Wissenschaftlicher Sammelband, herausgegeben von Thomas Tinnefeld - unter Mitarbeit von Matthias Ballod, Jan Engberg, Katja Lochtman, Günter Schmale, Veronica Smith. Saarbrücken: htw saar 2016. ISBN 978-3-942949-11-8

Self-Repair in Language Learners’ Oral Discourse


An Vande Casteele & Katja Lochtman (Brussels, Belgium)


Abstract (English)

In the present paper, different strategies of self-initiated self-repair in oral productions of L2 learners of Spanish are examined. In order to do self-repair, learners first need to detect a problem and interrupt their speech flow, and then formulate an alternative. In second language acquisition research, language learners’ self-repair in oral discourse is often linked to learners’ proficiency level and their monitoring ability and as such is considered as facilitative of learning. This article is based on a study in the course of which language learners were assigned a story-telling task based on a Spanish TV series. After viewing the video compilation, the participants were asked to retell the story in their own words. The aim of the study was to investigate which kinds of spon­taneous self-repair strategies were frequently used by these language learners, which types of problems were mostly noticed and whether detection always lead to an appro­priate correction. Finally, it is discussed in what way the self-initiated self-repairs were related to language proficiency in general and learners’ fluency in particular.
Keywords: Self-initiated self-repair, second language acquisition research, lan­guage proficiency, fluency, Spanish as the L2

Abstract (Deutsch)
In diesem Beitrag werden Strategien zur selbstinitiierten Selbstkorrektur (self-initiated self-repair) in der mündlichen Produktion niederländischsprachiger Spanischlernender untersucht. Damit Lerner sich selbst korrigieren können, müssen sie zuerst das Pro­blem identifizieren, den Gesprächsfluss unterbrechen und anschließend eine Alter­native formulieren. In der Fremdsprachenerwerbsforschung wird die Selbstkorrektur in mündlicher Kommunikation oft mit dem Niveau der Fremdsprachenkenntnisse und mit Monitoring-Kompetenzen in Verbindung gebracht und könnte somit als lernfördernd betrachtet werden. In dem vorliegenden Artikel wird über eine Studie berichtet, bei der Spanischlernende Geschichten formulieren mussten, die auf eine spanische Fernseh­serie zurückgingen. Nachdem sich die Lernenden einige Ausschnitte angesehen hat­ten, mussten sie die Geschichten nacherzählen. Dabei wurde untersucht, welche spon­tanen Selbstkorrekturstrategien häufig eingesetzt wurden, welche Probleme von den Lernenden am häufigsten identifiziert wurden und ob die Identifikation von Fehlern auch zu deren gelungener Korrektur führte. Zum Schluss wird der Frage nachge­gangen, wie sich selbstinitiierte Selbstkorrekturen zu Sprachbeherrschung und Rede­fluss verhalten.
Stichwörter: Selbstinitiierte Selbstkorrektur, Zweitsprachenerwerbsforschung, Sprach­beherrschung, Redefluss, Spanisch als Zweitsprache



1 Introduction
One way of treating language learners’ self-repair in oral discourse in second language acquisition research is by linking it to learners’ proficiency level and monitoring ability (Fincher 2006, Kormos 1999, Lennon 1994, O’Connor 1988, Smith 2008, van Hest 1996). As such it has also been linked to error-detection ability (noticing-the-gap, Schmidt 1995), error correction and the role of modified learner output in language learning (Swain 1995). Furthermore, self-repair is seen as a process that learners perform automatically (Simpson et al. 2013) and as such could be interpreted as an indirect measure of language proficiency in general and fluency in particular (Housen & Kuiken 2009).

Repair is defined as “practices for dealing with problems or troubles in speak­ing, hearing, and understanding the talk in conversation (and in other forms of talk in interaction)” (Schegloff 2000: 207). When a given repair is carried out by the current speaker, it is referred to as self-initi­ated (Schegloff et al. 1977). According to Shehadeh (2001), self-initiations occur when the language learner notices that the interlocutor has not understood or has misunderstood an utterance, or when the utterance is ill-formed. As such, the language learner realizes that he or she needs to reformulate or modify his or her output (Sato 2008). Self-initiated repairs are considered facilitative of learning. Along with Swain (1993, 1995), Lyster & Ranta (1997) argue that the process of attempting to produce more accurate and more comprehensible output will push the language learner to reprocess and restructure his or her interlanguage towards modified output. According to Swain’s Output Hypothesis (1993, 1995) the important functions of such learner output are:
the hypothesis-testing function, in which learners are potentially testing their hypo­theses about the target language; the metalinguistic function, enabling learners to control and in­ternalize linguistic knowledge; and a noticing-the-gap function, in which learners perceive a difference between what they can say and what they want to say. (Sato 2008: 224)
Moreover, output is assumed to enhance fluency through practice (Swain & Lapkin 1995), whereby fluency, the capacity to communicate meaning in real time (Skehan 1995), is often considered as one aspect of language proficiency next to accuracy and complexity (Housen & Kuiken 2009). What is more, the Common European Framework of Reference for Languages also takes fluency as a component for describing overall proficiency (Council of Europe 2001).

2 Repairs, Self-Repairs and Self-Corrections

Repair was initially studied in the context of L1 speaker discourse (Fox & Jas­person 1996, Schegloff et al. 1977, Schegloff 1979, 1997, 2000). Today, the study of repair has been extended to the analysis of learner discourse (Kazemi 2011, Simpson et al. 2013, van Hest et al. 1997, Wong 2000, 2005). However, as Simpson et al. rightly state,
this has been accompanied by a narrowing down of focus onto error monitoring and correction, partly due to the tendency for L2 speakers themselves to attend more to errors than L1 speakers”. (Simpson et al. 2013: 145)
An error is generally referred to as form-focused in nature (i.e. syntactic, mor­phological or lexical).
Generally, repair is classified into four categories based upon who initiates the repair and who takes steps to resolve the problem:
  • self-initiated self-repair (SISR),
  • other-initiated self-repair (OISR),
  • self-initiated other-repair (SIOR) and
  • other-initiated other-repair (OIOR). (Schegloff 1997, 2000)
According to Schegloff, SISR or “self-initiation and a same turn repair is the most common and most successful” Schegloff (1979: 268) in the L1. Modi­fications made by language learners are generally referred to as self-initiated, self-completed repairs (Simpson et al. 2013). The aim of this research is to offer more insight into the self-initiated self-repair. According to Postma (2000: 98) self-repair refers to repairing discourse elements that are considered as erroneous by something else without external prompting. This generally hap­pens “within a short span of time from the moment of error occurrence.” Self-repairing is not only typical of human speech, as Postma adds (2000), we also do it when driving a car or riding a bike, when playing a music instrument, etc. In brief, in many actions, corrective strategies come into play.

Self-correction occurs when learners correct themselves; this means that errors are replaced by an alternative that is correct. Self-repair is considered as a more general and broader concept, since it refers to an alternation with respect to what was uttered initially. It implies corrections, but also clarifications of previous elements in a given discourse. Moreover, self-repair does not always imply a successful correction. Specifically focusing on self-initiated modified output, this study examines which kinds of self-repairing strategies are frequently used by language learners, which kinds of problems are noticed and whether detection leads to appropriate corrections.


3 Self-Repairs and Proficiency

Self-repairs by language learners have often been linked to gauging language proficiency (Sato 2008, Wong 2000, 2005), since they are considered to involve error-detection (noticing-the-gap, Schmidt 1995) and the production of modified learner output in language learning (Swain 1995). Self-initiated self-repair can be viewed as an indicator of L2 proficiency “based on the link found between proficiency and monitoring(Fincher 2006, Simpson et al. 2013, van Hest 1996). In this study, we are especially concerned with fluency as one aspect of language proficiency, also referred to as ‘fluency in the narrow sense’, which is often contrasted with the linguistic complexity and accuracy of learner language (Housen & Kuiken, 2009):
Fluency in the narrow sense is usually described in terms of speedy and smooth delivery of speech without (filled) pauses, repetitions, and repairs. (De Jong et al. 2015: 224).

As Housen & Kuiken (2009: 462) put it, fluency is often defined as
the ability to process the L2 with native-like rapidity or ‘the extent to which the language produced in performing a task manifests pausing, hesitation, or refor­mulation’. (Ellis 2003: 342)
In brief, in order to make a repair, learners must first detect a problem, interrupt their speech flow, and then formulate an alternative. This procedure has an influence on the language learners’ fluency.

Fluency is also linked to the learners’ control of their linguistic L2 knowledge, with “control improv[ing] as the learner automatizes the process of gaining access” (Wolfe-Quintero et al. 1998: 4). The construct is not uncontroversial, however. In L2 research, mainly oral production data have been analyzed to de­termine which quantifiable linguistic phenomena contribute to fluency in L2 speech (Housen & Kuiken, 2009). According to such studies, speech fluency is even considered a multi-componential construct in which different sub-dimen­sions can be distinguished,
such as speed fluency (rate and density of delivery), breakdown fluency (number, length and distribution of pauses in speech) and repair fluency (number of false starts and repetitions). (Tavakoli & Skehan 2005) (Housen & Kuiken, 2009: 265).
In this sense, the present study focuses on repair fluency, since in the case of self-initiated self-repairs, language learners realize that they need to reformulate or modify their output, which, according to Swain’s Output Hypothesis (1993, 1995), is facilitative of learning (see Introduction above). 
 

In L2 studies, some problems arose also concerning the operationalization of fluency, that is, how fluency can be validly, reliably and efficiently measured (Housen & Kuiken 2009; for inventories of fluency measures, Ellis & Barkhuizen 2005; Wolfe-Quintero et al. 1998). Accordingly, self-repairs are often found as part of formulas for measuring fluency as one aspect of proficiency. Examples of such fluency measures would be the number of:
  • words per minute,
  • pauses and fillers in a given number of words,
  • repetitions in a given number of words and - above all –
  • self-repairs in a given number of words.
However, such quantitative measures do not provide us with the information needed to answer our research questions. The present study, therefore, is a qualitative analysis of self-initiated self-repairs by L2 learners of Spanish in Brussels. 

4 The Study

4.1 Research Questions


Specifically focusing on self-initiated modified output, in the current study, we examine which kinds of self-repair strategies are frequently used by language learners, which kinds of problems are noticed and whether their detection leads to appropriate corrections. It is a descriptive study focusing on how learners successfully or un­successfully self-repair their utterances and what this means for their fluency as a function of their overall language proficiency.

To examine the functions of self-repair practices of Spanish L2 learners, we posited the following research questions:
   1. What kinds of self-repairing strategies (SISR) are used
       by  our language learners?
Our aim was to determine how learners oriented themselves to trouble sources (i.e. instances in need of repair) in their own talk and how they structured their subsequent self-repairs.
   2. What kinds of errors are noticed? Does error-detection 
       lead to appro­priate corrections?
Our aim was to improve our understanding of how participants used self-repair as a resource for modified output, which is believed facilitative of language learning.
   3. How do self-initiated self-repairs influence fluency as  
       part of language proficiency?
Our aim was to improve our understanding of the role self-initiated self-repairs play in measuring fluency as a part of language proficiency.

4.2 Unit of Analysis


A repair formally consists of two segments: a reparandum and the alteration (Heeman & Allen 1999). The reparandum is “the stretch of speech that the speaker is replacing” (Heeman & Allen 1999: 3). The second part is the alteration and serves as the replacement for the reparandum. The end of the reparandum is called the “interruption point” or “cut-off” and this point is not always situated at the end of a word, it can also appear in the middle of a word (Schegloff 1987: 212). This ‘cut-off’ is a sudden interruption of the flow of speech and assumes the role of repair initiator (Schegloff 1987: 212).
Example:1

(1) Beatriz oiga / oye todo esto. (X2)2

reparandum / alteration
A particular phenomenon of self-repairing is the use of editing terms. Editing terms are fillers such as uhm, no or pues. They appear after the detection of the problem, i.e. after the interruption of the speech flow. They are often considered as a sign of speech disfluency, indicating false starts or restarts, but they also have a communicative purpose. They enable the speaker to hold the floor and offer him some time to reflect. Instead of editing terms, there can also be a silent or unfilled pause (van Hest 2000: 77).
Example:

(2) Hablo con uhm habla con las modeles. (X12)

reparandum / editing term alteration
Next to uhm we may find other editing terms such as no and no sí.
Examples:

(3) Sí hay un un uhm algunos algunas mujeres que que son feas.

(4) es la fiesta de jubilación de el jefe anciano Francisco que va a dar una agradación, no las gracias a todo el mundo

(5) y todo el mundo está enfadado, no, sí enfadado.
In other examples, instead of a pause, an extra-linguistic element appeared, such as laughter. This can be seen as another non-verbal kind of pause filler.
Example:

(6) y que esta esto esta persona (risa) va a ayudarse con uhm el medicamentos

4.3 Research Method and Description of the Task


The current investigation is a descriptive study of the discourse of university students in the Bachelor program of Linguistics and Literary Studies at the Vrije Universiteit Brussel (Belgium). The participants were all students learning Spanish as a foreign language. They were between 18 and 21 years of age and they all speak other languages (Dutch L1, French L1 or L2, English L2 and sometimes German L2).

The participants were assigned an oral story-telling task, based on the Spanish TV series Yo soy Bea (Ugly Betty). In particular, the story-telling task was based on film fragments chosen from a compilation of extracts from this TV series found on YouTube.


The task was assigned individually and took about 20 minutes to complete. It consisted of a first viewing session of the video compilation including audio, in order to give the participants a general overview of the storyline. Then, the participants were asked to retell the story in their own words while watching the video fragment a second time, but this time, without sound. This procedure allows a story-telling time almost equivalent to the video’s time, which put some (time) pressure on the learners to process the L2 with native-like rapidity. By doing so, we avoided participants only offering a brief summary of the story; so they really had to retell the story by describing what they were seeing. Another characteristic of the task was the impossibility of adjusting the produced discourse after performing the task (contrary to the production of a written assignment). The participants had to retell the story without any extra pre­paration, so the L2 speech fragments obtained were comparable to what they would produce in real-time interaction. Next to the time pressure described, there was the unavailability of exterior help. As such, the focus of the study could be fully directed to the fluency of the participants’ L2 discourse.


Finally, the participants were not informed about the purpose of the study. The activity was presented to them as a normal oral exercise within a language course.


The participants were also asked to fill in a background questionnaire. In this survey, they were requested to give basic information such as their L1, age, gender, their previous language studies, and to rate their proficiency in other foreign languages. 
 

The data were recorded and transcribed. In order to give the reader an idea of the data used in this research, this is a small excerpt of a transcription:
y es Beatriz la secretaría fea

y uhm dan beses eh besos

y hay una conservación conversación y uhm Álvaro quiere dar muchísim- muchas bes- bes- muchos besos a todos las personas que eh están guapa guapas

uhm las personajes rían se rían uhm

pero hay también Beatriz que está feo fea y no quiere dar una bes- dan dar una no quiere no quiere dar una besa ella
The corpus is compiled of 20 transcriptions and consists of approximately 12,000 words, which results in an average of 600 words per transcription.

5 Results


5.1 Identification and Analysis of Self-Repairs


In the following section, the nature of the different self-repairing strategies will be identified, i.e. the types that frequently occur in our database of oral narra­tives produced by language learners of Spanish, and the efficiency of these self-repairs. 
 

Due to the nature of the task and the absence of interaction, all instances of self-initiated self-repairs were spontaneous self-repairs, without any external prompting. We found a total amount of 194 self-initiated self-repairs.


The repairs retrieved in our database were classified according to the nature or level of repair. Repairs were distinguished on the morpho-syntactic level such as morpheme modifications (e.g. verbal forms, agreement within noun phrases, and repairs in sentence structure), repairs on the lexico-semantic level (inclu­ding word choice), repairs on the phonological level (changes in pronunciation or word stress) and repairs on the discourse level (concerning information structuring and referent identification).


In Table 1, the respective number of these types of self-repair is displayed:

Type of Self-Repairs
Absolute Numbers
Percentages
Repairs on morpho-syntactic level
108
55.6%
Repairs on lexico-semantic level
69
35.6%
Repairs on phonological level
7
3.6%
Repairs on discourse level
10
5.2%
TOTAL
194
100%

               Table 1: Types of self-repair in absolute numbers

Repairs on the morpho-syntactic level appeared most frequently in our data­base (in 108 cases), followed by modifications on the lexico-semantic level (in 69 cases). Seven examples illustrated a phonological repair and in ten cases, a repair on the discourse level was found. 
 

An even more detailed analysis of the self-repairs used by the language lear­ners offers the following results (Table 2).

Within the group of morphological repairs, most of them are related to verbal forms (47 examples of the 108 repairs in this category (43,5%)). The following extracts illustrate this category.
Examples:

(7) Los dos hombres entran pero no quiero no quieren decir qué qué ha ocurrido uhm (X19)

(8) y los dos chicas le ayuden ayudan (X2)

(9) Las otras personas también están pensando pens- sí no piensando pensando (X18)
Many self-repairs in verbal morphology concern a repair of the morpheme in­dicating a person, but repairs of tense and mode appeared in the corpus as well.

Type of Self-Repairs
Absolute Numbers
Percentages
Linguistic Level
Absolute Numbers
Percentages
Repairs on the morpho-syntactic level
108
55.6%
Verb
47
24.2%
Determiner
34
17.5%
Noun
4
2.1%
Adjective
14
7.2%
Pronoun
2
1.0%
Preposition
2
1.0%
Sentence structure
5
2.6%
Repairs on the lexico-semantic level
69
35.6%
Verb
21
10.8%
Determiner
4
2.1%
Noun
21
10.8%
Adjective
12
6.2%
Pronoun
6
3.1%
Ser/estar
5
2.6%
Repairs on the phonological level
7
3.6%
Word stress
7
3.6%
Repairs on the discourse level
10
5.2%
Choice
referent
10
5.2%
TOTAL
194
100%

194
100%
Table 1: Analysis of different types of self-repairs
Apart from these, various agreement repairs were found, mostly with determi­ners (in 34 cases), as in (10).

Example:

(10) no sabe qué hacerlo con este información esta información (X13)

In other cases, adjectives were repaired.

Examples:

(11) ella no fue contente contenta (X8)

(12) las chicas son muy contento contento contentas (X12)

The category of lexico-semantic repairs represents 69 instances in our data­base (35.6%). Within this group, we distinguish single word choices and the use of longer items, such as collocational aspects and the use of ser and estar.
The self-repairs of single word choices are mostly verbs and nouns.

Examples:

(13) Quiere pelear pegar Gonzalo (X1)

(14) Bea oye un conversación entre Alvaro y su compañón compañero (X15)

(15) Está haciendo uhm una soliciture solicitud para trabajar (X1)

(16) se trata de uhm la secretaría uhm fea de un uhm un hombre de uhm de businessman (X20)

(17) la mujer dice que Beatriz no está alle no es no se se siento bueno (X12)

(18) pienso que Bea es un poco uhm amorada euhm enamorada de Alvaro (X15)

(19) que los chicos son estúpidos están estúpidos (X1)

We also noticed one instance of a lapsus.

Example:

(20) Y hay una conservación conversación (X18)

As for phonological interventions, only a few examples could be discerned: seven examples of repairs (3.6%) in the pronunciation were found and those were often erroneous, since the learners sometimes changed a correct form into a wrong one. Frequently, this was related to a peculiarity in Spanish verbal morphology: the flectional morpheme changes but the respective syllable is not stressed. This seems particularly difficult for language learners as they tend to focus on the verbal morpheme, but this part cannot be emphasized.

Example:

(21) la gente uhm escucha escuchá (X18)

Finally, some corrections on the discourse level were identified. Those were corrections of extra-linguistic referent indication. Generally, such repairs are due to some confusion in indicating the correct protagonist, in this particular case the person who tried to hit the other one.

Example:

(22) Diego quiere pegar a Die.. no no Álvaro quiere pegar a Diego (X13)

5.2 Efficiency of Self-repairs


This last section is dedicated to the efficiency of self-repair. It is clear that not all attempts at self-repair turned out to be successful. So only the success rate of self-repairs in our database will be analysed. Numerous successful corrections could be observed, but also several examples of erroneous self-repairs or failed attempts at self-repair could be identified. In some cases, a correct item was “repaired” into an erroneous one (as in (23)). On other occasions, we found attempts to correct linguistic items, but they were finished unsuccessfully (e.g. (24)).
Examples:

(23) La chica trae agua para Beatriz para que se sienta siente mejor (X2)

(24) Uhm Beatriz es está muy uhm no habla y uhm nadie sabe qué qué ha pasada (X14)
Still, an important aspect concerning the success rate apparently is the con­dition under which the task had to be performed. The participants were under time pressure as the video fragment went on and there was no time for them to reflect on and correct what they had said. Erroneous self-repairs therefore both hamper the fluency and the accuracy of the utterance.

Looking at the success rate of the self-repairs (Table 3), we can conclude that in 81% of the cases, the repair was successful. Out of the 194 instances of self-repair retrieved from our learner corpus, 157 (80.9%) were successful and only 37 (19.1%) were unsuccessful. These findings show that when learners notice an erroneous form, chances are high that they are able to change it into a correct version:


Successful Repairs
Failed Repairs
Total Repairs
Repairs on the morpho-syntactic level
91
(46.9%)
17
(8.8%)
108
(55.6%)
Repairs on the lexico-semantic level
56
(28.9%)
13
(6.7%)
69
(35.6%)
Repairs on the phonological level
1
(0.5%)
6
(3.1%)
7
(3.6%)
Repairs on the discourse level
9
(4.6%)
1
(0.5%)
10
(5.2%)
TOTAL
157
(80.9%)
37
(19.1%)
194
(100%)
Table 3: Success rate of self-repairs

A more detailed analysis of the success rate within the different categories in­dicates that repairs on the discourse level and on the morpho-syntactic level turned out to be most successful, whereas phonological repairs predominantly failed.
Agreement corrections with determiners were mostly corrected successfully, fol­lowed by modifications of verbal forms. Within the category of lexico-semantic changes, nouns were mostly corrected successfully. 
 
More details on the results within the different categories are documented in Table 4:


Successful Repairs
Failed Repairs
Repairs on morpho-syntactic level
Verb
91
(46.9%)
36
17
(8.8%)
11
Determiner
32
2
Noun
4
0
Adjective
12
2
Pronoun
1
1
Preposition
2
0
Sentence structure
4
1
Repairs on lexico-semantic level
Verb
56
(28.9%)
16
13
(6.7%)
5
Determiner
3
1
Noun
20
1
Adjective
8
4
Pronoun
6
0
Ser/estar
3
2
Repairs on phonological level
Word stress
1
(0.5%)
1
6
(3.1%)
6
Repairs on discourse level
Choice referent
9
(4.6%)
9
1
(0.5%)
1
TOTAL
157
(80.9%)
37
(19.1%)
Table 2: Analysis of the successful and unsuccessful repairs

Finally, within the category of failed attempts at self-repair, the following sub-types can be distinguished:

In 15 cases (7.7%), the error was not modified into a correct version, so the corrective attempt still led to an erroneous version:
Example:

(25) la gente son emocionado emocionados (X17)
In 14 examples (7.2%), the repaired form was slightly better than the original form, but still not correct.

Examples:

(26) Beatriz uhm oída ha oída (X13)

(27) la mujer dice que Beatriz no está alle no es no se se siento bueno (X12)

In 16 cases (8.2%), a correct form was changed into a wrong form.

Examples:

(28) esto es una un reunión de todo el personal de la firma

(29) la chica trae agua para Beatriz para que se sienta siente mejor

Within the category of phonological repairs, in four cases (2.1%), a correct item was changed into an erroneous one.

Example:

(30) la gente uhm escucha escuchá

Because of the emphasis in the verbal form, pronunciation turned out erro­neous.
In some cases, several repairs followed each other. For instance, within the category of successful repairs, in three cases (1.5%), there was first a correct form, then modified into an error after which it was modified again into the original correct form.
Examples:

(31) Las otras personas también están pensando pens- sí no piensando pensando (X18)

(32) y que esta esto esta persona (risa) va a ayudarse con uhm el medicamentos (X18)

(33) puede uh responder uhm respondar no responder (X18)

6 Conclusions

The present study on self-repairing strategies showed how L2 learners of Spanish try to regulate and alter discourse while speaking. Self-initiated self-repairs occur when discourse elements are repaired or modified without exter­nal prompting. The aim of the investigation was to find out what kinds of self-repairing strategies were frequently used by our language learners, what kinds of errors were noticed and whether error-detection led to appropriate cor­rections. Our research aim was to improve understanding of how participants used self-repair as a resource for modified output, which is considered facili­tative of language learning.

The results of our study showed that repairs on the morpho-syntactic level appeared most frequently and usually represented modifications of verbal mor­phemes or determiner agreement. On the lexico-semantic level, repairs of single nouns and verbs were predominant. As for the success rate of self-repairs, morpho-syntactic repairs turned out to be most successful, whereas phonological changes usually failed. With regards to our learners’ overall language proficiency, our results suggest that self-initiated self-repairs improve morpho-syntactic accuracy at the expense of learners’ fluency. As opposed to native speakers, language learners interrupt their speech flow by morpho-syntactic corrections accompanied by multiple editing terms such as uh, uhm or no in order to gain time. Nevertheless, in our study, self-repairs frequently led to correctly modified output and, may as such, be facilitative of learning (Swain 1993, 1995).



References

Council of Europe. (2001). Common European Framework of Reference for Lan­guages: Learning, teaching, assessment. Cambridge: Cambridge University Press.

De Jong, N.H., R. Groenhout, R. Schoonen & J. Hulstijn (2015). Second language fluency: Speaking style or proficiency? Correcting measures of second language fluency for first language behavior. In: Applied Psycholinguistics, 36, 223-243.

Ellis, R. (2003). Task-based Language Learning and Teaching. Oxford: Oxford Univer­sity Press.

Ellis, R. & G. Barkhuizen (2005). Analysing Learner Language. Oxford: Oxford Univer­sity Press.

Fincher, A. (2006). Functions of self-initiated self-repair in an advanced Japanese language classroom. Unpublished PhD Dissertation. Griffith University: Australia.

Fox, B. & R. Jasperson (1996). A syntactic exploration of repair in English conversa­tion. In: P. Davies (ed.), Alternative Linguistics. Descriptive and Theo­retical Modes. Amsterdam: John Benjamins, 77-134.

Heeman, P. A. & Allen, J.F. (1999). Speech repairs, intonational phrases and discourse markers: Modeling speakers’ utterances in spoken dialog. Computational Lingui­stics, 25 (4), 1-45.

Housen, A., &f Kuiken (2009). Complexity, accuracy and fluency in second language acquisition. In: Applied Linguistics, 30, 461-473.

Kazemi, A. (2011). An investigation into the relationship between the type of self-repair and structural complexity of utterance. In: Journal of English and literature, 2.4, 96-102.

Kormos, J. (1999). Monitoring and self-repair in L2. In: Language Learning 49 (2), 303-342.

Lennon, P. (1994). Self-correction and error in advanced learner spoken narrative. In: Bartelt. G. (ed.), The Dynamics of Language Processes. Essays in Honor of Hans W. Dechert. Tübingen: Gunter Narr Verlag, 85-103.

Lyster, R. & L. Ranta (1997). Corrective feedback and learner uptake: Negotiation of form in communicative classroom. In: Studies in Second Language Acquisition 19, 37-66.

O’Connor, N (1988). Repairs as indicative of interlanguage variation and change. In: Walsh, T. J. (ed.), Synchronic and Diachronic es to Linguistic Variation and Change. Georgetown University Roundtable 1988, 251-259. Washington, D.C.: Georgetown University Press.

Postma, A. (2000). Detection of errors during speech production: a review of speech monitoring models. In: Cognition 77, 97-131.

Sato, R. (2008). Self-initiated Modified Output on the Road to Comprehensibility. JALT Journal, 30 (2), 223-240.

Schegloff, E. (1979). The relevance of repair to syntax-for-conversation. In Givon, T. (ed.). Syntax and semantics: Vol. 12. Discourse and syntax. New York: Academic Press, 261-286.

Schegloff, E. (1997). Practices and actions: boundary cases of other-initiated repair. In: Discourse Processes 23 (3), 499-545.

Schegloff, E. (2000). When ‘others’ initiate repair. In: Applied Linguistics, 21 (2), 205-243.

Schegloff, E., G. Jefferson & H. Sacks (1977). The preference for self-correction in the organization of repair in conversation.In: Language, 53, 361-382.

Shehadeh, A. (2001). Self- and other-initiated modified output during task-based inter­action. In: TESOL Quarterly 35, 433-457.

Skehan, P. (1995). Analyzability, accessibility, and ability for use. In: Cook. G. & B. Seidlhofer (eds.), Principles and Practice in Applied Linguistics. Oxford: Oxford University Press.

Schmidt, R. (1995). Consciousness and foreign language learning: A tutorial on the role of attention and awareness in learning. In: Schmidt, R. (ed.), Attention and awareness in foreign language learning. Honolulu: University of Hawai’i, Second Language Teaching and Curriculum Center, 1-63.

Simpson, R., S. Eisenchlas & M. Haugh (2013). The functions of self-initiated self-repair in the second language Chinese classroom. In: International Journal of Applied Linguistics 23, 2, 144-165.

Smith, B. (2008). Methodological hurdles in capturing CMC data: The case of the missing self-repair. In: Language Learning & Technology 12 (1), 85-103.

Swain, M. (1993). The output hypothesis: Just speaking and writing aren’t enough. In: The Canadian Modern Language Review 50, 158-164.

Swain, M. (1995). Three functions of output in second language learning. In: Cook, B. & B. Seidlhofer (eds.). Principles and practice in applied linguistics: Studies in honor of H. Widdowson. Oxford, UK: Oxford University Press, 125-144.


Swain, M. & S. Lapkin (1995). Problems in output and the cognitive process they gene­rate: A step towards second language learning. In: Applied Linguistics 16, 371-391.

Tavakoli, P. & P. Skehan (2005). Strategic planning, task structure, and performance testing. In: Ellis, R. (ed.) Planning and Task Performance in a Second Language. Amsterdam: Benjamins, 239-273.

van Hest, E. (1996). Self-repair in L1 and L2 Production. Tilburg: Tilburg University Press.

van Hest, E., N. Poulisse & T. Bongaerts (1997). Self-repair in L1 and L2 production: an overview. In: ITL, Review of Applied Linguistics 117/118, 85-115.

Wolfe-Quintero, K., S. Inagaki & H.-Y. Kim (1998). Second Language Development in Writing: Measures of Fluency, Accuracy, and Complexity. Honolulu, HI: University of Hawai’i, Second Language Teaching and Curriculum Center.

Wong, J. (2000). Delayed next turn repair initiation in native/non-native speaker English conversation. In: Applied Linguistics 21 (2), 244-267.

Wong, J. (2005). Sidestepping grammar. In: Richards, K. & P. Seedhouse (eds.). Applying Conversation Analysis. Basingstoke, Hampshire: Palgrave Macmillan, 159-173.
____________________

1 The examples have been transcribed according to the original data and without 
   punctuation marks. So, they may contain linguistic errors.
2 These codes refer to the users, i.e. the students who produced these texts.