COMPUTER-ASSISTED GRAMMAR PRACTICE

FOR ORAL COMMUNICATION

Stephen Bodnar, Catia Cucchiarini and Helmer Strik

Department of Linguistics, Radboud University Nijmegen, Nijmegen, The Netherlands

Keywords:

Computer-assisted language learning, Corrective feedback, Grammar instruction, Oral proﬁciency.

Abstract:

Gaining the ability to speak proﬁciently is an important goal in second language learning, and grammatical

correctness is an important dimension of oral proﬁciency. To acquire the ability to produce grammatically

correct speech in everyday conversational situations, learners must practice producing speech until they can

do so with little to no conscious effort. For maximum pedagogical effectiveness, practice exercises should

challenge learners to produce spoken output and provide corrective feedback (CF) on their productions so that

learners may notice and correct their mistakes. In this paper, we survey the ﬁeld of Intelligent Computer-

assisted Language Learning (ICALL) to examine the extent to which current offerings meet the pedagogical

requirements for training grammatical accuracy in oral communication. Our analysis shows that few grammar-

focused systems support oral practice, and that systems which do offer oral practice tend to train conversational

ﬂuency. In response to these ﬁndings, we present our position that grammar accuracy should be addressed in

ICALL systems and that, in spite of technological limitations, it is possible to deploy speech technology in

ICALL systems to support spoken interaction and allow individualized oral grammar practice and feedback.

1 INTRODUCTION

Gaining the ability to speak proﬁciently is an impor-

tant goal in second language (L2) learning, and gram-

matical correctness is an important dimension of L2

proﬁciency (Housen and Kuiken, 2009). (Brumﬁt,

1984) distinguished between accuracy-oriented ac-

tivities, those aimed at promoting the production of

grammatically correct language, and ﬂuency-oriented

activities, those aimed at stimulating spontaneous L2

production. Exposure to the target language and

usage-based learning are viewed as essential elements

in Second Language Acquisition (SLA) (Ellis and

Bogart, 2007). However, a considerable body of

SLA research has also shown that language input and

usage-based learning are not enough and that focus

on linguistic form and accuracy is often required to

help L2 learners achieve target-like levels of proﬁ-

ciency. Through instruction, output, intensive prac-

tice and corrective feedback L2 learners can acquire

the ability to produce grammatically correct speech in

everyday conversational situations. In general, learn-

ers do not have many opportunities to engage in the

kind of practice they need, namely spoken interac-

tions where feedback is provided. In language class-

rooms the emphasis is typically on meaning instead

of form, the difﬁculty level and the curriculum are not

tailored to the needs and interests of each learner, and

large class sizes limit the amount of practice available.

Recent developments in Intelligent Computer-

Assisted Language Learning (ICALL) may provide

new opportunities for offering intensive practice in

speaking the target language and personalized cor-

rective feedback. However, an analysis shows that

the majority of systems that train grammar skills

do not support spoken interaction, and that speech-

interactive systems typically do not individualize lan-

guage instruction and tend to focus on other aspects

of oral proﬁciency such as pronunciation or conversa-

tion skills.

It is our opinion a) that grammar accuracy, as

an important dimension of oral proﬁciency, should

be addressed in ICALL systems and b) that in spite

of technological limitations, it is possible to deploy

speech technology in ICALL systems to support spo-

ken interaction and allow individualized oral gram-

mar practice and feedback.

This paper is organized as follows. In section 2,

we draw on literature from Second Language Acqui-

sition (SLA) research to explore the pedagogical re-

quirements for providing sound and effective gram-

mar practice. Section 3 employs the features pre-

355

Bodnar S., Cucchiarini C. and Strik H..

COMPUTER-ASSISTED GRAMMAR PRACTICE FOR ORAL COMMUNICATION.

DOI: 10.5220/0003402503550361

In Proceedings of the 3rd International Conference on Computer Supported Education (CSEDU-2011), pages 355-361

ISBN: 978-989-8425-49-2

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

sented in the previous section as desiderata for oral

grammar practice to review recent advanced computer

applications from ICALL. In Section 4, we present

our position that pedagogically sound ICALL systems

for oral grammar practice can be developed by using

Automatic Speech Recognition (ASR).

2 REQUIREMENTS FOR ORAL

GRAMMAR PRACTICE

2.1 Literature Overview

In the ﬁeld of SLA there has been considerable de-

bate on the role of input, output, practice and cor-

rective feedback as factors in promoting the acqui-

sition of a second language. Proposers of the In-

put Hypothesis and the Natural Approach (Krashen,

1982) (Krashen and Terrell, 1983) maintain that lan-

guage acquisition is driven by exposure to compre-

hensible language input, and that “speech (and writ-

ing) production emerges as the acquisition process

progresses” (Krashen and Terrell, 1983, p. 53). In

line with this reasoning, language learners should not

be stimulated to produce language output because this

only plays a marginal role in the acquisition process.

However, immersion studies in which language

learners were exposed to huge amounts of language

input have revealed that this is not enough to achieve

target-like performance in productive skills (Swain

and Lapkin, 1982) (Ranta and Lyster, 2007). This

resulted in a renewed emphasis on the importance of

output in language learning and the possible functions

it may serve (Swain, 1985) (DeBot, 1996).

Relevant in this respect is the connection to skill

acquisition theory in cognitive psychology (Ander-

son, 1982) and in particular the distinction between

declarative and procedural knowledge. In this view

people acquire skills ﬁrst through explicit information

or rules (declarative knowledge) which are then grad-

ually transformed into internalized routines (proce-

dural knowledge) which eventually are automatized.

Output and practice play an important role in con-

verting declarative knowledge into procedural knowl-

edge thus promoting ﬂuency and automaticity (De-

Bot, 1996) (Gass, 1997). An important distinction

between declarative and procedural knowledge is that

while declarative knowledge is generalizable, pro-

cedural knowledge is skill-speciﬁc, and in spite of

claims that it is input practice that leads to acquisi-

tion (VanPatten and Cadierno, 1993), various studies

have indicated a lack of transfer from receptive to pro-

ductive skills, thus underlining the importance of lan-

guage output and practice (DeKeyser and Sokalski,

1996) (Izumi, 2002) when it comes to learning to

speak and write a second language.

Another debated issue in SLA research is the func-

tion of corrective feedback in general and the spe-

ciﬁc features, effectiveness and timing of different

feedback moves (Nicholas et al., 2001). Some re-

searchers claim that the evidence in favour of correc-

tive feedback is insufﬁcient, but various studies and

meta-analyses (Russell and Spada, 2006) tend to in-

dicate that, in general, corrective feedback has a pos-

itive effect on language learning. In addition, there

are indications that, to be effective, corrective feed-

back should be unambiguous, consistent, intensive,

should provide opportunities for self-repair and mod-

iﬁed output (Tatawy, 2002), should not be too asyn-

chronous as it is important to give feedback while the

procedural “knowledge” that led to the error is still ac-

tive in memory (DeKeyser, 2007) and should prefer-

ably take account of speciﬁc learner characteristics.

2.2 Results

The brief literature overview presented above has pro-

duced interesting results with respect to the desirable

features of ICALL systems intended to help L2 learn-

ers acquire the ability to produce grammatically cor-

rect L2 speech. First, these systems should provide

comprehensible input. Second, they should stimu-

late skill-speciﬁc output and practice. Third, they

should produce accurate corrective feedback that is

clear, consistent, intensive, timely, individualized and

that provides opportunities for self-repair and modi-

ﬁed output. In the following section we proceed to

analyzing a number of available systems to determine

to what extent they satisfy the criteria outlined above.

3 COMPUTER SYSTEMS FOR

GRAMMAR PRACTICE

To investigate how computer applications might be

able to help language learners improve the gram-

matical accuracy of their speech, we searched the

ICALL literature (e.g., Computer Assisted Language

Learning, ReCALL, Speech Communication, Sys-

tem, Language Learning and Technology, CALICO

Journal, and others) for reports of ASR-enabled sys-

tems that train grammar accuracy. Finding relevant

systems proved more difﬁcult than expected (an early

search yielded just one system), and as a result we

relaxed search criteria to include two categories of

systems: a) those that offer grammar training regard-

less of whether they utilize ASR (we refer to these

CSEDU 2011 - 3rd International Conference on Computer Supported Education

356

as “accuracy-focused” systems) and b) ASR-enabled

systems which implement exercises that do not fo-

cus on grammar training but which involve use of L2

grammar knowledge (we refer to these as “ﬂuency-

focused” systems). Relaxing these criteria increased

the number of relevant systems, and from these candi-

dates we selected those that come closest to meeting

the pedagogical requirements presented above.

This section provides an overview of the sys-

tems with respect to practice exercises offered, types

of feedback provided, and ability to accommodate

learner differences for the two categories of systems.

In the accuracy-focused category, we have the Ger-

man Tutor (Heift, 2001), CALLJ (Wang et al., 2009),

Te Kaitito (Vlugter et al., 2009), and a collection

of web-based language games from the CSAIL MIT

lab (Seneff, 2007); and for ﬂuency-focused systems,

we review the TLCTS (Johnson and Valente, 2009),

SPELL (Morton and Jack, 2005), CandleTalk (Chiu

et al., 2007), a prototype version of Let’s Go adapted

for language learners (Raux and Eskenazi, 2004),

and an unnamed prototype from the University of

oteborg (Bergenholtz, 2004).

3.1 Practice Activities

We ﬁrst present accuracy-focused systems before ex-

amining the potential for grammar training in systems

oriented towards ﬂuency.

Practice in the accuracy-focused systems is char-

acterised by narrow, skill-focused exercises that de-

velop the learner’s ability to produce grammatically

accurate utterances. Systems such as the German

Tutor and CALLJ are good examples of accuracy-

focused systems: learning is structured into dif-

ferent lessons that cover vocabulary and grammar

items found in language learning materials (e.g. the

Japanese Language Proﬁciency Test). Each lesson

contains activities that challenge the learner to apply

target language grammar rules to produce an error-

free phrase or sentence; when the learner arrives at

an answer they can submit it to the program for

evaluation. Activities typically resemble activities

found in traditional classroom materials, such as cloze

tests, and sentence-building and translation exercises.

A different approach in accuracy-focused instruction

has been to provide more realistic contexts (in the

sense that the exercises more closely resemble the ev-

eryday situations where the learners will be expected

to use the target language to communicate). An exam-

ple of this approach can be found in Te Kaitito, which

implements an accuracy-focused exercise that takes

the form of text-based dialogues between the learner

and a virtual tutor character. The dialogues are speci-

ally designed to elicit language that requires use of

target L2 forms known to be difﬁcult.

Practice in ﬂuency-focused systems stands in con-

trast to the previous group. These systems focus on

helping learners to become functional in the target

language, so as to be able to satisfy their everyday

communicative needs, such as describing one’s likes

and dislikes or asking for directions. Exercises are

task-based and allow the learner to converse with vir-

tual characters in a target language. Systems differ in

the richness of the simulation: in the Let’s Go system,

learners telephone an automated spoken dialogue sys-

tem for bus information; CandleTalk and the Bergen-

holtz system provide virtual two-dimensional charac-

ters with whom to interact; SPELL and TLCTS place

learners in a three-dimensional virtual world (naviga-

ble via a player-controlled avatar) populated with vir-

tual characters.

With respect to training for oral communica-

tion, one aspect that differentiates some applications

from others is the modality supported by the pro-

gram. Many accuracy-focused systems require that

the learner produce written output by using the key-

board, or through selection exercises that rely on use

of the mouse. CALLJ and the collection of web-based

games are notable exceptions, as they are the only

accuracy-focused systems we are aware of that em-

ploy ASR to implement grammar training exercises

that allow the learner to speak to the system with a

microphone.

In the process of completing exercises like those

presented above, learners will inevitably produce lan-

guage that contains errors. Next, we examine how

ICALL systems respond to errors with corrective

feedback.

3.2 Corrective Feedback

ICALL systems can help learners to reduce the fre-

quency of errors in their speech by drawing attention

to, explaining and supplying corrections for errors

through the provision of automatic corrective feed-

back. In the context of a practice exercise, an appli-

cation provides feedback by utilizing specialized lan-

guage or speech technology to monitor learner output

for errors. When errors are discovered, the system

processes the error to prepare a feedback response,

which can be described as a message or behaviour that

communicates that one or more errors were detected

and provides information to guide the learner to cor-

rect the error(s). Naturally, the details of the feedback

message vary from system to system, and may be in-

ﬂuenced by pedagogical and technical considerations.

One factor that inﬂuences feedback is the pedago-

COMPUTER-ASSISTED GRAMMAR PRACTICE FOR ORAL COMMUNICATION

357

gical focus of the system. In accuracy-focused sys-

tems, emphasis on the reduction of grammar errors

seems to motivate system designers to use unam-

biguous, explicit feedback. The German Tutor and

CALLJ systems use metalinguistic descriptions to

provide error-speciﬁc feedback that includes explana-

tions, such as in the example “Watch out! The verb

WOHNEN and the subject ICH do not agree. ICH is

singular” from the German Tutor. Other options are

explicit recasts or clariﬁcation requests which omit

detailed metalinguistic language and instead supply

corrected versions of an utterance. An example of

this kind of feedback from the Te Kaitito system is

“There’s a mistake in that sentence. Maybe you mean

-suggested-correction-? Let’s try again”.

For ﬂuency-focused systems, we observed that

there is a tendency towards implicit feedback, feed-

back that is less recognizably corrective and more eas-

ily integrated into the simulation. Perhaps the most

simple example can be found in TLCTS, in which

characters respond to errors with simple (and perhaps

realistic) requests for clariﬁcation such as “Sorry, I

don’t understand”. A second example, this time of an

implicit recast from the Bergenholtz system, can be

seen in the following interaction: “Student: How old

years is Dirk? System: How old is Dirk? Dirk is 30

years (old)”. An explanation for the tendency towards

implicit feedback is that ﬂuency-focused practice ex-

ercises are designed to be authentic and that frequent

error correction with overly explicit feedback such as

metalinguistic descriptions may detract from the real-

ism of the simulation.

A second factor that inﬂuences feedback in

ICALL systems is technical feasibility. Reliable per-

formance of the language and speech processing mod-

ules that detect and classify errors is important, as

poor performance may result in errors going unno-

ticed or the potentially worse case of errors being re-

ported for well-formed utterances. For this reason,

system developers must take the limits of ASR com-

ponents into consideration when designing feedback

properties of a system. The current shortage of stud-

ies investigating the reliability of corrective feedback

for grammar in ASR-enabled systems prevents strong

conclusions, but existing studies suggest that correc-

tive feedback is more reliable in constrained practice

of the type seen in accuracy-focused systems. Perfor-

mance results for CALLJ suggest adequate reliabil-

ity, while data for the SPELL and Let’s Go systems

suggest otherwise, leading the authors of the ﬁrst sys-

tem to state that performance is “not high enough to

support precise analysis of learner errors” (Anderson

et al., 2008, p.616).

In some systems, feedback properties may also be

affected by the proﬁciency level of the learner. In the

next section, we look at how feedback and other as-

pects of the system can be adapted to meet individual

learner needs.

3.3 Individualized Instruction

Computer systems have excellent logging capabili-

ties which can be used to make detailed recordings

of the interaction between the system and the learner.

ICALL systems employ student models to organize

the information they collect into data structures that

can capture different information about the learner.

The type of information captured in a student model

varies between systems, but some of the more com-

mon types concern study activity, including lessons

completed and time on task, and performance on ex-

ercises and quizzes (for examples, see TLCTS and

the German Tutor). Additionally, some systems la-

bel practice exercises with metadata related to peda-

gogical objectives, with the beneﬁt being that results

of completed exercises can then be taken as an indi-

cation of the degree of mastery with respect to the

objectives. Examples of metadata types are linguistic

items, such as particular classes of words or gram-

matical constructions (Heift and Nicholson, 2001), or

task-oriented skills the learner is expected to acquire

(Johnson and Valente, 2009). More advanced student

models, in addition to the above, also capture type

and frequency information about errors the learner

produces, such as spelling, word order, and speciﬁc

grammar errors (Heift and Nicholson, 2001).

On its own, the information captured by a student

model is valuable to the student and educators as a

form of summative feedback which can serve as an

indicator of progress towards the goals of the system.

However, the most advanced systems go one step fur-

ther and use this information to adapt different aspects

of the system with the aim of optimizing the instruc-

tion to a learner’s speciﬁc needs. For example, the

German Tutor uses a student model to adjust proper-

ties of the feedback by varying the amount of detail

that is included in the error message according to the

learner’s skill level. Other uses of the student model

in this system include guiding decisions about what

errors to provide feedback on when multiple errors

are detected in learner input and generation of reme-

dial exercises based on observed student difﬁculties.

Summarising our observations, we ﬁnd that prac-

tice in the systems surveyed can be grouped into

accuracy- and ﬂuency-focused systems. Practice in

both groups provides opportunities for written and

spoken output. Feedback types differ, with grammar-

focused systems use explicit feedback to point out er-

CSEDU 2011 - 3rd International Conference on Computer Supported Education

358

rors to learners, while ﬂuency-focused systems use

implicit feedback forms likely to integrate well with

simulated spoken exchanges. Elements of individu-

alised instruction can be found in both groups, includ-

ing adaptation of course content, selection of errors

for feedback treatment and type of feedback provided,

and suggestion of remedial exercises.

The purpose of the work so far has been to evalu-

ate the suitability of current ICALL systems to pro-

vide grammar practice for oral communication. In

the next section we discuss the results of this analysis

and present recommendations that support our posi-

tion that developing an ICALL system that provides

grammar practice for oral communication is indeed

possible.

4 RESULTS AND

RECOMMENDATIONS

In section 3 we have reviewed a number of ICALL

systems. This analysis has revealed that each of the

systems reviewed falls short of fulﬁlling the require-

ments for optimal computer-based grammar practice

for oral proﬁciency.

With respect to type of practice and interaction,

we have observed that few systems allow the learner

to speak their output and are capable of processing the

learners speech to detect errors and provide detailed

corrective feedback on grammatical errors, while it

seems that it is exactly the combination of spoken

interaction and form-focused activities with detailed

corrective feedback that is particularly required to im-

prove grammatical accuracy in L2 speaking. Find-

ings that support the skill-based character of procedu-

ral knowledge underline the importance of speaking

practice to learn to speak an L2 (DeKeyser, 2007).

The attested failure of immersion students to achieve

target-like grammatical accuracy in speaking empha-

sizes the importance of form-focused activities and

feedback to draw attention to less salient linguistic

features.

In spite of these ﬁndings, we are convinced that it

is possible to develop ICALL systems that exhibit the

desirable properties presented above. In the remain-

der of this section we explain how.

First of all, it is important to take account of tech-

nical limitations. ASR performs less-reliably on non-

native speech, and at the moment, best practices rec-

ommend limiting the ASR task to discriminating be-

tween a small number of candidate utterances. While

this may seem a considerable limitation, it is sufﬁcient

to allow the type of focused grammar practice seen

in the German Tutor and CALLJ. Previous research

from our lab on ASR-based corrective feedback on

L2 pronunciation has indicated that it is possible to

develop systems that supply accurate feedback (Cuc-

chiarini et al., 2009), provided that constrained and

pedagogically sound exercises are developed. In this

connection it is important to point out that provid-

ing accurate corrective feedback on pronunciation is

more challenging than providing feedback on gram-

mar (Morton and Jack, 2005). In more recent research

we are developing technology for providing feedback

on grammar (Strik et al., 2010).

As to the precise form of feedback to be used,

our analysis has indicated that ICALL systems em-

ploy a variety of different techniques to draw atten-

tion to learner errors, including simple clariﬁcation

requests, implicit and explicit recasts and explicit cor-

rection, with varying degrees of error-speciﬁc met-

alinguistic information. Studies that have examined

human-provided feedback do not provide deﬁnitive

answers as to which form of feedback is most effec-

tive (see section 2). Research on feedback provided

by ICALL systems seems to indicate that feedback

that is explicit, error-speciﬁc and indicates the loca-

tion of errors result in larger learning gains than less

speciﬁc types of feedback (see (Hanson, 2007) for a

review; see also (Heift, 2004)). Complicating the is-

sue is the fact that these studies refer to practice done

in the typed medium. To the best of our knowledge

no research exists which has looked into the effective-

ness of different feedback types in oral practice with

an ICALL system.

Because of the lack of conclusive ﬁndings regard-

ing feedback effectiveness, we recommend an ap-

proach that involves developing a library of feedback

techniques. This library would be useful for at least

two reasons: ﬁrst, it would allow researchers to con-

duct their own experiments to investigate the effect of

different feedback types in spoken grammar practice.

Second, the library could be used later as a resource

for individualising instruction, as different feedback

types may be more effective for different learners or

exercises.

Concerning individualised instruction, we make

the observation that the German Tutor, an accuracy-

focused system which does not include ASR, pos-

sesses superior adaptability compared to other sys-

tems reviewed. The student model inﬂuences the

amount of detail included in feedback messages,

guides error selection when input contains multiple

errors, and informs the module that generates reme-

dial exercises, all functions that greatly enhance the

adaptability of the system to individual learners and

which we endorse for ASR-enabled grammar prac-

tice systems. To achieve this functionality, a sys-

COMPUTER-ASSISTED GRAMMAR PRACTICE FOR ORAL COMMUNICATION

359

tem should implement a student model and supporting

framework capable of capturing information regard-

ing study activity, performance on practice exercises,

and detailed error-related data, in addition to attach-

ing metadata to practice exercises to label them with

the grammar feature(s) they target.

To summarize, on the basis of our reviews of SLA

literature on the one hand and of existing ICALL sys-

tems for oral grammar training on the other, we would

like to stress the importance of employing ASR tech-

nology to allow spoken interaction in relatively con-

strained exercises in which plausible, realistic com-

municative contexts are created, but where the learn-

ers output is predictable and for this reason can be an-

alyzed automatically in detail with acceptable levels

of accuracy. In turn, accurate error detection through

ASR constitutes the essential basis for providing skill-

speciﬁc individualized instruction, practice and feed-

back, thus meeting SLA requirements as much as pos-

sible.

At the moment, we are in the process of develop-

ing an ICALL system in line with the recommenda-

tions above (see Figure 1). The system uses ASR in

meaning-oriented question-and-answer exercises de-

signed to elicit forms known to be problematic for

Dutch L2 learners. A pilot experiment planned for the

spring of 2011 will investigate the effect of computer-

provided corrective feedback on grammar accuracy,

as well as the reliability and accuracy of the underly-

ing technology.

Figure 1: A screenshot of the ICALL system under devel-

opment. Learners speak into a microphone to answer ques-

tions posed by a virtual tutor about a ﬁlm clip. In this ex-

ample, the tutor is asking ‘How did Tom actually come to

the store?’, and the learner responds using the prompt and

phrases located in the bottom half of the screen.

ACKNOWLEDGEMENTS

This work is part of the research program Corrective

Feedback and the Acquisition of Syntax in Oral Proﬁ-

ciency (FASOP), which is funded by the Netherlands

Organisation for Scientiﬁc Research (NWO).

REFERENCES

Anderson, J. N., Davidson, N., Morton, H., and Jack,

M. A. (2008). Language learning with interactive vir-

tual agent scenarios and speech recognition: Lessons

learned. Comput. Animat. Virtual Worlds, 19(5):605 –

619.

Anderson, J. R. (1982). Acquisition of cognitive skill. Psy-

chological Review, 89:369 – 406 (37).

Bergenholtz, J. (2004). An interactionist and task-based ap-

proach to computer-assisted language learning mate-

rials. Master’s thesis, University of G

otheborg.

Brumﬁt, C. (1984). Communicative Methodology in Lan-

guage Teaching: The Roles of Fluency and Accuracy.

Cambridge University Press, Cambridge.

Chiu, T.-L., Liou, H.-C., and Yeh, Y. (2007). A study

of web-based oral activities enhanced by automatic

speech recognition for eﬂ college learning. Computer

Assisted Language Learning, 20 (3):209 – 233 (24).

Cucchiarini, C., Neri, A., and Strik, H. (2009). Oral pro-

ﬁciency training in dutch l2: The contribution of asr-

based corrective feedback. Speech Communication,

51 (10):853–863 (11).

DeBot, K. (1996). The psycholinguistics of the output hy-

pothesis. Language Learning, 46:529 – 555 (26).

DeKeyser, R. (2007). Practice in a second language, chap-

ter Introduction: Situating the concept of practice.

New York: Cambridge University Press.

DeKeyser, R. M. and Sokalski, K. J. (1996). The differ-

ential role of comprehension and production practice.

Language Learning, 46 (4):613–642 (29).

Ellis, N. C. and Bogart, P. S. H. (2007). Speech and lan-

guage technology in education: The perspective from

sla research and practice. In SLaTE-2007, pages 1–8

(8).

Gass, S. (1997). Input, interaction, and the second language

learner. Mahwah, NJ: Lawrence Erlbaum.

Hanson, R. M. (2007). Feedback in Intelligent Computer-

Assisted Language Learning and Second Language

Acquisition: A study of its effect on the acquisition

of French past tense aspect using an Intelligent Lan-

guage Tutoring System. PhD thesis, University of Ed-

inburgh.

Heift, T. (2001). Intelligent language tutoring systems

for grammar practice. Zeitschrift fr Interkulturellen

Fremdsprachenunterricht (online), 6(2):1 – 15 (15).

Heift, T. (2004). Corrective feedback and learner uptake in

call. ReCALL, 16(2):416–431 (15).

CSEDU 2011 - 3rd International Conference on Computer Supported Education

360

Heift, T. and Nicholson, D. (2001). Web delivery of

adaptive and interactive language tutoring. Interna-

tional Journal of Artiﬁcial Intelligence in Education,

12(4):310–324 (14).

Housen, A. and Kuiken, F. (2009). Complexity, accuracy,

and ﬂuency in second language acquisition. Applied

Linguistics, 30 (4):461 – 473 (13).

Izumi, S. (2002). Output, input enhancement, and the notic-

ing hypothesis. Studies in Second Language Acquisi-

tion, 24:541 – 577 (36).

Johnson, W. L. and Valente, A. (2009). Tactical language

and culture training systems: Using ai to teach foreign

languages and cultures. AI Magazine.

Krashen, S. (1982). Principles and Practice in Second Lan-

guage Acquisition. Oxford: Pergamon Press.

Krashen, S. D. and Terrell, T. D. (1983). The natural ap-

proach: Language acquisition in the classroom. Lon-

don: Prentice Hall Europe.

Morton, H. and Jack, M. A. (2005). Scenario-based spo-

ken interaction with virtual agents. Computer Assisted

Language Learning, 18(3):171–191 (20).

Nicholas, H., Lightbown, P., and Spada, N. (2001). Recasts

as feedback to language learners. Language Learning,

51 (4):719–758 (39).

Ranta, L. and Lyster, R. (2007). Practice in a second

language, chapter A cognitive approach to improv-

ing immersion students’ oral langauge abilities: The

Awareness-Practice-Feedback sequence. New York:

Cambridge University Press.

Raux, A. and Eskenazi, M. (2004). Using task-oriented spo-

ken dialogue systems for language learning: Potential,

practical applications and challenges. In ICALL-2004.

Russell, J. and Spada, N. (2006). Synthesizing research on

language learning and teaching, chapter The effec-

tiveness of corrective feedback for second language

acquisition: A meta-analysis of the research, pages

133–164 (31). Amsterdam: Benjamins.

Seneff, S. (2007). Web-based dialogue and translation

games for spoken language learning. In Proc. of

the Speech and Language Technology in Education

(SLaTE) Workshop, Farmington, Pennsylvania.

Strik, H., van de Loo, J., van Doremalen, J., and Cuc-

chiarini, C. (2010). Practicing syntax in spoken in-

teraction: Automatic detection of syntactic errors in

non-native utterances. In Proceedings of the SLaTE-

2010 workshop, Tokyo, Japan.

Swain, M. (1985). Input in Second Language Acquisition,

chapter Communicative competence: some roles of

comprehensible input and comprehensible output in

its development, pages 235 – 253. Rowley MA: New-

bury House.

Swain, M. and Lapkin, S. (1982). Evaluating bilingual edu-

cation: A Canadian case study. Multilingual Matters

Ltd., Clevedon, UK.

Tatawy, M. E. (2002). Corrective feedback in second lan-

guage acquisition. Working papers in TESOL and Ap-

plied Linguistics, 2 (2):1 – 19 (19).

VanPatten, B. and Cadierno, T. (1993). Explicit instruction

and input processing. Studies in Second Language Ac-

quisition, 15 (2):225 – 243 (19).

Vlugter, P., Knott, A., McDonald, J., and Hall, C.

(2009). Dialogue-based call: a case study on teach-

ing pronouns. Computer Assisted Language Learning,

22(2):115–131 (16).

Wang, H., Waple, C. J., and Kawahara, T. (2009). Com-

puter assisted language learning system based on dy-

namic question generation and error prediction for au-

tomatic speech recognition. Speech Communication,

51(10):995–1005 (20).

COMPUTER-ASSISTED GRAMMAR PRACTICE FOR ORAL COMMUNICATION

361