This study examined the L2 acquisition of the three telicity-related constructions.
On the free translation task, however, untargeted patterns were noted. An analysis of the unexpected production data might shed some light on the issue. Hence, this section is devoted to the analysis of the untargeted patterns elicited.
Table 4-6 below presents other patterns of resultative constructions, where the number stands for the frequency of each untargeted pattern elicited.
Table 4-6. Frequency of other patterns of resultatives elicited
Type Example H M L
Omission
(17) He kicked the door ___. (H4) 4 7 6
Substitution (53)
That boy cleaned up that desk. (H5) The boy made the table clean. (H37) That table was cleaned by that boy. (L24)
17 14 22
Paraphrase (46)
He kicked the door and it opened. (H12)
He opened the door by kicking it. (H24) 13 22 11
The worker threw that door green. (L17)
She cut the bread pieces. (L1) 14 46 45 No/incomplete elicitation (38) 6 14 18
Total 71 131 134
There were five major untargeted patterns elicited from the production data of resultatives. The most prevailing pattern was grammatical errors. Subsequent to this error type were substitution and paraphrase, followed by no/incomplete elicitation and omission.
Grammatical errors fell into four subtypes, including incorrect word choice, S-V disagreement, misuse of present tense, and wrong word order. Incorrect word choice accounted for the largest proportion of grammatical errors. This error type mainly bore on the misuse of verbs (e.g. The worker painted that door green. Æ The worker
threw that door green.) and the wrong form of resultative complements (e.g. She cut
the bread into pieces. Æ She cut the bread pieces. / She cut the bread for a slice.). The
findings verify that the variability of resultative complements, AP, NP, and PP, gives rise to the great complexity of resultative constructions in L2 acquisition, as proposed in the prior section. S-V disagreement was also a common grammatical error (e.g. The
worker paint the door green.). Since S-V agreement is not overtly marked in
Mandarin Chinese, the results entail the trace of negative L1 transfer.
As regards the misuse of tense, since completion of a telic event is expressed in simple past tense in English, the expected tense of the three telicity-related constructions is simple past. However, it was found that some subjects translated resultative sentences in simple present tense (e.g. She cuts the bread into pieces.), showing that telicity was a latent property of resultative constructions, of which our subjects were unaware.
Also, some of our subjects made wrong word order errors (e.g. He kicked the
door open. Æ He kicked open the door.). As the word order is characteristic of
Chinese resultative constructions, these results argued for the effect of negative L1 transfer on the subjects’ L2 acquisition. It was observed that the most occurrences of wrong word order were identified from the production data of the H group, who was
expected to make fewer grammatical errors. This might be ascribed to the fact that more resultative sentences were elicited from the H group and that all the three groups had not yet achieved the level of full attainment in their interlanguage grammar. On the whole, the results on grammatical errors showed that the frequency of grammatical errors descended with the increase in the subjects’ proficiency levels; the H group outperformed the other proficient groups in most cases.
In addition to grammatical errors, it was found that the subjects employed the avoidance strategy of substitution of other constructions in replace of resultatives, which were novel to them (cf. Butterworth and Hatch 1978). Three types of constructions were used to substitute for resultatives. The first type was verb-particle constructions (e.g. That boy wiped that desk clean. Æ That boy cleaned up that
desk.). This result lends support to our regression analysis that the successful
acquisition of verb-particle constructions was well predicted by that of resultative constructions. The second type was causative constructions (e.g. The boy made the
table clean.). This finding indirectly echoes Snyder and Stromswold’s (1997)
conclusion that causative constructions form a cluster with other telicity-related constructions. The third type was passive constructions (e.g. That table was cleaned
by that boy.). This may be attributed to the high degree of affectedness denoted by
resultative constructions, in which the postverbal argument NP undergoes a high degree of affectedness. Similarly, passive constructions are characterized by the disposal effect, and the affected argument NP is the focus of passive constructions.
This may account for the substitution of passive constructions for resultatives.
Like substitution of other constructions, the subjects were found to draw on the strategy of paraphrase for resultatives as well (e.g. He kicked the door open. Æ He
kicked the door and it opened. / He opened the door by kicking it.). Paraphrase can be
also regarded as an avoidance strategy due to their unfamiliarity with resultative
constructions.
Occasionally, the subjects omitted resultative complements (e.g. He kicked the
door open. Æ He kicked the door ___.), which were the parts they were not familiar
with. To the extreme of omission, it was found that the subjects chose to produce nothing or incomplete sentences for resultatives. These two strategies also confirm the high degree of difficulty of resultative constructions.
In the case of verb-particle constructions, elicited untargeted patterns and their frequency are tabulated in Table 4-7.
Table 4-7. Frequency of other patterns of verb-particles elicited
Type Example H M L
Omission
(27) The student took ___ the book. (H12) 6 10 11 Paraphrase
(32)
The woman found her key on the ground
and then caught it. (M26) 13 6 13
That woman found her key on the ground
and pick it up. (L9) 12 23 24 Word order
(7)
The lady found her key on the floor and
picked up it. (M20) 0 3 4
As shown in Table 4-7, four main patterns were elicited from the verb-particle translation data. Like resultatives, grammatical errors accounted for the largest proportion of the unexpected production data, followed by paraphrase, omission, and no/incomplete elicitation.
With respect to grammatical errors, S-V disagreement was the most common type (e.g. That woman found her key on the ground and pick it up.). Next was the incorrect word choice. One error type involved the inappropriate selection of verbs (e.g. Mr. Wang bought a coat and put it on. Æ Mr. Wang bought a coat and wore it
on.). This kind of error entails that the subjects could not distinguish the nuances of
meaning between the verb wear and the verb-particle put on. This error type might result from negative L1 interference, for the verb-particle put on corresponds to the Chinese verb-particle compound chuan-shang, literally meaning ‘to wear on’. As Chinese verb particles can co-occur with some stative verbs, the subjects, as a result of negative L1 transfer, might have no inkling of the incompatibility of stative verbs and telic particles in English. The other type was wrong collocations of verbs and particles. A typical example was the wrong use of the particle up in sentences like Mr.
Wang bought a coat and put it up. This error type supports our analysis that the
variability of particles augments the complexity of verb-particle constructions.
Besides, misuse of present tense was observed in the verb-particle data as well (e.g. That student takes the book away.), implying that our subjects did not recognize the telic property of the constructions. The last grammatical error type was misordering (e.g. The lady found her key on the floor and picked up it.). English particles obligatorily follow a pronoun, whereas Chinese verb-particles exhibit the fixed word order V-Particle-NP. The finding points to the effect of negative L1 transfer. Generally speaking, the H group made fewer grammatical errors than the M and L groups on verb-particle constructions.
Like resultatives, it was found that the subjects drew on the strategy of paraphrase for verb-particle constructions (The woman found her key on the ground
and then caught it.). They were also observed to omit some particles, which varied
according to different verbs (e.g. The student took away the book. Æ The student took
___ the book.). Finally, 13 occurrences of no/incomplete elicitation were identified in
the verb-particle production data.
Finally, unexpected patterns of dative/double object constructions elicited are shown in Table 4-8.
Table 4-8. Frequency of other patterns of datives/double objects elicited
Type Example H M L
That teacher tell her student one thing.
(L2) 6 13 27
Word order (2)
The teacher told to her student one thing.
(M5) 0 2 0
Word choice (25)
The teacher talked one thing to her student. (M7)
John threw a pen at her. (H10)
5 7 13
No/incomplete elicitation (3) 1 2 0
Total 19 34 53
Only grammatical errors were elicited in the constructions. S-V disagreement constituted the largest amount of the errors (e.g. That teacher tell her student one
thing.), indicating that L2 common errors interfered in the L2 acquisition of the
constructions. The misuse of present tense (e.g. John throws a pen to her.) showed that our subjects were still unaware of the telic property of dative/double object constructions, as they were not on resultatives and verb-particles. In the case of incorrect word choice, one error type was related to the verb choice (e.g. The teacher
told one thing to her student. Æ The teacher talked one thing to her student.). It is
indicated that the subjects could not distinguish (di-)transitivity of the two semantically similar verbs talk and tell. The other error type bore on the form of the case marker (e.g. John threw a pen to her. Æ John threw a pen at her.). The finding
showed that a few subjects did not differentiate prepositions from the case marker to, both of which were similar in form. Finally, two occurrences of wrong word order were elicited (e.g. The teacher told to her student one thing.). The word order [V-to-I.O. D.O.] is typical of one variant of Chinese dative constructions—another piece of evidence of negative L1 transfer. By and large, the higher L2 proficient learners performed better on dative/double object constructions than the lower counterparts.
Table 4-9 shows the overall distribution and frequency of elicited untargeted patterns of the three telicity-related constructions.
Table 4-9. Other patterns of the telicity-related constructions elicited
Type Resultative Verb-particle Dative/
double object
Omission (40) 17 23 0
Substitution (53) 53 0 0
Paraphrase (78) 46 32 0
Grammatical error (431) 182 167 107
No/incomplete elicitation (54) 38 13 3
Total 336 232 110
As shown in Table 4-9, five main untargeted patterns were noted in this study. They coincide with the error taxonomy by Dulay, Burt, and Krashen (1982) and Butterworth and Hatch (1978), viz. omission, substitution, misordering, and misinformation.
Among the five types, grammatical errors accounted for the largest proportion of unexpected production data. They involved incorrect word choice (essentially verbs, resultative complements, and particles, etc.), misuse of present tense, S-V disagreement, and wrong word order. Intriguingly, the four subtypes of grammatical errors were found across the three constructions. On the one hand, the
construction-universal misuse of present tense shows the covert property of telicity of the three constructions. On the other hand, the wrong word choice, incorrect word order, and S-V disagreement throughout the constructions entail L2 complexity of the telicity-related constructions and negative L1 interference in L2 acquisition, as suggested above.
Besides, paraphrase and omission were found in resultative and verb-particle constructions but not in dative/double object constructions. In accord with the above-observed trend, these findings suggest that datives/double objects were the least difficult constructions and need not be rephrased or avoided by means of other patterns or circumlocution. As for substitution of constructions, the avoidance strategy was specific to the resultative constructions; verb-particle constructions are often used to replace resultatives. This corroborates the formerly established predictive relationship that verb-particle constructions are a precursor of resultatives. Moreover, the frequency of no/incomplete elicitation of the three constructions is consonant with the overall acquisition sequence of the three telicity-related constructions.
As the production analysis suggests, on the whole, resultative constructions exhibit the highest incidence of unexpected patterns, and dative/double object constructions the lowest frequency. The analysis of the untargeted production patterns accords with the relationship of the three telicity-related constructions observed in the prior sections. In the formerly established difficulty hierarchy, resultatives are the hierarchically highest constructions, followed by verb-particles and by datives/double objects. The hierarchy accounts for the ascending acquisition sequence of the three telicity-related constructions: resultatives < verb-particles < datives/double objects.