’s typology, we find that the two-morpheme minimum required 
for grammaticalized evidentiality is problematic. In Uzbek and Kazakh, as well as in most other 
languages of Eurasia, only non-firsthand information source is marked, and the use of any other 
term does not necessarily imply firsthand information source. In the division of semantic space 
proposed by Aikhenvald (2004), these languages would appear to fall into type A2: 
Table 6: Aikhenvald's Typology of 2-Term Systems 
I 
Visual 
II 
Sensory 
III 
Inference 
IV 
Assumption 
V 
Hearsay 
VI 
Quotation 
A1 
firsthand 
non-firsthand 
A1 
firsthand 
non-firsthand 
A1 
firsthand 
non-firsthand 
other or  
A2 
 
indirect 
A3 
 
reported 
A4 non-visual 
 
reported 
Into this category A2, Aikhenvald places Turkish, Bulgarian, and Tajik, languages that are 
similar to Uzbek and Kazakh, inasmuch as their indirect terms bear a wide variety of non-
confirmative meaning and that there exists no morpheme in any of these languages that expresses 
firsthand information source. These unmarked terms are the same terms that Aronson (1967) 
and Friedman (1977; 1978) consider confirmative, but, confirmativity as a subvariety of 
STATUS
does not (necessarily) imply firsthand information source. Aikhenvald’s claims that a language 
cannot have only one marker of evidentiality are contradicted not only by her allowance for 
unmarked terms (which are not, technically, part of a true paradigm at all), but also by the non-
existence of forms in Uzbek, Kazakh, and related languages that express firsthand information 
source. 
Within generative linguistics, there exist a variety of opinions as to the status of 
evidentiality. On the basis of data from St’át’imcets, Matthewson et al. (2007), argue that 

---

20 
EPISTEMIC MODALITY
, may encode a two-fold distinction. Languages may choose either to 
encode quantificational force (resulting in a reading of status or modality) or information source.
Under this approach, status/modality and evidentiality are mutually exclusive sets of meaning 
that should not co-occur. As shown in the following chapters, Uzbek and Kazakh may 
simultaneously express status/modality and information source, which renders this theory 
problematic. 
A number of other generative theories posit a dedicated head position for evidentiality.
Cinque (1999) places evidentiality within a broader configuration of moods and modes: 
[Speech Act Mood [Evaluative Mood [Evidential Mood [Epistemological Mode […]]]]] 
Such an approach implicitly assigns evidentiality to a distinct category, albeit a category of 
“Evidential Mood.” It is unclear whether Cinque intends for 
MOOD
here to be a distinct category 
with three sub-types, or whether “mood” is merely a convenient label for certain higher order 
sets of meanings. 
Speas (2004) proposes dedicated heads for a number of pragmatic features, one of which 
is evidentiality. She retains Cinque’s (1999) configurations of heads, but adds four pragmatic 
roles that act as implicit arguments of these heads: 
(4) 
Speech Act Phrase - 
SPEAKER
Evaluative Phrase - 
EVALUATOR
Evidential Phrase - 
WITNESS
Epistemological Phrase - 
PERCEIVER
The configuration of these arguments and their co-indexations produce sets of meanings that are 
realized either as logophoricity or evidentiality. Various types of evidentiality (personal 
experience, direct evidence, indirect evidence, and hearsay) may also be expressed via these co-

---

21 
indexations. 
Many recent works have focused on the languages of Americas, which often express 
evidential meaning in ways very different from those of Eurasia. Furthermore, it is quite 
common to discuss evidential meaning as though it could be separated from the morphemes that 
express it. While Cinque (1999) and Speas (2004) make convincing arguments that evidential 
meaning has a place within a semantic hierarchy, the semantics of evidential morphemes are 
often too complex and too closely associated with 
STATUS
or 
EPISTEMIC MODALITY
for this 
hierarchy to necessarily correspond to the morphosyntactic behavior of these forms.