• Keine Ergebnisse gefunden

6.2 The experimental task

6.3.2 Individual analysis

The group analysis presented above allows for the individuation of general tendencies within groups, while overlooking individual performances. In what follows, I will now present some data based on an individual analysis in order to deepen the understanding of participants’ performance.

Table 6.3 provides an overview of the number of participants in each group who perform above chance-level in the different conditions. Participants complete eight items per condition; accordingly, their performance is classified as 'above-chance' whenever the number of correct answers is equal to six or more.

198

Table 6.3: Number of participants who perform above chance-level (per group and per condition).

CO (=21) PAD1 (=4) PAD2 (=21) PAD3 (=11)

WhoS 21 3 8 3

WhoO 21 4 3 6

WhichS 21 4 13 4

WhichO 21 4 13 3

As expected, all participants in the control group perform above chance-level in all conditions. The same observation holds true also for PAD1, except for one participant who has a poor comprehension of WhoS questions. The behaviour has already been detected at the group analysis level, as a comparison between COs and PAD1 actually points out a significant difference between the two groups with respect to the comprehension of WhoS questions.

Also the individual analysis for participants in the PAD2 group confirms data found at the group level. Only three patients (out of twenty-one) perform above chance-level in correspondence to WhoO questions. The specific impairment on the object condition with bare Wh-element is therefore singled out at both levels of analysis. In contrast, thirteen patients perform well on both conditions characterized by a lexically-restricted Wh-element, namely WhichS and WhichO.

Finally, the individual data from patients in the PAD3 group offer interesting insights. Although at the statistical level no significant difference is to be found across conditions in PAD3, the distribution of above-chance performances is uneven. Half of the participants (six out of eleven) performs above chance on WhoO questions, while the same result is not obtained in the other three conditions: three participants perform above chance on WhoS, four perform well on WhichS and three are accurate on WhichO. The image is rather fragmented, with an unequal distribution of above-chance performances across conditions.

The following sections draw further observations from the individual analysis on PAD2 and PAD3 participants.

6.3.2.1 Individual analysis: PAD2

In what follows, contingency tables will now help us to reach a better understanding of the behaviour of participants in the PAD2 group, the only one in which asymmetries

199

among conditions were detected. In particular, I will take into consideration comparisons between conditions that already emerged as significant at the group level: WhichS vs WhoO (Wilcoxon related samples, Group 2, WhichS vs WhoO: Z=-3.428, p=.001), WhichO vs WhoO (Wilcoxon related samples, Group 2, WhichO vs WhoO: Z=-2.366, p=.018), and WhichS vs WhoS (Wilcoxon related samples, Group 2, WhichS vs WhoS:

Z=-3.318, p=.001).

Table 6.4 presents results for the first relevant comparison, namely the one between WhichS and WhoO. The contingency table compares the performance of each patient in the two conditions. In the first row it shows the number of participants who perform well both on WhichS(+) and on WhoO(+), and then the number of participants who perform well on WhichS(+) but not on WhoO(˗). In the second row we can see the number of participants that perform well on WhoO(+) but not on WhichS(˗) and the number of participants who do not perform well in either of the two conditions (WhichS(˗) and WhoO(˗)).

Table 6.4: Target comprehension of WhichS and WhoO in PAD2.

WhoO + WhoO ˗

WhichS + 2 11

WhichS ˗ 1 7

Participants mainly distribute across two conditions: eleven patients retain a good understanding of WhichS questions in combination with a poor performance on WhoO;

alternatively, seven participants do not reach above-chance comprehension in either of the two conditions. The converse patterns are very rare, as only two participants perform well on both conditions and only one performs well on WhoO but not on WhichS.

Altogether, data in table 6.4 confirm that participants perform better on WhichS questions than on WhoO questions.

Table 6.5 shows a very similar pattern in correspondence to the comparison between WhichO and WhoO questions. The specific impairment on WhoO questions is confirmed again by the fact that half of the participants (ten out of twenty-one) perform accurately on WhichO but not on WhoO. Eight patients perform at chance-level in both conditions, while three perform well on both. None of the participants in the PAD2 group has a good comprehension of WhoO in combination with a poor comprehension of

200

WhichO. Individual data therefore comfirm that PAD2 participants are overall more accurate on WhichO questions than on WhoO questions.

Table 6.5: Target comprehension of WhichO and WhoO in PAD2.

WhoO + WhoO ˗

WhichO + 3 10

WhichO ˗ 0 8

Finally, Table 6.6 offers a comparison within the subject conditions. The group analysis revealed that WhichS questions are statistically better understood than WhoS questions.

The claim holds true at the individual level too because the reversed pattern (namely good comprehension of WhoS and poor comprehension of WhichS) emerges only in one participant. However, it should also be pointed out that the other twenty participants evenly distribute across the three remaining possible patterns of performance. Seven patients have an accurate comprehension of both conditions; seven comprehend poorly both, and six have a good comprehension of WhichS questions but not of WhoS questions.

Table 6.6: Target comprehension of WhichS and WhoS in PAD2.

WhoS + WhoS ˗

WhichS + 7 6

WhichS ˗ 1 7

Altogether, data from the contingency tables confirm the observations already drawn at the group level, in particular with respect to the comparison between WhichS and WhoO questions and the one between WhichO and WhoO questions. For what concerns the last comparison, a strong preference for WhichS over WhoS is not completely borne out from the individual analysis because PAD2 participants evenly distribute across different patterns of performance.

6.3.2.2 Individual analysis: PAD3

Table 6.3 shows the number of PAD3 participants who perform above chance in the different conditions: three out of eleven PAD3 participants perform above chance on WhoS questions, six perform above chance on WhoO questions. As for WhichNP