Neurolinguistic & psycholinguistic investigations on evidentiality in Turkish

2.3.2.
 
Source identification
Agrammatic participants’ source identification was worse than that of 
NBDs, who also made errors on this task. However, for the NBDs, the 
errors were equally distributed over the conditions (all comparisons 
p 
> 
.05). The mean numbers and ranges of correctly answered source 
identification questions are given in Table 2.4. 
 
Table 2.4.
Mean number of correctly identified categories of information 
source by agrammatic and NBDs. Ranges in brackets (max=10 per 
condition)
Source for 
Direct perception 
Inferential 
Reportative 
Agrammatic 
7.71 (5 - 10) 
2.29 (0 - 5) 
2.71 (0 - 6) 
Control 
8.43 (6 - 10) 
7.57 (6 - 10) 
8.00 (4 - 10) 

---

44 
The agrammatic speakers were significantly worse in judging the 
correct information source than NBDs for the inferential condition 
(
Wilcoxon, z = 
-3.14, 
p = 
.001), and for the reportative condition (
Wilcoxon, 
z = 
-2.96, 
p = 
.001). However, they did not differ from the NBDs in 
recognition of the source for the directly perceived events (
Wilcoxon, z = 
-
.716, 
p = .
535). Comparisons of agrammatic speakers’ performance across 
the three conditions confirmed that the direct perception as information 
source is easier to identify than the inferential (
Wilcoxon, z = 
-2.37, 
p = 
.018
), and the reportative (
Wilcoxon, z = 
-2.37, 
p = .018
). Inferred and 
reported events as information source conditions were both difficult to 
identify, and there was no significant difference between them (
Wilcoxon, z 
= 
-.542, 
p = .
588).
Table 2.5 demonstrates the total number of source errors and 
substitutions. The agrammatic participants made two types of source 
identification errors: (1) no-recognition (i.e., no answer or ‘I do not know’ 
responses), and (2) source misattributions. We observed three types of 
source misattributions types: (a) thought to have heard, (b) thought to have 
inferred, and (c) thought to have seen.