Chinese Speakers' Acquisition Of The English Resultative Construction:the Role Of Input Type Frequency And Input Distribution

There is broad consensus in SLA research that input is the key to L2 learning,and that learners generally require vast amounts of rich and meaningful input for successful acquisition.However there remain some disputes as to the exact kind of input structure that is most facilitative for L2 learners to acquire specific constructions.And a number of studies have found that skewed input is more beneficial in construction learning for it features the frequency distribution in natural language.There have also been some proposals which state that type frequency exerts influences on construction learning combined with other variables.However,there is far less consensus as to the role of input type frequency and input distribution despite the studies conducted over the issue.The present study investigated the effects of Input Type Frequency(high/middle/low)and Input Distribution(balanced/skewed)as well as their interactive effects on the learning of the English resultative construction.The participants were 108 Chinese middle school students without any prior knowledge of the target construction.They were randomly assigned to seven groups according to the conditions generated by the independent variables.There were altogether 150 instances in the input.The participants received different types of input(HIGH-BALANCED,30 verb types with 5 tokens for each;HIGH-SKEWED,30 verb types,with 15 tokens for 4 verb types and 3-4 tokens for 26 verb types;MIDDLE-BALANCED,18 verb variations,with 8-9 tokens for each;MIDDLE-SKEWED,18 verb variations,with 20 tokens for 3 verb types and 6 tokens for15 verb types;LOW-BALANCED,10 verb types,with 14-16 tokens for each;LOW-SKEWED,10 verb variations,with 35 tokens for 2 and 10 tokens for 8 verb types).There was also a control group designed to testify and eliminate the testing effects.The participants in the control group received linking verb plus adjective structures(30 verb types with 5 tokens for each)rather than the target construction and their input was also balanced.A pretest was implemented to indiscriminate the discrepancy which may exist prior to the manipulation.After the training,the participants took the immediate posttest and two weeks later,they finished the delayed posttest.For the influence of Input Distribution,this experiment found little evidence for the facilitative effects of the skewed input.Instead,when the learners received high type frequency input,construction learning in the balanced frequency condition is significantly superior than the skewed input condition.Besides,when the learners were exposed to middle or low type frequency input,there was no significant difference between the balanced and skewed groups.To be more specific,when exposed to middle type frequency input,learners in the balanced group showed insignificantly better learning outcome than those in the skewed group.While for the low type frequency condition,the results were the opposite,but the discrepancy between the balanced condition and the skewed condition was much smaller than that in the middle input type frequency condition.The writer supposed that when exposed to high type frequency input,despite the advantages in the amount of the verbs,there were only three or four verbs which acquire the adequate number of examples for the generalization of the construction while others not.On the contrary,the participants in the balanced group took the advantage of the amount of the verbs with their adequate number of examples and they obtained significantly better results.For those participants in the middle and low input type frequency conditions,in spite of the existing differences among the conditions as to the amount of the verbs and the example sentences for each verb,there were enough exemplars to demonstrate the information of the construction.Thus,the learning results of these groups exhibited no significant difference.This finding challenges the proposal that skewed input facilitates construction learning more than balanced input.When it comes to the role of Input Type Frequency,under the same input distribution,the high type frequency input was significantly facilitative than the other three conditions and the middle and low type frequency input also substantially outperformed the zero type frequency condition.This finding confirms the proposal that high input type frequency facilitates generalization,abstraction of the construction.The best learning results took place with a combination of the high type frequency and the balanced distribution.Therefore,construction learning is most successful if the input not only contains vast amounts of the target pattern,but if these input floods come with favorable frequencydistributions.Long-term effects showed that the learners earned a significantly better learning outcome from the pretest to the immediate posttest and despite the loss from the immediate posttest to the delayed posttest,the eventual state of their learning outcome was substantially better than the pretest.The control group showed no statistically better learning results across the three tests.Concerning that the testing instrument tested both the interpretation and the production of the target construction,the writer analysed the data according to the testing target of the instrument.And the results showed that as long as the learners received target construction related input,their understanding of the resultative construction outperformed that prior to the training.That is to say,the interpretation of the target construction was more influenced by the input type frequency and little evidence was found as to the role of input distribution.While for the production of the target construction,the overall tendency for the influence of the two factors was almost the same with the general results.That is,high type frequency and balanced input was more facilitative for the construction learning and there was no significant difference among the other five experimental groups while all the experimental groups outperformed the control group.Long-term learning effects also showed that all the experimental groups outperformed in their immediate posttest and the delayed posttest scores also reached a statistical state despite the loss from the immediate posttest to the delayed posttest.The control group showed no significant learning outcome across the tests.On one hand,the present study enriches the studies under the instruction of the usage-based linguistic model.To be more specific,the paper not only addresses the dispute over the role of input distribution on construction learning,but sheds some light on the interactive effects of the input distribution and input type frequency.On the other hand,the study generates some implications for construction and English patterns teaching,and it provides instructional routines for the teaching and learning of English as an L2.