Henry S. Cheang and Marc D. Pell Sarcasm

Recognizing sarcasm without language 27
Humor
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Cantonese
English
Language of Tokens
Pr
op
or
tio
n C
orr
ec
t
Native
Cantonese
Listeners
Native
English
Listeners
Figure 1c.
Neutrality
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Cantonese
English
Language of Tokens
Pr
op
or
tio
n C
orr
ec
t
Native
Cantonese
Listeners
Native
English
Listeners
Figure 1d.

28 Henry S. Cheang and Marc D. Pell
6.2 
Cross-linguistic analysis of accuracy data for each attitude
For sarcasm, the 2 x 2 ANOVA yielded significant main effects of LISTENER 
GROUP (F(1, 38) = 37.18, p < .001) and LANGUAGE (F(1, 38) = 53.97, p < .0001) 
and a significant interaction of these two factors (F(1, 38) = 189.88, p < .0001). 
The interaction was explained by the fact that the two listener groups were al-
ways significantly more accurate at identifying sarcastic sentences spoken in their 
native language than in a foreign language. Cantonese listeners recognized the 
Cantonese exemplars of sarcasm significantly better than English listeners, and 
English listeners recognized the English exemplars of sarcasm significantly better 
than Cantonese listeners. These patterns are illustrated in Figure 1a.
For sincerity, the ANOVA yielded main effects of LISTENER GROUP 
(F(1, 38) = 11.66, p = .0002) and LANGUAGE (F(1, 38) = 17.34, p = .0002), as well 
as a significant interaction between the two factors (F(1, 38) = 68.74, p < .0001). 
Posthoc tests established that participants in each listener group were always sig-
nificantly better at identifying sincerity when spoken in their native language when 
compared to the foreign language. Expressions of sincerity in English were recog-
nized more accurately by English than Cantonese listeners, whereas there were 
no significant group differences in the recognition of sincerity from Cantonese 
(although there was a trend for Cantonese listeners to be more accurate in this 
condition, see Figure 1b). For humor, the ANOVA produced a significant main 
effect of LANGUAGE (F(1, 38) = 38.88, p < .0001) and a significant LANGUAGE 
by LISTENER GROUP interaction (F(1, 38) = 4.15, p = .0487). Surprisingly, the 
interaction demonstrated that Cantonese listeners identified humor significantly 
less accurately when listening to Cantonese versus English tokens. As one might 
expect, English listeners were better at recognizing humor in English sentences 
versus Cantonese sentences (see Figure 1c).
Finally, for neutrality there was a significant main effect of LANGUAGE 
(F(1, 38) = 4.62, p = .0381) and a significant interaction between LANGUAGE and 
LISTENER GROUP (F(1, 38) = 39.90, p < .0001). The interaction was explained by 
the fact that like sincerity, neutral sentences spoken in English facilitated the per-
formance of the English listeners rather than the Cantonese listeners. There were 
no significant differences between the listener groups in the recognition of neu-
trality from Cantonese utterances (Figure 1d).

tải về 1.21 Mb.

Chia sẻ với bạn bè của bạn: