The foremost research question the present study addresses is the L2 acquisition of the three telicity-related constructions, resultatives, verb-particles, and datives/double objects. Specifically, the issue concerns whether the three constructions are acquired approximately concurrently, sequentially, or in a piecemeal fashion—the relationships of the constructions.
In the literature, Snyder (1995) notes semantic similarities of the three constructions in English such that they all denote telic interpretations. On the one hand, Snyder puts forth a null telic morpheme to group the three constructions together, providing a compositional analysis of the telic semantics of the constructions. On the other hand, he resorts to the complex verb approach to accounting for the syntactic structures of the null telic morpheme and the three related constructions. In light of Snyder’s telic proposal, the present study links the three constructions jointly under investigation.
In Chapter Two, the three telicity-related constructions in English and Chinese have been compared and contrasted. As summarized in Table 2-1, English and Chinese dative/double object constructions display a similar constraint on the verb types; the constructions are incompatible with any stative verbs in the two languages.
Structurally, English dative/double object constructions [V D.O. to I.O.] / [V I.O.
D.O.] are available in Chinese and are parallel to Chinese counterparts [V D.O. gei I.O.] / [V I.O. D.O.]. The difference observed between the two languages lies in the fact that Chinese dative constructions have another variant [V gei I.O. D.O.], which exhibits reverse word order to that of the English counterpart.1
On the contrary, verb-particle and resultative constructions demonstrate relatively a higher degree of differences between English and Chinese than datives/double objects. The verb-particle constructions in these two languages differ in their structures and verb types. English particles follow the main verb and they can appear on either side of the direct object, while Chinese particles form a compound with the action verb and they are not subject to particle movement. Besides, English verb-particles are incongruent with any stative verbs, whereas some Chinese counterparts can apply to this verb type. As regards resultative constructions, English and Chinese resultatives also display different morphosyntactic structures; the action verb and the resultative expression come apart in the former but are compounded together in the latter. Also, English resultative constructions are incompatible with any stative verbs and are predicated only of the direct object. Chinese, in contrast, allows some stative verbs to appear in resultative constructions, which are more flexible in predication of the direct object or the subject than the English counterparts.
1 Another difference between English and Chinese dative/double object constructions is their constraints on dative alternation. However, dative shift is not the research focus of the present study. To factor out the effect of different constraints on dative alternation, all the ditransitive verbs under investigation were dativizable verbs, i.e. ditransitive verbs that can apply to both types of constructions.
On the whole, dative/double object constructions display a relatively lower degree of L1-L2 differences than verb-particles and resultatives, and the latter two constructions are quite alike with respect to the extent of variations between the languages under consideration. Although the three English constructions purportedly form a big family of telicity-related constructions, the different degrees of L1-L2 similarities and differences may lead to varying positive/negative L1 transfer and contribute to the sequential acquisition of the three constructions, as predicted in (1).
(1) Prediction of the acquisition sequence of the constructions based on L1 transfer DATIVES/DOUBLE OBJECTS > VERB-PARTICLES = RESULTATIVES
It is predicted that dative/double object constructions will be acquired earlier by our Chinese learners of L2 English than verb-particle and resultative constructions and that the latter two constructions, due to their comparable differences between L1 and L2, will be acquired around the same time.
The present study drew on two experimental tasks to collect pertinent data, and the subjects of this study were regrouped into three different levels of English proficiency. To paint a holistic picture of the acquisition of all the constructions, English proficiency and task effects were not considered for the time being. Figure 4-1 below shows the subjects’ overall performance on the three telicity-related constructions. The subjects as a whole performed the best on dative/double object constructions (Mean=.85, SD=.14), while they did the worst on resultative constructions (Mean=0.49, SD=.14). Their mean scores on verb-particle constructions reached 0.69 point (SD=.15), lying in between those of the former two constructions.
0.49
0.69
0.85
0 0.2 0.4 0.6 0.8 1
Resultative Verb-particle Dative/double object Construction
Mean score
Figure 4-1. The overall results on the telicity-related constructions
The statistical significance of the differences among the three constructions is presented in Table 4-1.
Table 4-1. Mean differences between the overall results on the constructions
Construction p
Resultative Verb-particle .000**
Dative/double object .000**
Verb-particle Dative/double object .000**
Note: The mean difference is significant at the .05 level (*p<.05, **p<.01).
The mean differences between any two of the three constructions arrived at the statistically significant level (p=.000** throughout). As the results suggest, the three telicity-related constructions display divergent degrees of complexity and difficulty;
thus, construction effects on the second language acquisition were borne out. The L2 acquisition of the three constructions exhibit the following sequence: dative/double object constructions precede verb-particle constructions, which in turn precede resultative constructions. The findings corresponded to our prediction of the acquisition of dative/double object constructions, but do not accord with that of verb-particle and resultative constructions.
The present study proposes that although resultatives, verb-particles, and datives/double objects belong to the family of telicity-associated constructions, the acquisition of the constructions is underlain by the acquisition of the subcategorization frame of each family member. For dative/double object constructions, the subcategorization frame is [V, ___ NP NP], in which the two postverbal NPs are a constant (e.g. Haegeman 1994, Van Valin 2001). In contrast, the subcategorization frame of resultative constructions involves a variable X, as in [V, ___ NP X], where the variable stands for different forms of resultative complements, such as AP, PP, and NP (see Carrier and Randall 1992).2 As for verb-particle constructions, particles vary according to verbs. Verbs in the constructions subcategorize for an NP and a variable X, where X represents different particles, such as on in put…on and away in take…away. Among these three constructions, both resultative and verb-particle constructions contain a variable X, whereas dative/double object constructions are typified by a constant in a salient frame. The marked variable may augment structural complexity of verb-particle and resultative constructions, which leads to the later acquisition of the two constructions. On the other hand, the fixed frame of constant NP-NP constructions may account for the earlier acquisition of dative/double object constructions.
As the subcategorization frames entail, verb-particle and resultative constructions take a transitive verb. To acquire the two constructions, the subjects had to differentiate phrasal verbs from non-phrasal verbs, in addition to the subcategorization frames. Verb-particle constructions take a phrasal verb (see Quirk et al. 1985, Biber et al. 1999), while resultative constructions take a non-phrasal verb.
2 According to Carrier and Randall (1992), although resultative complements come in different forms, AP, PP, and NP, they are not necessarily substitutable in all the resultative sentences. Since the c-selection cannot satisfactorily account for the restriction on resultative XPs, Carrier and Randall propose that the verb itself s-selects the resultative phrase, which must therefore be an argument of the verb. Based on Chomsky’s (1986) assumption that argumenthood requires sisterhood, they put forth a ternary analysis that the verb, the postverbal NP, and the resultative XP must all be sisters.
The marked nature of phrasal verbs and the addition of resultative complements to ordinary non-phrasal verbs may also give rise to the greater complexity of verb-particle and resultative constructions than that of datives/double object constructions.
So far, the much higher scores on dative/double object constructions than the other two constructions have been discussed in terms of L1 transfer and L2 structural complexity. However, these variables could not satisfactorily explain our subjects’
contrasting performance on verb-particle and resultative constructions. In the literature, some studies have suggested the role of linguistic inputs in language acquisition (e.g. Ioup 1984, White et al. 1991, Trahey and White 1993, Elbers 2000, Hara 2007). In Taiwan’s secondary school English education, verb-particle constructions are one of the constructions at which English curricula are targeted.
English textbooks provide adequate inputs for the acquisition of verb-particle constructions. However, no trace of these inputs bearing on resultatives could be identified in the high school English formal teaching material. Perhaps with the trigger of verb-particle inputs, the subjects made relatively faster progress toward the acquisition of verb-particle constructions than that of resultative constructions in their interlanguage system. This holds true of the earliest acquisition of the dative/double object constructions, which are introduced in the high school English textbooks.
Adequate L2 inputs may be a trigger for the acquisition of the constructions as well.
In a nutshell, the earlier acquisition of English dative/double object constructions than that of the other two constructions could be ascribed to the lower L2 complexity, the higher saliency of the subcategorization frame, and positive L1 transfer. On the other hand, the absence of L2 inputs might account for the later acquisition of resulative constructions than verb-particles.
When the variable of L2 proficiency was considered, the results of the
acquisition of the three constructions consistently approximated to the foregoing overall findings, as shown in Figure 4-2.
0.83
Figure 4-2. Each group’s scores on the telicity-related constructions
The higher proficient learners of English received the highest mean score on dative/double object constructions (Mean=.92, SD=.05) and the lowest score on resultative constructions (Mean=.58, SD=.14), with the mean score of the verb-particle constructions being midway between them (Mean=.77, SD=.13). The middle and lower groups demonstrated a similar acquisition order of the three constructions: dative/double object constructions (Mid: Mean=.83, SD=.13; Low:
Mean=.80, SD=.10) were followed by verb-particle constructions (Mid: Mean=.67, SD=.13; Low: Mean=.62, SD=.15), which in turn were followed by resultative constructions (Mid: Mean=.45, SD=.12; Low: Mean=.42, SD=.12). The variances between the subjects’ performance on the three constructions are tabulated in Table 4-2. Table 4-2 indicates that the H group performed significantly differently than the M and L groups on the three constructions (p<.05* throughout). On the contrary, the M and L groups showed no significant differences on the three telicity-related
Table 4-2. Mean differences between the experimental groups’ scores on the constructions
Group Resultative Verb-particle Dative/
double object
H M .000** .02* .001**
L .000** .000** .000**
M L .59 .25 .60
Note: The mean difference is significant at the .05 level (*p<.05, **p<.01).
constructions (p>.05 throughout). The results showed that the higher English proficient group outperformed the middle and lower proficient groups at the statistically significant level and that the M and L groups were at the similar developmental stage. Thus, the variable of L2 proficiency in acquisition is substantiated.
Basically, the results are in tune with Slabakova’s (2001) findings with respect to the acquisition of the three constructions. In the present study, all the three groups demonstrated a parallel tendency on the acquisition order of the three constructions (datives/double objects > verb-particles > resultatives). Slabakova reported that the advanced and high-intermediate Bulgarian groups performed better on double object constructions than verb-particles and resultatives and similarly on the latter two constructions (double objects > verb-particles = resultatives), while the low-intermediate learners performed mostly at chance, with the exception of higher accuracy on double objects. According to Slabakova, there are no corresponding resultative, verb-particle, and dative/ double object constructions in Bulgarian. In light of the identical constructions and different L1 subjects under investigation, it is suggested that the L2 complexity of the three constructions is a universal tendency.
In light of the consistent acquisition sequence of the three constructions shown by the holistic results and the between-group analysis, it is worthwhile to probe into
the relationships of these constructions and embark on a regression analysis on the pertinent individual subjects’ data. Table 4-3 shows the correlation coefficient of the three constructions and the statistical significance.
Table 4-3. Correlations of the telicity-related constructions
Construction r p
Resultative Verb-particle .32** .001**
Dative/double object .42** .000**
Verb-particle Dative/double object .40** .000**
Note: The correlation is significant at the .05 level (*p<.05, **p<.01).
It was indicated that the three constructions were significantly correlated (p<.01**
throughout). To cope with the relationships of the telicity-related constructions, individual subjects’ data were examined further. Those who successfully acquired each of the three constructions were singled out for the in-depth analysis. The criterion for successful acquisition of one construction was 75% accuracy both on the grammaticality judgment task and on the free translation task (Mean=.75).3 Figure 4-3 below shows the number of the subjects who demonstrated successful acquisition of each construction. As Figure 4-3 exhibits, the number of the successful acquirers of the constructions accorded with the observed acquisition sequence: resultatives (7), verb-particles (34), and datives/double objects (85).
Based on these subjects’ data, this study looked into their acquisition of one telicity-related construction in relation to the other two constructions. For the seven successful learners of resultatives, five of them also demonstrated successful acquisition of verb-particle constructions (71.4%), and all the seven subjects showed a perfect command of dative/double object constructions as well (100%). It was
3 According to Slabakova (2001:167-168), the criterion of 75% accuracy as the demonstration of certain knowledge at issue achieved the statistically significant level, where the subjects’ performance
‘would be different from chance at p=.02069’.
Figure 4-3. The acquisition of one construction vis-à-vis the other two constructions
suggested that the success in the acquisition of resultative constructions would well predict successful acquisition of verb-particle constructions and that of dative/double object constructions.
Of the 34 successful learners of verb-particles, 32 subjects displayed successful acquisition of dative/double object constructions (94.1%), and only five learners succeeded in acquiring resultative constructions (14.7%). The results indicated that verb-particle constructions were a significant predictor of dative/double objects but not of resultatives.
With respect to the successful acquires of datives/double objects, 32 learners demonstrated their success in the acquisition of verb-particles (37.7%), and far fewer subjects met the acquisition criterion of resultatives (8.2%). The findings showed that although dative/double object constructions were well predicted by verb-particles and by resultatives, they were not a significant predictor of the latter two constructions.
As the consistent results showed, the three telicity-related constructions were found significantly correlated and exhibited a difficulty hierarchy, with resultatives as the hierarchically highest construction and datives/double objects as the lowest, as formulated in (2).
51 27
5
2 2
resultative verb-particle
dative/double object
(2) The difficulty hierarchy of the telicity-related constructions
RESULTATIVES > VERB-PARTICLES > DATIVES/DOUBLE OBJECTS
On the one hand, the difficulty hierarchy was negatively correlated with the acquisition order. The higher the constructions were in the hierarchy, the greater degree of difficulty they would exhibit and the later they would be acquired. On the other hand, the hierarchically higher constructions were found to be a significant predictor of the lower constructions, but not vice versa. The success in the acquisition of resultatives could predict the successful acquisition of verb-particles and of datives/double objects. Verb-particles were also found to be a significant predictor of datives/double objects but not of resultatives. Datives/double objects were well-predicted by verb-particles and by resultatives, but they could not predict the latter two constructions.
The results of individual scores were to some extent different from the analysis of individual scores by Slabakova (2001). Slabakova reported that each one of the three constructions was significantly well predicted by the other two constructions.
Despite the differences, the present study chimes in with Slabakova’s in the sense that resultatives, verb-particles, and datives/double objects were significantly correlated and exhibited a predictive relationship. Also, this study indirectly lends empirical support to the theoretical proposal that the three constructions bear a close relationship and should be treated as the same family members (cf. Hoekstra 1988, Snyder 1995).