Explicit and implicit knowledge of infinitival and gerundival verb complements in L2 speech

From LearnLab
Revision as of 15:42, 31 August 2011 by Mbett (talk | contribs) (Reverted edits by Pierrehernandez (Talk); changed back to last version by Ndjong)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

[This page is under construction]

Abstract

The present study set out to investigate whether students were able to use their explicit knowledge of verb complements during spontaneous speech. Spontaneous speech data was elicited from 32 high-intermediate English as a Second Language (ESL) students. Results show that students made errors that were not typical of explicit knowledge, in that they did not merely switch verb complement types. Instead, they used many unmarked complements or doubly-marked complements. This suggests they relied on implicit knowledge, possible acquired from exposure to the target language. Variable input due to matrix verbs that can take either type of complement may have affected the acquisition of the correct complement forms.

Teachers cannot assume that the variable matrix verbs are the easiest ones because they allow both forms. Instead, these verbs may need additional instruction. In addition, instruction needs to include sufficient practice in order to stimulate the acquisition of implicit knowledge.

Background and significance

It is a common phenomenon that second language (L2) learners know many grammatical rules but do not apply them correctly and consistently when speaking. In class, students often acquire explicit knowledge of grammatical structures, but this is slow to use and requires attentional resources (De Jong, 2005; DeKeyser, in press; Ellis, 2005, 2006; Hulstijn, 2002). Therefore, to be able to speak with high fluency and grammatical accuracy, it is necessary for students to acquire implicit knowledge, because it is faster to use and does not require as much attentional resources.

The present study set out to investigate whether students were able to use their explicit knowledge of verb complements during spontaneous speech. If explicit knowledge is used, L2 learners may choose the two verb complement forms equally; if implicit knowledge is used, L2 learners might show preference for one of the two forms. If explicit knowledge is being used, it can be expected that errors would involve switching the two types of complements. However, if other errors would occur, implicit knowledge is likely being used and possibly L1 transfer may be involved.

Research question

  • Do high intermediate ESL students rely on their explicit knowledge of a variable grammatical structure (verb complements) during spontaneous speech, or do they rely more on implicit knowledge instead?

Method

Participants were 32 students enrolled in high intermediate Speaking classes. They had been taught the grammatical structure of English verb complements, in that some verbs take a to-infinitive, some take a gerund, and others take either a to-infinitive or gerund as a complement. The students took part in three sessions of the 4/3/2 task, in each of which they spoke three times, for four, three, and two minutes, respectively. Fifteen students repeated the same topic during a session, while 8 other students spoke about three different topics. In total, 27 minutes of speech were elicited from each student.

The speeches were transcribed in PRAAT, and annotated in CHAT/CLAN (the software of the Childes project). Codes were added for parts of speech, errors, and retracings (repetitions, corrections, and reformulations). All verb complements (correct and errors) were retrieved to be further analyzed.

Independent variables

  • Type of matrix verb requirement
    • to-infinitive
    • gerund
    • either

Dependent variables

  • Type of complement produced
    • to-infinitive
    • gerund
    • ambiguous
  • error
    • appropriate
    • not appropriate

Hypotheses

  • If students rely more on explicit knowledge of the target structure, errors will involve mostly switching of the verb complement types.
  • If students rely more on implicit knowledge of the target structure, other types of errors will be made, e.g., use of the default form.

Findings

Verb complement use and accuracy All participants attempted the target structure; the range was 7-44 verb complement attempts per student. The mean accuracy per student was 82.2% (12%) with a range of 60 – 100%.

Matrix verbs that require a gerund were produced least often (16) but most accurately. More instances were produced of matrix verbs that require an infinitive (152) or allow either a gerund or a to-infinitive (352). The error rate for both groups of these matrix verbs was around 15%. The error pattern for these types of matrix verbs was very similar.

Errors were rarely a result of a mismatch between matrix verb and verb complement, as when a gerund would be used instead of a to-infinitive or vice versa (e.g., I enjoy watch TV, I want studying). Instead, the most common error was producing only the root verb with neither the infinitival to marker nor the gerundival -ing, while the next common error was using both markers.

L1 transfer To explore the origins of the preference for using gerunds and to-infinitive markers, we investigate L1 influence. The Korean L1 and Chinese L1 speakers had much lower means of verb complement structures per student (about 16) than the other L1 groups (over 30). The Korean L1 students and the single Russian speaker had slightly lower accuracy (about 73%) than the other L1 groups.

It might be predicted that students from L1s with less morphology produced the ambiguous verb complements with neither morpheme. This expectation was not supported because L1 did not seem to impact the type of errors.

Explanation

If explicit knowledge were used, it would be likely that students knew the correct verb complement forms, but made errors selecting the appropriate complement for a particular matrix verb. Thus, gerunds would be used instead of to-infinitives, and vice versa. In addition, there would be an even distribution of the two options. Instead, the most common error was a lack of either morphological marking, only producing the root verb for the verb complement.

A possible explanation is that the online demands of the production task and trying to apply explicit knowledge about the matrix verb requirements were too great, resulting in no morphological marking. Therefore, the students may have relied on their implicit knowledge during this production task.

One type of implicit knowledge may have stemmed from experience with the target language. Of the matrix verbs that can take either verb complement form, 13% of the verb complements were ambiguously produced. This finding might reflect the difficulty of learning a structure that has varied input; there is no clear collocation with the matrix verbs, especially in the variable category (like, prefer, love) despite the high frequency of these verbs.

Teachers cannot assume that the variable matrix verbs are the easiest ones because they allow both forms. Instead, these verbs may need additional instruction. In addition, instruction needs to include sufficient practice in order to stimulate the acquisition of implicit knowledge.

Further information

The results of this study were presented at the GURT conference in March 2009, and are under review for the proceedings. The study was performed by Mary Lou Vercellotti from the University of Pittsburgh and Dr. Nel de Jong from Queens College of CUNY.

This study was part of the project Fostering fluency in second language learning by Nel de Jong, Laura Halderman, and Charles Perfetti.