This third research question addresses whether agency is a factor affecting the acquisition of the English progressive aspect. According to Givón’s thematic hierarchy, it was predicted that progressives with an agentive subject would be easier to acquire. To investigate the agency effects, each of the five major verb types was further classified into agentive and non-agentive types. This section presents and discusses the subjects’ performances on agentive and non-agentive types in the two tasks.
4.3.1 Overall Findings
Tables 4-13 and 4-14 present the overall results for agentive and non-agentive types in the AJ Task and the BF Task, respectively.
Table 4-13 The Overall Results for Agentive and Non-agentive Types in the AJ Task
Agency M SD t p-value
Agentive 2.83 0.41 6.678 .000
Non-agentive 2.51 0.48
Table 4-14 The Overall Results for Agentive and Non-agentive Types in the BF Task
Agency M SD t p-value
Agentive 0.69 0.24 6.687 .000
Non-agentive 0.54 0.24
In the AJ Task, the participants as a whole gave significantly higher ratings to sentences with an agentive subject (2.83) than to those with a non-agentive subject (2.51). Although the mean ratings for the two types ranged from 2 (somewhat acceptable) to 3 (acceptable), the distribution of acceptability ratings showed that our subjects preferred progressives with an agentive subject to a non-agentive one. The agentive types were acceptable to 37% of our subjects and strongly acceptable to 30%
of our subjects. (Please refer to Table (v) in Appendix E for the distribution of ratings.) That is, near 70% of our subjects showed high acceptability to the agentive types.
Furthermore, only 14% considered the agentive types unacceptable. For the non-agentive types, however, the percentage of “Acceptable” and “Very acceptable”
lowered to 33% and 21%, respectively. Moreover, the rate of unacceptability increased to 25%. In a nutshell, the agentive types were considered acceptable by more subjects than the non-agentive types. In the BF Task, our subjects scored significantly higher in the agentive types (0.69) than in the non-agentive types (0.54), indicating that the agentive types were easier to acquire.
The overall results of the two tasks provided clear evidence that agency played a role in the L2 acquisition of the English progressive aspect. The performance of the individual groups was in conformity with the overall results, as shown in Tables 4-15
and 4-16 below:
Table 4-15 Each Group’s Scores for Agentive and Non-agentive Types in the AJ Task
Group Agency M SD t p-value
L Agentive 2.75 0.42 2.148 .045
Non-agentive 2.54 0.52
LI Agentive 2.62 0.33 3.839 .001
Non-agentive 2.28 0.4
HI Agentive 2.97 0.39 3.688 .002
Non-agentive 2.54 0.44
H Agentive 2.98 0.41 3.912 .001
Non-agentive 2.68 0.48
Table 4-16 Each Group’s Scores for Agentive and Non-agentive Types in the BF Task
Group Agency M SD t p-value
L Agentive 0.66 0.24 6.272 .000
Non-agentive 0.36 0.24
LI Agentive 0.58 0.25 3.213 .005
Non-agentive 0.45 0.19
HI Agentive 0.72 0.23 2.236 .038
Non-agentive 0.62 0.16
H Agentive 0.79 0.23 2.382 .028
Non-agentive 0.71 0.21
As can be seen from the above tables, the results obtained from every group demonstrated the same tendency. Compared with the non-agentive types, the agentive types were given higher ratings in the AJ Task and received more correct answers in the BF Task. Moreover, the differences between the two types in every group were statistically significant.
In the AJ Task, the contrasts between the agentive and non-agentive types were as follows: 2.75 vs. 2.54 in the L group, 2.62 vs. 2.28 in the LI group, 2.97 vs. 2.54 in the HI group, and 2.98 vs. 2.68 in the H group. In comparison with the non-agentive
types, the agentive types were considered very acceptable by more subjects and unacceptable by fewer subjects in every group. (Please refer to Table (vi) in Appendix E for the distribution of ratings.) In line with the results of the AJ Task, the agentive types were also performed significantly better than the non-agentive types in the BF Task. The contrasts between the agentive types and the non-agentive types were as follows: 0.66 vs. 0.36 in the L group, 0.58 vs. 0.45 in the LI group, 0.72 vs. 0.62 in the HI group, and 0.79 vs. 0.71 in the H group.
L2 proficiency was found to be an influential factor in the subjects’ performances on the two types. With respect to the agentive types, one-way ANOVA showed significant differences among groups in the AJ Task (F＝ 4.103, p＝.009) as well as in the BF Task (F＝2.858, p＝.042). The post hoc analysis by Tukey HSD showed that the LI group performed significantly worse than the H and HI groups in the AJ Task and that LI group performed significantly worse than the H group in the BF Task.
As for the non-agentive types, one-way ANOVA revealed no significant differences among groups in the AJ Task (F＝2.604, p＝.058) but in the BF Task (F＝12.154, p
＝.000). However, Tukey HSD indicated that the LI group performed significantly worse than the H group on the AJ Task and that the L and LI groups performed significantly worse than the H and HI groups in the BF Task.
In sum, the results of our two tasks showed that the agentive types were easier for the subjects than the non-agentives types. This tendency could be seen from the overall results as well as the results within each proficiency group. Besides, the subjects’ performances on the two types were influenced by their L2 proficiency.
4.3.2 General Discussion
As noted by Givón (1984), there is a thematic hierarchy for the subject position:
agent>recipient/benefactive>theme/patient>location>instrument. Grimshaw (1990)
also proposed a thematic hierarchy: agent>experience>goal/source/location>theme.
Although the orderings are not exactly the same, the two proposals agree that agent should be ranked in the highest position. In the present study, significant differences were found between progressives with an agentive subject and those with a non-agentive subject. The overall performances of our subjects followed the prediction of the hierarchy. Our subjects gained higher score for questions that contained an agentive subject. Moreover, the agency effects were observed in all the four groups. The findings indicated that the agentive types were easier to acquire and less challenging for the learners.
The agency effect was also found in Sato (2003) and Hsin and Lin (2006). In the two studies, the subjects were tested with verbs which required subjects of different thematic role: agent, experiencer, and theme. The L1 Japanese learners in Sato’s study were more accurate in answering English questions whose subject was an agent. The L1 Chinese learners in Hsin and Lin’s study also performed best on questions which contained an agentive subject, showing that verbs with an agent subject were more readily acquired.
In fact, it was not surprising that agency effects were obvious in the L2 acquisition of English progressives when we considered the nature of the progressive aspect. Comrie (1976:49-50) has proposed that the use of the progressive aspect in English typically restricts to continuous and dynamic situations which require continuous input of energy from the subject. Smith (1991:222) has also noted that progressives “are typically durative and have the connotations of dynamisms and volition.” Take our test sentences for example, the act of building a sandcastle in (13) cannot sustain without a constant input of energy from the boy, who is voluntary in doing this action. In the same vein, the event of knitting a sweater in (14) cannot continue without Jane’s effort.
(13) The boy is building a sandcastle. (BFQ20)
(14) Jane is knitting a sweater for her mother’s birthday. (AJQ18)
According to Biber et al. (1999: 473), one distinguishing feature that is associated with the verbs that commonly occurred with the progressive marker is that these verbs “typically take a human subject as agent, actively controlling the action expressed by the verb.” On the contrary, verbs that rarely take the progressive marker either have no human subject at all or “take a human subject as experiencer, undergoing but not controlling the action expressed by the verb.” In the BF Task, the L2 learners needed to describe the current condition of the subject in each picture.
Because the present moment was emphasized, all of the expected answers were in present progressive. However, our subjects tended to use simple present instead of present progressive when the subject was not a human or when the subject was a human experiencer, as shown in the following examples, taken from our subjects’
responses in the BF Task.
(15)*The light on the fire engine flashes. (HIS9) (16)*The boy sneezes. (LIS12)
The light in (15) is not human so it cannot control the actions. The boy in (16) is a human subject but the action of sneezing is beyond his control. Therefore, our subjects preferred simple present to present progressives under these circumstances despite the fact that they were asked to describe the current condition of the subject.
Because progressives commonly take an agentive subject, it is likely that the L2 learners are exposed to more input of this type. As a result, they acquired the agentive types earlier than the non-agentive types.