Judit Karmas* Attention and monitoring in a second language A qualitative analysis Abstract. This paper aims to refine the understanding of the attentional mechanisms of second language learners' self-repair behaviour by means of the qualitative analysis of two Hungarian learners' output and the detailed retrospective comments they made on their performance. The results show that the distribution of attention at these two levels of proficiency differs to a considerable extent. The preintermediate learner was found to pay special attention to the application of grammatical mies, whereas the upper-intermediate student's attention was directed at the lexical level. lt is argued that the difference between the learners' self-repair behaviours is due to the differing degree of automaticity of the participants' speech. 1. Introduction lt is almost a common place in applied linguistics that attention plays an essential role in L2 learning. In a recent study, SCHMIDT (2001: 15) claims that "the concept of attention is necessary in order to understand virtually every aspect of second language acquisition". In his overview, he convincingly argues that attention plays a central role in language development, variability, fluency, individual differences andin the effectiveness of instruction. Research evidence shows that conscious attention to input is necessary for learning to take place (SCHMIDT 1990, 1993, 1994). ROBINSON (1995) refined the conditions that are essential for acquisition by asserting that input will become intake if the detection of input is followed by rehearsal in short term memory. VANPATTEN (1990, 1994, 1996; V ANPATTEN/ CADIORNO 1993) conducted a number of experiments in which he examined how attention is divided between form and content in input processing. TARONE (1983) and TARONE/ PARRISH (1988) analyzed the accuracy of L2 speakers' performance in tasks that required varying amount of attention to be paid to linguistic form and meaning. The results of these studies indicated that the availability of attention influences the accuracy of leamers' output. The findings of another related research project showed that the discourse salience of a linguistic form also affects how much attention is paid to its correct production (TARONE 1985). Researchers investigating the production mechanisms involved in task based language learning examined how task characteristics (FOSTER! SKEHAN 1996, SKEHAN/ FOSTER 1996), planning conditions (FOSTERISKEHAN 1996, SKEHAN/ FOSTER 1997) and taskrepetition (BYGATE 1996) affect Korrespondenzadresse: Dr. Judit KORMOS, Ph.D., Eötvös Lorand University, Department of English Applied Linguistics, Ajt6si Dürer sor 19-21, 1146 BUDAPEST, Hungary E-mail: dolgos.l@mail.datanet.hu Areas ofwork: Psycholinguistics, Task-based language learning, Language testing. IFLllllL 32 (2003) Attention and monitoring in a second language. A qualitative analysis 117 whether L2 learners devote more attention to form or to the meaning conveyed. Attention is also related to the concept of second language fluency which SCHMIDT (1992: 358) defined as "automatic procedural skill" (based on CARLSON/ SULLIVAN/ SCHNEIDER 1989) that "does not require much attention or effort". As available attention is affected by working memory capacity, individual differences in this respect (HARR! NGTON/ SA WYER 1992) can influence to what extent L2 learners can attend to new input and their own speech. Attention also plays an important role in explicit L2 instruction, since teachers can direct learners' attention to specific features of the target language. The most important characteristics of attention are that attention is limited and is therefore selective (BROADBENT 1958, KAHNEMAN 1973). Attention can be controlled voluntarily (KAHNEMAN 1973, WICKENS 1989). KAHNEMAN (1973) and WICKENS (1989) claim that the allocation of attention is determined by two forces: conscious decisions based on the demands of the task to be performed and enduring pre-dispositions. WICKENS (1989) argued that performance on concurrent tasks deteriorates if both tasks draw on the same pool of attentional resources (for example maintaining two parallel conversations). Upon processing their speech, L2 learners need to rely on the same verbal resource pool, therefore the various phases of speech production need to compete with each other for attentional resources. TOMLINNILLA (1994) pointed out that attention has three functions, and consequently can be understood as alertness; orientation and detection. They defined alertness as "general readiness to deal with incoming stimuli or data" (1994: 190). Based on the expectations concerning the information to be processed, the orienting function of attention takes care of the allocation of resources. Upon detection, attention is focused on specific units of information. TOMLINNILLA claimed that alertness and orientation are not necessary for detection in L2 learning, whereas SIMARD/ WONG (2001) argued that the three functions of attention cannot be separated. Since my paper is mainly concerned with the detection function of attention, this debate will not be pursued here any further. Tue present paper has two main objectives. First, it provides a summary of previous mainly quantitative and theoretical research on the role of attention in monitoring second language speech. Secondly, it aims to refine our understanding of the attentional mechanisms in L2 self-repair behaviour by means of the qualitative analysis of two Hungarian learners' output and the detailed retrospective comments they made on their performance. The paper starts with a brief overview of earlier research findings concerning the role of attention in L2 monitoring, which is followed by a description of the research methods. Next, a detailed discussion of the transcripts is provided from which conclusions concerning the role of attention in monitoring L2 speech are drawn. 2. Review of previous research on the role of attention in monitoring 12 production The role of attention in speech monitoring has become the focus of interest only recently. Many of the previous studies in this field were only concerned with establishing the lFLlllL 32 (2003) 118 Judit Karmas distribution of various types of self-repairs, and did not attribute much importance to the discussion of the allocation of attention (e.g. FATHMAN 1980, LENNON 1984, VAN REST 1996). Another shortcoming of the research in this field has been that with the exception of a few studies (GREEN/ HECHT 1993; KORMOS 2000; POULISSE 1993, 1999) conclusions concerning the monitoring skills of L2 learners were drawn without the examination of the frequency and the correction rate of errors and their relationship to the frequency of self-repairs. Despite these problematic issues of research methodology, most researchers of L2 production claim that L2 learners pay considerably more attention to lexical appropriacy than to grarnmatical accuracy (e.g. FATHMAN 1980, LENNON 1984, POULISSE 1993, POULISSE/ BONGAERTS 1994, VAN REST 1996). KORMOS (2000), however, found that Rungarian L2 learners paid approximately equal attention to the appropriacy and adequacy of the informational content of their utterance as to linguistic accuracy. Nevertheless she added that the similarity of the proportion of corrected lexical and grammatical inaccuracies did not necessarily mean that her participants' attention was equally divided between monitoring for the lexical appropriacy and the grammatical accuracy of their message. She acknowledged that the lack of observable differences between the correction rate of grammatical and lexical errors might have been caused by the fact that in her study covert repairs were not investigated. The retrospective comments suggested that speakers made conscious decisions concerning the implementation of the repair in L2. lt was pointed out that this decision can be influenced by several factors such as the accuracy demand of the situation, the learner' s perception of how seriously the error impedes successful communication, and to what extent the correction decreases the fluency of the utterance. Tue present study aims to investigate this issue in more detail by means of a qualitative analysis of two transcripts. Investigations concerning L2 self-repairs also revealed that the frequency of repairs concerning the information content of the message varies across different types of tasks (POULISSE 1993, VAN REST 1996). To date, two large-scale research projects exarnined the effect of proficiency on the allocation of attention in monitoring. In line with earlier small scale studies (e.g. O'CONNOR 1988), VAN REST (1996) found that with the development of L2 competence, L2 speaker' s monitor becomes more sensitive to discourse level problems than to lower level structural errors. This finding was also replicated in KoR- MOS (2000). These results suggest that owing to the high level of automaticity of the speech encoding mechanisms of advanced learners, these speakers have additional attention available for monitoring which they use for checking the discourse level aspects of their message. On the basis of the quantitative analysis of the correction rate of lexical and grammatical errors, KORMOS (2000) also found that the amount of attention paid to the linguistic accuracy of the message remained constant in the three proficiency groups (intermediate, upper-intermediate and advanced) she investigated. She explained this finding by claiming that formally instructed foreign language speakers in countries where explicit grammar teaching plays a significant role can allocate their attentional resources and make decisions concerning error corrections in a different way from learners in a second lFLl.lL 32 (2003) Attention and monitoring in a second language. A qualitative analysis 119 language environment and from students instructed with a purely communicative method. The qualitative analysis of a pre-intermediate and an upper-intermediate students' behaviours described in this paper, however, provide a more refined picture about the distribution of attention at these two levels of proficiency. 3. Method 3.1 Participants The participants of this study were two Hungarian learners of English. The pre-intermediate learner called Janos had been studying English for 3 years when the study was completed (names were changed to protect participants' anonyrnity). He worked in the Hungarian armed forces and was participating in a preparation course for an intermediate level state exam in Veszprem (a city located approximately 100 kilometers from Budapest). Learning English was essential for his career in the rnilitary. He was 24 years old at the time ofthe study. He scored 21 points out ofthe 63 in the C-test, which placed him among the pre-intermediate learners. The other participant called Margit was a retraining Russian teacher who participated in a three year TEFL (Teaching English as a Foreign Language) course at Eötvös Lorand University, Budapest. She had been learning English for 8 years at the time ofthe study. She was 35 years old, and she was teaching English in an elementary school, and as a part time job she worked as an assistant for a company. She scored 46 points of the 63 in the C-test, which meant that she was an upper-intermediate speaker. I selected these two participants from my database for several reasons. First of all, they represent two very different types of learners in terms of proficiency, gender and education, and can therefore illustrate two different cases. Secondly, both of them provided detailed metalinguistic comments on their performance, thus the retrospective data I gained from them is very rich. 3.2 Procedures First a C-test, which had been validated by DöRNYEIIKATONA (1992) was adrninistered to the participants to measure their level of proficiency. The C-test contained three texts with 21 gaps each. The participants of the study who scored higher than 54 points out of 63 were classified as advanced speakers. Students with scores between 53 and 41 points were considered upper-intermediate speakers, and learners with scores below 40 points pre-intermediate speakers. Following the adrninistration of the C-test, the participants were interviewed one by one. First, they were asked to act out an approximately 5 rninute long information gap type role-play activity adapted from JONES (1991: 218) with me being the interlocutor. The participants had to assume the role of the manager of a restaurant, who was to answer an enquiry concerning a private room in the restaurant. I played the role of the customer. Instructions and the necessary background information were provided in the JFLIIIL 32 (2003) 120 Judit Kormos native language of the participants. The task was designed in a way that it would reflect real-life interaction. Its aim was to ensure genuine exchange of information, and to elicit the maximum possible amount of speech from the students while keeping my input to the minimum. The task could be divided into four distinct phases. First, the participants were requested to provide information on the conditions of hiring a private room. Next, the participant and me had to reach a compromise, as the customer (played by me) wanted to hire the room for a fewer number of people than specified in the conditions of the restaurant. Thirdly, I asked the participant to recommend some dishes from the menu, and finally, to describe the private room. These two questions were unexpected to the students as no information conceming these topics was provided on the information sheet. This task involved unpredictable interaction and considering new information, which seemed to place heavy cognitive load on the participants. Consequently, it was assumed that participants in the research would focus on meaning rather than on form. The retrospective interview was conducted on the basis of the guidelines set up by ERICSSON/ SIMON (1980, 1993) immediately after the task was performed. The interview was partly controlled as participants were asked to comment on specific aspects of their performance only, but the information they could provide was not predetermined. The recall of relevant information was aided by asking students to verbalize their thoughts upon listening to their speech on a tape recorder. The retrospective report was to a certain degree self initiated because the participants were requested to stop the tape when they found instances ofbreakdowns or self-repairs and comment on them. Nevertheless, I also asked questions if the students failed to reflect upon relevant hesitation phenomena or instances of self-correction. Due to the fact that I did not inform the participants that they would need to comment on their performance before carrying out the task, the request to provide retrospective comments was not supposed to influence task performance. The retrospective interview was carried out in Hungarian, which was the native language of both participants. The performance of the task and the subsequent retrospective interview were both video and audio recorded. The transcriptions of the tasks were done by trained transcribers, and I checked the transcriptions and transcribed the retrospective interviews. 3.3 Analysis First, all instances of overt self-repairs and errors were identified and classified. In this study four major groups of overt self-repairs were distinguished: different-information, appropriacy-, errorand rephrasing-repairs. In the case of different-information (D) repair, the speaker decides to encode different information than he/ she is currently formulating (LEVELT 1983). In the speech ofmy participants two types ofD-repairs were found: message-replacement and inappropriate information repairs. Message-replacement takes place if the speaker abandons the originally intended message due to lack of linguistic resources and replaces it with a different one. In the case of inappropriate information repairs, the speaker repairs the message because its information content is faulty (LEVELT 1983). IFLl.! L 32 (2003) Attention and monitoring in a second language. A qualitative analysis 121 Appropriacy- (A-) repairs are employed when the speaker decides to encode the originally intended information but in a more precise way (LEVELT 1983). Speakers resort to Arepairs when they have encoded inaccurate, ambiguous information that needs to be further specified (appropriate level of information repair), or if they have used either incoherent terminology (coherent terminology repair) (LEVELT 1983), ambiguous reference (ambiguous reference repair) (LEVELT 1983), or pragmatically inappropriate language (pragmatic appropriacy repair) (BREDART 1991). In the case of error-repairs, speakers repair an accidental lapse (LEVELT 1983). Such lapses can occur at every stage of speech processing, that is, during accessing words, grammatical and phonological encoding, and articulation (for a detailed review of mechanisms of language processing see LEVELT 1989, LEVELT/ ROELOFSIMEYER 1999). Repairs of lapses occurring at these different stages are called lexical (3 ), grammatical (4), and phonological error-repairs (4) respectively. The fourth main type of repair, rephrasing-repair involves the revision of the form of the speaker' s original message without changing its content. In this case the speaker repeats a slightly modified version of a word or phrase by adding something and/ or using paraphrase because of uncertainty about its correctness, but tries to convey the same original message. Rephrasing-repair is different from error-repair in that error-repairs merely involve the correction of accidental lapses, and, consequently, the issuing of the same pre-verbal plan in an unmodified form, whereas rephrasing-repairs are signs oflack of L2 competence (K0RM0S 2000). Covert repairs, that is, the corrections of inappropriate or erroneous output before articulation, were also identified. Since the reparandum is not articulated, one can infer that a covert repair was made from its indirect manifestations such as word or phrase repetitions, blocking, prolongation, syllabic repetition or silent pauses (P0STMA! K0LK 1993). The retrospective comments made by the participants also provided guidance in identifying these repairs. For the analysis of errors, LENN0N's (1991) definition of errors was used. In this definition an error is "a linguistic form or combination of forms, which in the same context and under similar conditions of production, would in all likelihood, not be produced by the speakers' native speaker counterparts" (1991: 182). On the basis ofthis definition, I identified all the possible instances of errors in the transcripts of the participants. In the English transcripts all the cases where no unambiguous judgments could be made were collected and were shown to two (an American and a British) educated native speaker judges. In an interview, which I conducted separately with the two judges, the native speakers were informed about the task the students had to perform and were shown the errors in their contexts. The judges were asked to decide whether they would produce the specific utterance in the given context. Only the cases that both of the judges considered unacceptable were counted as errors. Phonological errors were not studied, as the systematic differentiation of inaccuracies due to the accent of the speaker and those originating from lack of knowledge concerning the phonological form of lexical entries would have caused serious problems. After the identification of errors, all the instances of erroneous utterances were classilFILlllL 32 (2003) 122 Judit Karmas fied as either grammatical or lexical errors. In the present project, errors of lexis and grammar were studied from a psycholinguistic perspective, which means that errors were not classified on the basis their formal features, but according to where the error occurred in the psycholinguistic process of encoding the message. In order to obtain comparable results with the distribution of self-corrections, erroneous utterances were classified similarly to grammatical and lexical error-repairs. Grammatical errors were defined as inaccuracies that are the results of faulty grammatical encoding processes, whereas lexical errors were assumed tobe caused by inappropriate lexical access (see Appendix 1 on page 129). 4. Results and discussion First I compared the performance of the two participants as regards the frequency of selfrepairs, number of errors and the correction rate of errors (see Table 1 on the following page) as well as some other performance variables. Next I performed a qualitative analysis of the transcript which was aimed to complement the information I gained from the quantitative data (see Appendix 2 for the transcripts on page 130-132). The results of the quantitative and qualitative analyses will be discussed in parallel. Table 1 shows that repairs in general occur twice as frequently in the intermediate learner's speech than in the output of the upper-intermediate student. Error-repairs are even more frequent in Janos' s performance than in that of Margit. Margit repairs 11.76% of her lexical and grammatical errors, while Janos 21.05%. Janos seems to be more conscious ofhis grammatical errors because he repairs 29.63% ofthem, whereas Margit corrects only 18.18% ofherfaulty grammar. Theresults also show thatJanos corrects the information content of his message more frequently than Margit. ]F[,1JL 32 (2003) Attention and monitoring in a second language. A qualitative analysis 123 Rawcount Per 100 words Rawcount Per 100 words Total of repairs 8 1.64 19 3.40 Sum of error-repair 4 0.82 11 3.70 Lexical error-repair 0.21 3 1.01 Grammatical error-repair 2 0.41 8 2.69 Phonological error-repair 0.21 0 0 Sum of appropriacy-repair 0.21 0.34 Appropriate level of inforrnation 0 0 0.34 repair Arnbiguous reference repair 1 0.21 0 0 Sum of different inforrnation 2 0.41 6 2.02 repair Inappropriate inforrnation repair 0.21 4 1.35 Message replacement repair 1 0.21 2 0.67 Rephrasing repair 0.21 1 0.34 Sum of errors 44 7.01 57 19.19 Grammatical error 11 2.27 27 9.09 Lexical error 23 4.74 30 10.10 Grammatical error correction rate 18.18% 29.63% Lexical error correction rate 8.70% 13.33% Total correction rate of errors 11.76% 21.05% Speech rate (syllables/ sec) 1.97 0.98 Table 1: Comparison of the two participants' speech production behaviours On the one hand, some of the results are not unexpected. V AN REST (1996) also found that pre-intermediate learners make more repairs than advanced students. At the first glance, this would mean that Janos pays more attention to monitoring his speech than Margit. If we analyse the transcripts and the retrospective comments in more detail, however; it becomes apparent that a lotof the monitoring activities take place covertly in Margit's speech, whereas Janos makes no covert repairs. The following examples may illustrate this. FL1.IIL 32 (2003) 124 Judit Karmas (1) Margit: do you do you want to er eat together with the other er guests? Retrospective comment: The word guest did not come to my rnind first, then I was not sure whether somebody can be a guest in a restaurant, and all this went through my rnind very quickly. (2) Margit: do you change er your date Retrospective comment: As I was saying this sentence, I noticed that it was incorrect, but I did not want to repeat the whole sentence again. I was told by my teachers that repeating a sentence sounds strange, and it matters more that you produce the sentence fluently than the fact that it contains an error. (3) Margit: you have to er pay it in advance Retrospective comment: The expression in advance came automatically, and it just ran through my rnind whether this was correct, and I realised this was OK. (4) Margit: Er if you are ready to pay a little bit more er than the er ordinary or or ordinary er fee er it is possible of course Retrospective comment: lt did not come to my rnind, how average price is in English. I knew ordinary fee was not correct. I was searching for the right word, but I just could not find it. In addition to these four examples, another two instances of covert monitoring can be found in Margit's speech concem the pragmatic aspects of her message. (5) Margit: So you can you can choose er what you want. Retrospective comment: Immediately as I said this sentence, I realised this was not very polite, I just did not know how to say this more politely. (6) Margit: (Long pause) You're welcome. Retrospective comment: First the expression not at all occurred to me, but then I realised this was not appropriate in this situation. So I saidyou are welcome. lt seems that due to the higher level of automaticity of Margit' s encoding mechanisms, her speech is faster than that of Janos, and some of the processes can also run parallel (see also BÄRENFANGER 2002). This allows Margit to intercept errors before they are actually audibly produced. She is also able to pay attention to monitoring her speech withoufhaving to reduce the speed ofher speech significantly. Janos appears tobe unable to do this as no covert repairs were found in his speech. The high number of covert monitoring activities in Margit's speech therefore suggests that she pays considerable attention to monitoring, but this does not always surface in the form of overt repairs. Margit's attitude to errors is also different from that of Janos as she explicitly states that sometimes fluency is more important for her than accuracy (see Example 2 above). Having leamt English for a number ofyears already, she knows that getting her message across at an acceptable speed even if she makes errors is better than speaking very slowly but without any errors. Margit's attitude was probably formed and reinforced by the communication centred syllabus of the course she attended and by the several native speaker teachers she met at university. Janos, on the other hand, strives for accuracy and seems to correct almost every mistake he notices even if his message was perfectly understandable in its erroneous form. Janos does not have a long language leaming history, and his leaming experience was mainly shaped by the accuracy and grammarcentred teaching methods of typical Hungarian secondary schools (NIKOLOV 1999). The exam for which he was preparing also assesses the linguistic components of communica- IFLUJL 32 (2003) Attention and monitoring in a second language. A qualitative analysis 125 tive competence, and fluency constitutes only one mark out of the five scores given in the oral exam. If we compare Margit's repairs to some of the corrections in Janos's speech, the difference between the monitoring activities of the two learners becomes even more apparent. (7) Janos: we we can er put put it on into the er room some seat chair chairs uhm but er er not not much Retrospective comments: The word on came automatically, but I realised this was not correct; you should say put into. I thought that seats are fixed to the ground, for example, there are seats in cinemas or theatres, so I should use chair instead. I realised that I need to use plural after some. (8) Janos: there is there are some free room er there. Retrospective comment: I know there is plural after some, but I simply cannot say it correctly first. (9) Janos: but within within spitting distance in within er within spitting distance there is a very good hotel Retrospective comment: We learnt this expression not long ago, and I was not sure whether you need the in or not. (10) Margit: you can choose er a light music between light music or or music er by a er cassette player Retrospective comment: I realised that you need between if you have two options to choose from. I also noticed that you do not need the a before music because music is uncountable. Examples 1--4 above indicate that Margit's attention in monitoring is mainly directed at the lexical level. She is concemed with shades of meaning (such as whether someone can be a guest in a restaurant or whether ordinary fee is the same as average price) and the appropriate use of expressions. In her speech the application of most rules of grammar is automatic which is supported by the relatively low number of grammatical errors she makes, and she does not seem to have problems accessing the right words in everyday communication. Thus it can be assumed that a high number of grammatical and lexical encoding processes in her speech are automatised which allows her to pay attention to subtle lexical and grammatical issues. The two grammatical repairs in Example 10 are also related to lexis, since 'between' is the complement of the verb 'choose' and the repair of the article is also lexically driven. On the other hand, Janos is mainly concemed with grammatical errors that are rule based such as singular and plural marking and agreement, as well as with errors in basic level vocabulary. He admits in Example 8 that he knows the rule, but he cannot apply it correctly, andin Example 9 he is not able to retrieve a memorised unit correctly. These examples illustrate the lack of automaticity of Janos's speech encoding mechanisms; he has the declarative knowledge which has not become procedural knowledge yet. The figures in Table 1 would also suggest that Janos pays more attention to the information content of his message than Margit because in bis speech more different information repairs can be found. At a closer look, however, it tums out that out ofthe 6 Drepairs, two are message replacement repairs, which are made because Janos realises he does not have the linguistic means to express the intended message (Example 11), and he does not succeed in reformulating bis message. There is also one message replacement repair in Margit's speech (Example 12), but here the reformulation is necessary because FLllll]L 32 (2003) 126 Judit Kormos one word does not come to her rnind, and she produces a perfectly understandable alternative. In Janos's speech the rest of the D-repairs are false information. repairs (Example 13). lt seems that Janos is cognitively so much overloaded with linguistic problems that he is not able to pay proper attention to the information content of the task. Thus, he often provides incorrect information which he then needs to correct. (11) Janos: The room is free er all day er when arrive er here er the er we er we are uhm time Retrospective comment: I wanted to say that the party can start when you all arrive, but I could not say it in this form. (12) Margit: er if you are satisfied er with er my er with with this opportunity Retrospective comment: I wanted to say something like Jf you are satisfied with what I proposed or with my offer, but I could not say it so I decided to use other words. (13) J anos: And er it look like er uhm opened. Er we uhm ja december van. [it's December] (laughs). Er there is er a heater gas heater. Retrospective comment: I wanted to say that the windows can be opened but I realised that the conversation is taking place in December, so this did not make sense. Janos does not seem to have enough attentional resources to attend carefully to the level and quality of information he provides as in his speech no appropriacy repairs can be found. On the surface, Margit does not pay particular attention to this aspect of her speech either because among the overt repairs, there is only one appropriate level of information repair in her output. Examples 5 and 6 above, which are covert pragmatic repairs, however, indicate that she makes attempts to filter out pragmatic errors, but she seems to lack the appropriate pragmatic competence to do so successfully. 5. Conclusion Thein depth qualitative analysis of the self-repair behaviour of a pre-intermediate and an upper-intermediate learner has shown that the distribution of attention at these two levels of proficiency differs to a considerable extent. The low level of automaticity of the preintermediate participant's encoding mechanisms did not allow the interception of errors at the pre-articulatory level, thus no covert repairs were found in his speech. On the other hand, the upper-intermediate leamer often succeeded in correcting errors before they were articulated and performed a high number of covert monitoring activities. The preintermediate student was concemed with the correct application of grammatical rules, whereas the upper-intermediate participant's attention was mainly directed at lexical problems. Tue low proficiency participant seemed to be cognitively so overloaded by linguistic problems that he often failed to pay enough attention to the accuracy of information he provided. In contrast, the upper-intermediate student had enough attention to spare for some pragmatic problems, as well. The study reported in this paper has some lirnitations. First of all, as every case study, the description of these two leamers' self-repair behaviour is not generalisable. Moreover, as it was stated in the overview of previous research, the quantitative analysis of the same database from which these two cases were selected did not show that pre-intermediate ]F]Ll.l]]L 32 (2003) Attention and monitoring in a second language. A qualitative analysis 127 learners in general paid significantly more attention to grammatical accuracy than to lexical appropriacy (KORMOS 2000). This study, however, was not meant to produce generalisable results, but to provide a thorough analysis of two cases with the aim of understanding them. lt is hoped that the readers might recognise some of their students either in Janos or in Margit, so by understanding their self-repair behaviour, they can develop their students' fluency and accuracy. References BÄRENFÄNGER, Olaf. (2002): "Automatisierung der mündlichen L2-Produktion: Methodische Überlegungen". In: BÖRNER, Wolfgang/ VOGEL, Klaus (Hrsg.), Grammatik im Fremdsprachenerwerb. Kognitive, psycholinguistische und erwerbstheoretische Perspektiven. Tübingen: Narr 119-140. BREDART, Simon (1991): "Word interruption in self-repairing". In: Journal of Psycholinguistic Research 20, 123-137. BROADBENT, Donald. E. (1958): Perception and communication. Oxford: Pergamon. BYGATE, Martin (1996): "Effects of task repetition: Appraising the developing language of learners". In: WILLIS, Dave/ WILLIS, Jane (eds.): Challenge: Change inLanguage Teaching. London: Heinemann, 136-146. CARLSON, Richard A./ SULLIVAN, Michael A./ SCHNEIDER, Walter (1989): "Practice and working memory effects in building procedural skill". In: Journal of Experimental Psychology: Learning, Memory and Cognition 15, 517-526. DöRNYEI, Zoltan / KATONA, Lucia ( 1992): "Validation of the C-test amongst Hungarian EFL learners". In: Language Testing 9, 187-206. ERICSSON, K. Anders/ SIMON, Herbert A. (1980): "Verba! reports as data". In: Psychological Review 87, 215-251. ERICSSON, K. Anders/ SIMON, Herbert A. ( 1993): Protocol Analysis: Verbal reports as data. Cambridge, MA.: MIT Press. FATHMAN, Ann. K. (1980): "Repetition and correction as an indication of speech planning and execution processes among second language learners". In: DECHERT, Hans W./ RAUPACH, Manfred (eds.): Towards a crosslinguistic assessment of speech production. Frankfurt: Lang, 77-85. POSTER, Pauline / SKEHAN, Peter (1996): "The influence of planning and task type on second language performance". In: Studies in Second Language Acquisition 18, 293-323. GATHERCOLE, Susan E./ BADDELEY, Alan (1994): Working memory and language. Hillsdale NJ.: Lawrence Erlbaum. GREEN, Peter S. / HECHT, Karlheinz (1993): "Pupil self-correction in oral communication in English as a foreign language". In: System 21, 151-163. HARRINGTON, Michael/ SAWYER, Mark (1992): "L2 working memory capacity and L2 reading skill". In: Studies in Second Language Acquisition 14, 25-38. JONES, Leo (1991): Cambridge Advanced English. Cambridge UK.: Cambridge University Press. KAHNEMAN, Daniel (1973): Attention and effort. Englewood Cliffs, NJ.: Prentice Hall. KORMOS, Judit (2000): "The role of attention in monitoring second language speech production". In: Language Learning 50, 343-384. LENNON, Paul (1984): "Retelling a story in English as a second language". In: DECHERT, Hans W. / MÖHLE, Dorothea/ RAUPACH, Manfred (eds.): Second language productions. Tübingen: Narr, 50-68. LENNON, Paul (1991): "Error: Some problems of definition, identification and distinction". In: Applied Linguistics 12, 180-195. lFLl.lllL 32 (2003) 128 Judit Kormos LEVELT, Willem J. M. (1983): "Monitoring and self-repair in speech". In: Cognition 33, 41-103. LEVELT, Willem J. M. (1989): Speaking: From intention to articulation. Cambridge, MA: MIT Press. LEVELT, Willem J. M. / ROELOFS, Ardi / MEYER, Antje S. (1999): "A theory of lexical access in speech production". In: Behavioural and Brain Sciences 22, 1-75. NIKOLOV, Marianne (1999): "Classroom observation project". In: FEKETE, Hajnal / MAJOR, Eva/ NIKO- LOV, Marianna (eds.): English language education in Hungary: A baseline study. Budapest: The British Council, 221-245. O'CONNOR, Nadine (1988): "Repairs as indicative of interlanguage variation and change". In: WALSH, Thomas J. (ed.): Georgetown University Round Table in Languages and Linguistics 1988: Synchronie and diachronic approaches to linguistic variation and change. Washington DC.: Georgetown University Press, 251-259. POSTMA, Albert/ KOLK, Herman (1993): "The covert repair hypothesis: Prearticulatory repair processes in normal and stuttered disfluencies". In: Journal of Speech and Hearing Research 36, 472-487. POULISSE, Nanda (1993): Slips of the tongue and their correction in L2 learner speech: Metalinguistic awareness and second language acquisition. Paper presented at the 10 th AILA World Congress, Amsterdam. POULISSE, Nanda (1999): Slips of the tongue. Speech errors in first and second language production. Amsterdam: John Benjamins. POULISSE, Nanda/ BONGAERTS, Theo (1994): "First language use in second language production". In: Applied Linguistics 15, 36-57. ROBINSON, Peter (1995): "Attention, memory and the 'noticing' hypothesis". In: Language Learning 45, 283-331. SCHMIDT, Richard (1990): "The role of consciousness in second language learning". In: Applied Linguistics 11, 129-158. SCHMIDT, Richard (1992): "Psychological mechanisms underlying second language fluency". In: Studies in Second Language Acquisition 14, 357-385. SCHMIDT, Richard (1993): "Awareness and second language acquisition". In: Annual Review of Applied Linguistics 13, 206-226. SCHMIDT, Richard (1994): "Deconstructing consciousness in search of useful definitions for applied linguistics". In: AJLA Review 11, 11-26. SCHMIDT, Richard (2001): "Attention". In: ROBINSON, Peter J. (ed.): Cognition and second language instruction. Cambridge UK.: Cambridge University Press, 3-32. SKEHAN, PETER/ FOSTER, Pauline (1996): "Task type and task processing conditions as influences on foreign language performance". Paper presented at the AAAL Conference, Chicago, Ill. USA. SKEHAN, Peter/ FOSTER, Pauline (1997): "The influence of planning and post-task activities on accuracy and complexity in task based learning". In: Language Teaching Research l, 185-211. SIMARD, Daphnee / WONG, Wynne (2001): "Alertness, orientation and detection. The conceptualization of attentional functions in SLA". In: Studies in Second Language Acquisition 23, 103-124. TARONE, Elaine (1983): "On the variability of the interlanguage system". In: Applied Linguistics 4, 143-63. TARONE, Elaine (1985): "Variability in interlanguage use: A study of style shifting in morphology and syntax". In: Language Learning 35, 373-403. T ARONE, Elaine / PARRISH, Betsy (1988): "Task related variation in interlanguage. The case of articles". In: Language Learning 38, 21-44. TOMLIN, Russel/ VILLA, Victor (1994): "Attention in cognitive science and SLA". In: Studies in Second Language Acquisition 16, 185-204. VAN REST, Ema (1996): Self-repair in Ll and L2 production. Tilburg: Tilburg University Press. V ANPATTEN, Bill (1990): "Attending to form and content in the input". In: Studies in Second Language Acquisition 12, 287-301. ]F]Ll.ll]L 32 (2003) Attention and monitoring in a second language. A qualitative analysis 129 VANPATTEN, Bill (1994): "Evaluating the role of consciousness in SLA: Terms, linguistic features, and research methodology". In: AILA Review 11, 27-36. V ANPATTEN, Bill (1996): Input processing and grammar instruction. New York: Ablex. V ANP A TTEN, Bill/ CADIORNO, Teresa (1993 ): "Explicit instruction and input processing". In: Studies in Second Language Acquisition 15, 225-243. WICKENS, Carolle D. (1989): "Attention and skilled performance". In: HOLDING, David H. (ed.): Human skills. 2 nd edidtion. New York: John Wiley, 71-105. Appendixl The classification of errors (KORMOS 2000) Lexical error Inappropriate choice of content words my chef can make *cancer very good Inappropriate choice of prepositions and auxiliaif you need this room you need to tel1 me *before ries with independent conceptual specifications twenty hours Collocational errors we can cook er *to taste The erroneous production of a derivative form you have to write a *confirmament Grammatical errors Inaccurate use of inflectional morphology it *have to be er uhm uhm thirty five people Inappropriate choice of tense, aspect or voice of but I *don't mention er the room is er only on the the verb phrase er eighteen and on the nineteen of December free Faulty encoding of complements and specifiers Wrong word order you have to pay er the uhm twenty five er percent *the uhm the price Minimum er thirty five er people er have to be er then *can I er let it for you Inappropriate choice of prepositions and auxiliawe may make a contract if you er if you *will pay ries accessed by syntactic building procedures more * signals the beginning of the erroneous part of the utterance IFLwL 32 (2003) 130 Appendix2 Transcripts of the two participants' performance (silent pauses are not marked) lffi Margit (M), Researcher (R) Judit Kormos M: Good moming Madam! My name is Mrs X. I' m the owner of this eer er restaurant. And I would like to help you. R: Yes. We want to organise a party with my classmates and I would like to inquire about the possibilities in your restaurant. M: Aha er do you need er a spe a special separate hall er to arrange this meeting or er do you do you want to er eat together with the other er guests? R: Oh well, we would need a special separate room, yes, I agree. M: And er when will you er plan er to arrange this meeting? R: Well, we were thinking of either the 20 th of December or the thel9th of December. M: 20th and 19th? Just a minute, I have to er check my er information. Er I think we have to discuss it, because the separate hall is er free only on the er 18th and 19th of December. Er do you change er your date or? · R: Well, well we don't have to change the the the date, because you've mentioned that the 19 th is good. M: Andaha! R: And our second choice was M: aha, I missed it. Sorry. R: all right, Ok well. That's fine the 19th ofDecember, that's OK. All right. Can you tell me something about the costs and payment? M: Aha. Er this special hall is for about 40 people. Er but you have to pay er rent fee rent fee is the rent fee er for for a 35 people. And er you have to er pay it in advance. And er er if you if you er er if you are satisfied er with er my er with with this opportunity then you have to pay it er within er 24 er hours. And er yes. R: Oh well, OK. well that's fine, there is just one little problem namely that there will be uhm could between 24 and 35 people in our company. What happens if there is only 24 of us? M: Er I think er er we could er uhm we could make a compromise. Er if you are ready to pay a little bit more er than the er ordinary or or ordinary er fee er it is possible of course. R: All right, OK. And what is this ordinary fee? M: Ordinary? Eer er two h... er er two thousand er forints per per person. R: All right, OK. All right, that's fine then. And can you tell me something about the menu? M: Menu? Er we can make er er three course er I don't know what. So er so you can you can choose er er between er three types of meals meals. Er do I have to er encounter or? R: Oh well just tell me what this speciality or... M: Er the first is er er a fish menu the second is er er vegetarian menu and the third one is er er beef and and pork menu. So you can you can choose er what you want. I think, so it it must be satisfying. R: All right, OK. And we need to tell you in advance? M: Yes yes of course, because we have to plan er er things and and we have to arrange and decorate the hall er so as you er er to eajoy it your your meeting. R: All right, OK. Just can you mention decoration? Can you tell me something about the room? What does it look like? M: Er it is a special er er circle room. (coughs) And there is er er a very big er er table so you can you can be together and see each other er very well. Er and er you er can talk to each other of course and er you can choose er a light music between light music or or music er by a er cassette player. And er er the the major colours of the room are white and and blue. Er it can be a little bit cool but there are JF]Ll]]L 32 (2003) Attention and monitoring in a second language. A qualitative analysis 131 a lot of flowers plants. So it is it is very friendly, I think. And there there are a lot of er bulbs so it' s very light and cosy. So I can reco... recommend it. R: That sounds very good. Uhm OK so and what time should we arrive? M: Er I think er it would be better if you arrive er so er two hours before the er dinner. R: So I should arrive two two hours, but the whole group? M: Er I think er all of you should arrive er before er dinner because you can you can talk to each other. You can you can er enjoy the time before the dinner. But er it' s your choice. R: All right, OK. Just one more question. And how how long can we stay? M: How long can you stay? R: Yes. M: Er er the restaurant er closes er at er 4 o'clock er a.m„ So you can stay er until 4 o'clock. R: All right, OK. Well, thank you very much. M: You're welcome. lffi Janos (J), Researcher (R) R: Good morning. J: Er good morning Miss. Er can I help you? R: Yes. We would like to organize a party with my classmates and I wonder if you can help me. J: Yes. Er er you you have er I have we have (laughs) we have er a room er about er forty er er passenger and er er we can er eer er we can givee you er for at least thirsty five passengers er. Er and er you can er you can er pay er u... some money uhm before er the party. R: I see. All right. So would you say that we have to be 35, right? 35 people? J: 35, uhm good. R: Yes, but what happens if there is only 24 people? J: Uhm 24? U: hm er we we can er put put it on into the er room some seat chair chairs uhm but er er not not much. R: I see. All right. OK. And when is the room free? J: (sighs) Er the room is free er er only eeer er eighteen or nineteen of December. R: So it is only free on those two days. I see. J: But eer er then er the room is free er whole day. R: OK, all right, then it is free all day. But the 19th is good for us. J: 19? OK. Er uhm we we can give the room er 6 er hour er in moming next morning. R: Until 6 hour, so 6 o'clock in the moming. J: Until uhum yes. R: All right, OK. Well, I don't think we stay that late but that's OK. All right. Yes? J: (sighs) er we we can eer write a paper anderer fiftyfive percent er deposit er you can pay. R: So we have to pay 55%, did you say? J: No 50 er (laughs) twenty twentyfive percent. R: Aha 25%. OK, all right. And how much does the room cost? J: (sighs) The room is er uhm eer thirsty thirsty thousand er too much. (laughs) Er 10 thousand er forint er forints per day. R: 10 thousand forints for the whole night. J: And eer er you can er order uhm er some eat and beverage. R: And can you tell me something about the menu, what can we eat? J: (sighs) Er the menu is er a la carte or er we can er cook uhm er to taste. R: Uhum, uhum I see. And what do you recommend? What's your speciality? lFILi.nlL 32 (2003) 132 JuditKormos J: Uhm unfortunately er er there isn't er recommand er but er within within spitting distance in within er within spitting distance ther! ! is a very good er hotel and it is cheap and er er there is there are some free room er there. R: Aha. I was asking about the food. What do you recommend... J: Food yeah, aha. (laughs) Sorry. (laughs) Sorry. Er uhm eer there is some there are some specialists food er there are er Chinese food and European food er I recommend (wp) er er game games food. Because er my eer cook is very specialisf er games food in games food. R: Yeah, all right. lt sounds very good. And can you tel1 me something about the room? What is it like? J: (sighs) The room is er uhm very very light er uhm there i... there are er er er windows er very very much. And er it look like er uhm opened. Er we uhm ja december van. (laughs) Er there is er a heater gas heater and er uhm you can er heating you can heat er to taste. R: .Aha. That's good. All right, OK. And what time can we arrive? J: (sighs) Er er there are there is a big parking place er anderer er come er some bus er here. R: Uhum all right. And what time? When can we arrive? When is the room free for us? J: (sighs) The room is free er all day er when arrive er here er the er we er we are uhm time. R: All right. OK. Fine. OK then thank you and 1'11 get back to you with the paper. OK? J: OK. R: All right. See you. J: Thanks. See you? Nna. FLuL 32 (2003)