This article was downloaded by: [University Of Rhode Island] On: 13 February 2014, At: 18:53 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK The Language Learning Journal Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rllj20 Thinking aloud about L2 decoding: an exploration of the strategies used by beginner learners when pronouncing unfamiliar French words Robert Woore a a Department of Education, University of Oxford, UK Published online: 29 Mar 2010. To cite this article: Robert Woore (2010) Thinking aloud about L2 decoding: an exploration of the strategies used by beginner learners when pronouncing unfamiliar French words, The Language Learning Journal, 38:1, 3-17, DOI: 10.1080/09571730903545210 To link to this article: http://dx.doi.org/10.1080/09571730903545210 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Language Learning Journal Vol. 38, No. 1, April 2010, 3 17 Thinking aloud about L2 decoding: an exploration of the strategies used by beginner learners when pronouncing unfamiliar French words Robert Woore* Department of Education, University of Oxford, UK Decoding converting the written symbols (or graphemes) of an alphabetical writing system into the sounds (or phonemes) they represent, using knowledge of the language s symbol/sound correspondences has been argued to be an important but neglected skill in the teaching of second language (L2) French in English secondary schools. Several longitudinal and cross-sectional studies have highlighted poor levels of L2 decoding proficiency amongst beginner learners of French at Key Stage 3. However, there has been less investigation of these learners strategic reasoning when attempting to decode French words. Previous exploratory research in this area found that, in the absence of adequate knowledge of French decoding conventions, participants relied on English decoding processes to deal with French words. There was also some evidence that the most successful decoders were those who were aware of this influence of English and sought to move beyond it, thinking consciously about how to pronounce words in a more French way. The current small-scale, exploratory study set out to investigate in more detail the conscious strategies employed by participants as they tried to generate French pronunciations of unfamiliar words. Twelve beginner learners of French of varying attainment levels were asked to read aloud unknown French words, and to describe their thought processes. Reading the words aloud proved to be an effortful, conscious process for these participants, similar to a problem-solving task and very different from their automatic L1 decoding. They used a range of conscious strategies to support their L2 decoding, and there was a high degree of consistency among the strategies employed by the various participants. However, the strategies often led to incorrect outcomes because they were not underpinned by secure knowledge of French symbol/sound correspondences. Background Recent studies have argued that decoding proficiency plays an important part in second language (L2) learning for beginner learners of French in English modern foreign language (MFL) classrooms (Erler 2003, 2004; Woore 2009). There is also empirical evidence for a link between the L2 decoding proficiency of learners in this population and their wider motivation for language learning (Macaro and Erler 2008). Decoding refers here specifically to the process of converting the written symbols (or graphemes) of a language into the sounds (or phonemes) they represent using knowledge of the language s symbol/sound correspondences, such as when pronouncing an unknown word. 1 *Email: robert.woore@education.ox.ac.uk ISSN 0957-1736 print/issn 1753-2167 online Ó 2010 Association for Language Learning DOI: 10.1080/09571730903545210 http://www.informaworld.com
4 R. Woore French is still the most widespread MFL in England, taught in 99% of maintained schools (CILT 2008). However, despite a reassuringly familiar alphabet, it poses particular problems for English speakers in terms of decoding. Similarly spelled words may sound very different in the two languages (e.g. nation, chat). Recent evidence, both longitudinal (Woore 2009) and cross-sectional (Erler 2003, 2004, 2008; Macaro and Erler 2008; Woore 2006), suggests that Key Stage 3 (KS3) learners of French have poor L2 decoding skills, and make poor progress in this aspect of their language learning. If it is correct that L2 decoding proficiency is an important ingredient in learners overall motivation and success, then this could represent a serious impediment to their language learning. Erler (2004) argues that KS3 learners poor French decoding ability is a consequence of the lack of focus on this skill in classrooms, which mostly adopt a broadly communicative approach with an emphasis on oral skills (Klapper 2003, 34). Perhaps L2 decoding is assumed to be something which learners acquire implicitly, simply through contact with the language. However, this appears not to be the case (perhaps unsurprisingly given that pupils typically spend only about two hours per week in MFL lessons). Further, the introduction of the original KS3 Framework for language learning (DfES 2003) does not seem to have caused the improvement in pupils decoding proficiency that might have been hoped for, given its expectation that students receive explicit instruction in symbol-to-sound mapping. The evidence referred to above concerning KS3 learners poor L2 decoding proficiency was gathered several years after the launch of the Framework. Decoding strategies Despite the growing body of quantitative data on KS3 students decoding proficiency in L2 French, there has been little investigation into the strategies they use when decoding French words. Participants poor scores on decoding tests show that they were not correctly applying knowledge of French symbol/sound correspondences; however, the tests offer no insight into the actual mechanisms by which the incorrect pronunciations were generated. Understanding these strategies is important, because it may help in designing more effective programmes of instruction. Of course, learners L2 decoding may be partly subconscious. In their acquisition of L1 literacy, most KS3 pupils have reached a stage where they recognise many L1 words by sight, and decode unknown words automatically (see Ehri 2005). Because there is some overlap between French and English symbol/sound correspondences (especially in the case of consonants), applying these automatic, L1-based decoding processes to L2 words may yield partially correct pronunciations. For example, the French word gaz may be pronounced intelligibly by a monolingual English reader. However, the French and English systems also differ significantly. English-based decoding of French words often results in incorrect pronunciations. Beginner learners therefore need to move beyond English-based decoding and become acquainted with the new, L2 system. Just as in the early stages of learning to read in the L1, this will require conscious effort, drawing upon declarative rather than procedural knowledge (see Hulstijn 2005, 130f). It is the conscious mental actions taken by learners in order to produce French pronunciations of unknown words as opposed to simply pronouncing them as if they were English, in an unquestioning way that form the focus of the current study: learners decoding strategies (see Macaro 2006).
Language Learning Journal 5 Previous research To my knowledge, the only systematic investigation of the strategies used by KS3 learners of French when reading French words aloud was conducted by Erler (2003). Her research included interview-discussions with 23 Year 7 (Y7) students, all literate L1 English speakers, covering reading-related topics, including their experiences of L2 decoding. There was also a timed reading-aloud task using 10 individual French words, followed by discussions about how participants approached this task. A key overall finding was that most pupils appeared to be using their schema knowledge of English grapheme-phoneme conventions when pronouncing written French words, because they lacked the necessary knowledge of French decoding conventions (Erler 2003, 185). As noted above, this L1-based decoding may happen automatically and subconsciously, resulting in anglicised pronunciations of the kind familiar to many MFL teachers. This may apply not only to new words but also to words which students already know orally. It may generate incorrect pronunciations for these words, which conflict with the pronunciations they have heard in class. Indeed, some of Erler s (2003, 165) participants commented on this mismatch explicitly: when the written forms of words are first encountered, they look like a different word than what you ve been [...] repeating ; when Miss says it, it doesn t sound like [when I read it]. Similar comments arose when participants discussed the sounds of French words more generally: French words are spelt completely different to what I would usually think ; whereas English words are spelt... the same as how they re pronounced, (...) I know in French they re spelt different to how they re pronounced (Erler 2003, 166, 168). These last comments suggest a strong sense of English orthography as the norm, with French orthography perceived as puzzling and opaque. They also show that at least some learners are aware that the French and English orthographic systems differ, and therefore that English spelling/sound correspondences are an unreliable guide to pronouncing French words. In the discussions following the word reading task, too, English was found to play a strong part in participants approaches to dealing with French words. Almost all interviewees made at least one comment relating one or more of the cue words to English. Participants reactions to these perceived similarities between French and English words were classified into five categories (see Table 1, based on Erler 2003, 189f). Categories 4 and 5 show that at least some participants had developed a conception of French orthography as a distinct system. However, they face a dilemma: whilst they know that the French words should be pronounced differently than they would be in English, they lack secure knowledge of what the L2 Table 1. Categories of response to perceived similarities between English and French words in Erler (2003). 1. The word looks like English and is therefore easier to pronounce. 2. The word looks like English, so I ll pronounce it like the English word it resembles. 3. The word looks like English, but is not quite English; nevertheless I ll pronounce it like the English word it resembles. 4. The word looks like English, but is not quite English; so I ll add to the English pronunciation some sounds which I think might be French. 5. The word looks like English, but is not quite English; so I puzzle and struggle over deciding how it ought to be pronounced in French, rather than in English.
6 R. Woore pronunciations should actually be. In category 4, for example, the Frenchy touch which one participant says she adds to words appears vague and does not seem to reflect specific knowledge of French symbol/sound mapping. Similarly, in category 5, the same respondent says she knows that there are silent letters in French, but does not know specifically which letters or in which positions these are. Dealing with this uncertainty may explain the considerable, conscious effort on the part of learners in category 5 who struggle and puzzle over how to pronounce French words. A possibly related finding (Erler 2003, 183f) is that five of the seven pupils who took longest to pronounce the words on the timed reading-aloud task were higher achieving students who actually produced the best pronunciations. Further, these students spent longer pronouncing French words which resembled English than they did pronouncing words less similar to English. Erler suggests that this is because these learners were not content with their immediate English-based pronunciations; rather, they engaged in a conscious process of combating the heuristic or schema of English orthography-phonology conventions, trying to arrive at a French pronunciation of the word. Overall, Erler s findings paint a picture of students who lack L2 French decoding skills, and who perhaps unsurprisingly base their pronunciations of unknown French words on English symbol/sound correspondences. However, there are interesting individual differences between participants. At least some appear to know that their decoded output is wrong; some then engage in an effortful, conscious struggle to produce more authentic pronunciations. However, in many cases they lack a secure knowledge base for doing so, beyond a vague notion that French decoding should somehow be different from English. Erler s study was exploratory in nature, being based on a relatively small sample and focusing on 20 individual grapheme/phoneme correspondences. Further, several words used in the reading-aloud task resembled English words, possibly influencing participants approaches to decoding them. Further investigation is therefore required, to see whether this strong reliance on English-based processing is also found in other groups of learners when they decode words less visually similar to English words. Particularly interesting are the questions of whether learners are aware of any role being played by their L1 when dealing with French words, and if so, in what ways they react to this awareness. The current project addresses some of these issues. The research question is: What strategies do 12 beginner learners in Y7 use when decoding unfamiliar French words? Method The sample comprised 12 learners (five male, seven female) in the last few weeks of Y7, drawn from four French classes in a large, mixed comprehensive school in South-East England. Students at the school, chosen as a convenient sample, are atypical of the national population in certain respects: for example, students with English as an additional language, students from households of lower socio-economic status and students with special educational needs are all under-represented. The four participating classes were selected randomly from seven parallel classes in the year group, none of which was grouped by previous attainment. From each class, the French teacher nominated one higher-, one middle- and one lower-achieving
Language Learning Journal 7 learner. All participants were monolingual native speakers of English who reported that they had not studied languages before Y7. Consent was obtained from participants and their parents. Ten French words (Appendix) were selected as the focus for individual self-report discussions, on the basis that: (i) they contained graphemes known to pose problems for English-speaking learners (Woore 2006); and (ii) participants were unlikely to know them (later questioning confirmed this to be so). This meant that participants could not access pre-stored phonological forms for the words; they had to generate pronunciations via symbol-to-sound decoding. Participants were asked to read aloud each of the words, presented individually via computer, and to describe in as much detail as possible any thought processes relating to pronouncing them. Time and care were taken to make participants feel at ease; it was stressed that there was no question of assessing their thinking as right or wrong. After an initial prompt from the interviewer, the direction of the discussion on each word was determined by the participants themselves: they were free to pursue their own lines of thinking, reflecting the exploratory nature of the research. Additional prompts were given as necessary (e.g. Why do you think that? ). Discussions lasted around 25 minutes per participant. Not all of the 10 words were covered by all participants: the number of words covered in each case depended on how much s/he had to say about each one. All discussions were audio-recorded. Extracts deemed to yield interesting information were transcribed, giving approximately 10,500 words of verbal protocol comprising 89 discussions about individual words. A grounded approach was taken to analysing these data: they were not classified into pre-existing categories, but rather, the categories for the analysis emerged from the participants responses themselves. Conscious verbal reports have various limitations as a source of insight into internal processing (Seliger 1983). However, they are a well-established tool in researching problem-solving behaviour (Kormos 1998) including language learning strategies (Macaro 2001, 44ff). This was considered a valid sphere of enquiry for the current study: though the ultimate aim in language learning may be the fluent and automatic decoding of L2 words, similar to what happens in the L1, the first steps along the road to this end-goal may be rooted in conscious, strategic reasoning. Findings Occasionally, participants were unable to explain the pronunciation they had given for a particular word: there were nine comments in total, made by six of the participants, indicating that a pronunciation just came to them; they just said it. Three participants also said they had just guessed the pronunciations. This may reflect automatic processing of the L2 letter-strings, inaccessible to conscious scrutiny. However, comments of this kind were the exception. Three hundred and forty-two comments were recorded where participants engaged in strategic reasoning, either describing their thought processes in advance of pronouncing a word, or retrospectively justifying their pronunciation of it. These comments can be classified into categories which fall into three broad groups. A. Participants developing conception of L2 decoding The first group contains strategies where participants appeal to a developing though in many cases inaccurate conception of L2 decoding.
8 R. Woore A1. Appeal to declarative knowledge of L2 symbol/sound mappings All participants made explicit comments (43 in total) about how a particular French grapheme or group of graphemes should be realised phonemically, as illustrated in examples 1 2 below: 2 1. 5te moin4: the O and the I make a different sound than when they d just be on its own, together they make (...) like a /a/ /wa/ sound 2. 5houblon4: Ooh, and you don t pronounce H s This shows that declarative knowledge can be used successfully by beginner learners to support L2 decoding. However, it is obviously crucial that this declarative knowledge be correct. Whilst the verbal protocols contained 17 comments (made by nine participants) containing correct statements about particular symbol-to-sound correspondences, there were also 19 comments (made by 11 participants) containing incorrect declarative knowledge, as in examples 3 4: 3. 5te moin4: I knew that OI made a sort of /OI/ sound 4. 5te moin4: erm cos I think if it s got that dash thing on top of the E it s /i:/ Often, there is no way of determining the origin of this declarative knowledge, but in some cases participants comments link it to previous explicit instruction, as in 5 6: 5. 5caleçon4: the second C has got like a little 5 under it that s what Madame B told us 6. 5te moin4: she said that an accent changes the E to an /ei/ By contrast, example 7 may be attributable to implicit learning, which however in this case has resulted in unsound declarative knowledge. In order to help decode the 5c 4 in the unfamiliar word caleçon, the participant draws an analogy with the familiar question Ça va?, which also contains this grapheme. However, the participant has formulated an incorrect hypothesis about the effect of the cedilla on the pronunciation of this phrase, apparently assuming that it is connected to the rising intonation pattern characteristic of question forms in French. This example seems to paint a picture of a learner striving to make sense of the L2 decoding system, but who without explicit guidance has formed an incorrect hypothesis relating to a completely different aspect of speech production: 7. 5caleçon4: Participant: that bit there [pointing to the cedilla] it s like in erm when you say ça va? Researcher: and what does it do in the word ça va? Participant: it puts more of an accent to it (...) so you might say it like when it goes up Taken together, these examples could be used to support the argument that declarative knowledge about decoding should be taught to beginner learners, since it appears that, at least sometimes: (i) declarative knowledge can help learners to work out correct L2 pronunciations of unfamiliar written words; (ii) explicit instruction in particular grapheme phoneme correspondences can lead to successful retention of
Language Learning Journal 9 this knowledge by learners; (iii) in the absence of explicit instruction, learners may formulate their own, possibly inaccurate L2 decoding rules. A2. Appeal to declarative knowledge of general L2 patterns This category is distinguished from the previous one by the fact that the declarative knowledge relates not to specific grapheme phoneme correspondences, but rather to more general patterns in French spelling, pronunciation or symbol/sound mappings, as in examples 8 10 respectively: 8. 5gicle e4: cos like double E we use that a lot in French words don t we 9. 5te moin4: in French they sort of say /OI/ a lot in their words 10. 5goinfrez4: most words I ve heard with the O and I in it it s /OI/ The origin of this knowledge is again unclear, although the comments above suggest that the participants have noticed these patterns themselves through contact with the language. Again, however, the knowledge is sometimes incorrect, as in examples 9 and 10 relating to the pronunciation of 5oi4. This is despite the fact that, by the end of Y7, learners are likely to have encountered several high-frequency words and phrases containing this grapheme (e.g. moi, toi, choisis la bonne re ponse). This provides support for the view (section 1) that contact with the L2 alone may be insufficient for learners to acquire accurate knowledge (either implicit or explicit) of its symbol/sound correspondences. A3. Use the analogy strategy cluster The analogy strategy (see Woore 2007) involves using the pronunciation of a known word to support the pronunciation of an unknown word with which it has one or more graphemes in common. Comprising several mental actions, it is actually a cluster of individual strategies (Macaro 2006, 327) which must operate sequentially: participants (i) identify a particular written form as difficult to decode; (ii) identify a known French word (the source word ) containing the same written form, and recall the word s pronunciation; (iii) isolate the pronunciation of the target written form within the pronunciation of the source word as a whole; (iv) import this isolated phonological form into the target word, integrating it into the other decoded graphemes in that word. Each of these stages has potential pitfalls which may introduce inaccuracy into the final pronunciation of the target word (see examples below). Nine of the 12 participants made at least one comment which appeared to indicate use of the analogy strategy (24 comments in total). Seven participants made comments which appeared to indicate using the analogy strategy to support the pronunciation of individual graphemes, as in examples 11 13. They appear to have applied this strategy to L2 decoding spontaneously (their French teachers said that they had not taught them to use it), although further investigation is needed in order to find out whether they had learnt it via L1 literacy instruction, or whether it was something they had developed spontaneously (see Moustafa 1995; Goswami 1995): 11. 5gueulez4: cos I remember blue /blø/, B L E U cos that s got an E and U in it and it s /Y/
10 R. Woore 12. 5marteaux4: cos that X, I know that s a silent letter cos in Bordeaux it has a silent letter in it at the end 13. 5caleçon4: when you say français the C on that has got one under it as well Five participants also reported using the analogy strategy with larger written units. Several use the familiar word muse e to support their decoding of the target word museau (example 14), whilst others use familiar words such as qui, non and moins (known from time phrases like huit heures moins le quart) to help with the pronunciation of individual syllables in the words quignon (15 16) and te moin (17): 14. 5museau4: It looks a bit like the word /myzi:/ so... but it s got a U on the end and so it goes /myzi:y/ so like /myzjy:/ 15. 5quignon4: well I know what erm /n`n/ means (...) no doesn t it? 16. 5quignon4: yeah and then I know /kwi/ that means erm I ve forgotten it, like /kwis@pel/ 17. 5te moin4: I know the /moin/... yeah erm it s erm like eight /@:z mwa l@ ka:/ (¼ heures moins le quart ) Examples 11 17 are interesting for the different degrees of success with which the analogy strategy cluster is used. Whereas in 12 and 13 it results in correct pronunciations, in the other examples it is derailed by various problems. In 15 and 16, for example, participants stored phonological representations of the known words non and qui s appelle contain inaccuracies, which are then spread to the target words. This highlights the need for learners to have accurate pronunciations of source words used in the analogy cluster. However, even where this is the case, the sounds of the source words may not be transferred successfully to the target word. For example, the participant in 11 pronounces bleu accurately, yet seems unable to isolate the sound /ø/ successfully. In 17, the participant recalls the pronunciation of the word moins as /mwa/ (a relatively near miss for the nasalized /mw{~/), yet still pronounces this syllable as /moin/ when she transfers it into the new word context, and this form remains in the pronunciation she finally settles on (/temoin/). A4. Make explicit comparisons with English decoding Half the respondents (in 13 separate comments) explicitly compared the French and English decoding systems. In some cases, they highlighted the differences between the systems, as in 18. It is unclear here what the participant has based his French pronunciation on; this may illustrate the dilemma of learners who know that French and English spelling/sound correspondences differ, but do not know exactly in what ways: 18. 5goinfrez4: so/goin/ would sort of be more English I think, so /gwin/... /go~fwoiz/... cos /goinfòez/ sounds a bit English More often, however, participants comments seem to indicate the conscious use of English decoding to provide a basic phonological platform, to which specific modifications are then made. These modifications may reflect knowledge of particular French symbol/sound correspondences (examples 19 20), or they may
Language Learning Journal 11 be based on a vaguer notion of what sounds French (21). This last example recalls Erler s respondent, cited above, who reported giving L2 words a Frenchy touch : 19. 5caleçon4: well /keilk`n/ in English but then I see the C with the kind of... it s got that accent, that s /sei/ cos of français so /kalsei/ /kals`n/ yeah /kals`n/ I d probably say 20. 5gicle e4: er well the G I and the C and the L and the E will probably sound well apart from the accent E would probably sound mostly as it would in English 21. 5gicle e4: I just looked at it and thought kind of the English way you d say it but and then /gislei/... I kind of added in the French kind of thing Example 19 is particularly interesting because it shows how a learner s final pronunciation may be the result of a multi-stage process involving the progressive modification of an initial, English-based, decoded form. The participant evaluates and modifies the output at each stage, before reaching a pronunciation she considers acceptable. B. Break the target word up into smaller units This group contains all instances providing evidence that a participant has decided to focus on a specific part of a word s written form (one or more graphemes; one syllable). There is some overlap between this group and the previous one: for example, many strategies in categories A1 (appeal to declarative knowledge of particular symbol/sound mappings) and A3 (use the analogy strategy cluster) also involve consciously focusing on individual syllables or graphemes. Eleven participants (31 comments in total) explicitly describe this strategy of splitting the target word into smaller units, as in examples 22 24: 22. 5goinfrez4: I split it up into two words so /goin/ and then /fòez/ just put all the [inaudible] together 23. 5marteaux4: split it up between the R and the T so that s /ma:/ then the rest of it s like / tøw /. 24. 5quignon4: you would think that that N O N would be a separate like syllable itself so you d split it up between the G and the N so that d be like /kwig/ and the other [inaudible] would be like /n`n/. As in these examples, when participants explicitly describe splitting words up, it is almost always into syllables rather than into smaller units such as graphemes. Occasionally, however, participants segmentation of words does not follow French syllabification rules. In 24, for example, the French word is actually composed of the syllables /ki/ (5qui4) and/fio~/ (5gnon4). This problem in turn arises from an incorrect segmentation of the L2 grapheme 5gn4, which in French but not in English often constitutes a single grapheme. Having broken words up into smaller parts, eight participants then focused on these segments in a different order from that in which they appear when moving left to right through the word. For example, the participant in 25 scans the word for familiar graphemes; and the participant in 26 concentrates first on the vowels, which he considers particularly difficult. This non-linear processing seems to indicate conscious, strategic reasoning.
12 R. Woore 25. 5gicle e4: I first go round then I see if there s like any different bits about it that I notice and like know 26. 5gicle e4: [I m] deciding what the vowels sound like cos (...) they sound different sometimes in different words There are many other instances in the verbal protocols where participants focus on particular word segments, but without commenting explicitly that they are doing so. These instances can be grouped into two categories, depending on whether they focus on syllables (category B1) or graphemes (B2). B1. Focus on individual syllables There were 80 instances where participants focused on the pronunciation of individual syllables, as illustrated in 27 28. This strategy was used by all participants except two (one high achiever and one low achiever ). 27. 5houblon4: right so the H and the O and the U makes /hy/ 28. 5quignon4: then the second bit, the Q the U and the I, I sort of know, so that makes a /kxwi/ B2. Focus on individual graphemes Sixty instances were recorded where participants focused on the pronunciation of individual graphemes. Two-thirds of these were vowels. Seven participants focused explicitly on vowel graphemes (29 30), nine on consonant graphemes (31 32). Many of these latter instances related either to the grapheme 5ç4 which is understandable, because the cedilla does not exist in English and so marks the letter out as different or to silent consonants with no phonemic realisation. 29. 5gicle e4: I looked at the I... I is pronounced is pronounced /i/ 30. 5museau4: then it s that E the A and the U they make a like /øw/ sound 31. 5caleçon4: well the C has an accent I think erm so that would make it a /si:/ sound 32. 5goinfrez4: the Z is a silent letter and you would like wouldn t say the Z Having focused on the pronunciations of individual word segments, these sounds must then be integrated with the pronunciation of the rest of the word to give the overall pronunciation. Five participants make one or two explicit references each to this process, but two participants do so several times, describing it in terms of a mathematical process of addition (33 34). Here, the process of L2 decoding is clearly conceptualised as a conscious, problem-solving activity. 33. 5te moin4: the T, at the start, with the E with the little thing at the top, makes /te/ and then O I N is /we~/, so then you add the M in so then it s /temwe~/... and then you add it all together, so it s like a sum 34. 5gicle e4: that d be /i:/ cos the /e/ plus the other one makes an /i:/ sound, and then add the L and the C, makes /kli:/, /kli /, and then you add the G and the I and then you add it all up Interestingly, however, there is sometimes a mismatch between the pronunciation of a grapheme when it is focused on individually, and its pronunciation as part of the final reading of the word as a whole. This occurs in the verbal protocols of eight participants. The participant in 30, for example, says that the E the A and
Language Learning Journal 13 the U... make a like /øw/ sound, but finally pronounces museau as /mu:si:/. Similarly, another participant says that the 5e 4 in te moin makes an /e/ sound, but later pronounces the word as /timoi/. In all cases, participants seem unaware of the mismatching pronunciations they have generated for the same grapheme. Perhaps the sound which has been worked out via conscious reasoning is overridden by more automatic, L1-based processing when the word is read as a whole unit. Alternatively, the phenomenon may reflect limitations imposed by working memory, making it impossible to hold the relevant phonological form securely whilst other processing takes place. It should be acknowledged that the burden on participants working memory may be exacerbated by the verbal selfreporting. C. (Sub-)vocalise possible pronunciations of part or all of the target word Prior to settling on the final pronunciation of a target word, 10 out of 12 participants were found to produce intermediate vocalisations of the whole word or parts of the word. As discussed above in relation to example 19, this appears to be part of a monitoring process where learners evaluate their proposed pronunciation of a letter string to see whether it sounds right. This may also occur silently: three participants describe this explicitly, saying that they try and pronounce it in [their] head and then (...) say it out loud. Thus, during the discussions, faint whispering or lip movements were often detectable. However, such instances were not recorded. Over half of the participants who tried pronunciations out loud were found to say letter strings repeatedly within the same speech turn, with the same pronunciation (35 36). This suggests that they were finding it hard to decide whether their pronunciation was acceptable, reflecting in turn their lack of a secure basis for doing so: 35. 5goinfrez4: I think it s /fòe/ /e/ /ez/, /fòez/ 36. 5te moin4: /temoin, temoin/ In a related process, nine participants made repeated pronunciations of all or part of a word, but modified the pronunciation slightly each time (37 38). Often, these variant forms were embedded within an explicit commentary about the fact that different pronunciations were being weighed up (39 41): 37. 5caleçon4: /kiljeso~/, /kaleso~/ 38. 5goinfrez 4: or from the F as well, like /fòez/ like /ga~fòez/, /dza~fòez/ 39. 5houblon4: erm again I m thinking shall I call it /h@u/ or /hjyw/ I ll probably say I ll go with /hjyw/ 40. 5gicle e4: so it s /glisleji:/ it could be /glislei/ and it could be /glisli:/, /leji:/ 41. 5gueulez4: /gjywijyw/ erm I don t think that s how it s said erm /gjyw/ I think I d say, /gjyw/ /gjylez/ Conclusions Erler (2003) found that, when faced with L2 French decoding tasks, beginner learners in Y7 based their pronunciations overwhelmingly on L1 (English) symbol/ sound mappings. Many pronunciations produced by participants in the current
14 R. Woore study, too, sounded intuitively like the output of English-based decoding (e.g. /goinfòez/ for goinfrez; /kwign`n/ for quignon; /temoin/ for te moin). However, there were also many examples of participants trying to move beyond L1-based processing and to pronounce words instead according to the L2 decoding system. They employed various strategies to help them do this. There was a high degree of consistency in the strategies used by the different participants, with most being used by over half the sample. Almost all explicitly reported that they broke target words up into syllables to facilitate decoding. There were also many instances of learners focusing their attention on individual graphemes (usually vowels) which they considered particularly difficult, rather than simply processing the word in left-to-right order. Often, one or two identifiably French elements of a word (e.g. letters with diacritics like 5c 4 or 5e 4) were selected for focused attention, whilst the rest of the word (most consonants and some simple vowel graphemes) was apparently considered unproblematic and decoded with little conscious thought. The pronunciations that had been worked out for the difficult graphemes were then slotted into the framework of relatively more secure (English-like) grapheme phoneme correspondences. However, the parts of words which participants considered unproblematic were often mispronounced nonetheless. This occurred when the French and English realisations of particular graphemes differed, even though the participants thought that they did not. Previous findings of poor French decoding proficiency amongst this population are therefore supported. Having singled out particular graphemes for conscious focus, participants were found to draw upon a range of further strategies to generate what they considered an L2 pronunciation. The most common strategy (used by all participants) was to appeal to declarative knowledge of particular symbol/sound mappings in French. Some participants also referred to knowledge of more general patterns in French spelling, pronunciation or decoding. However, this declarative knowledge whether the result of previous explicit instruction or of implicit learning was not always sound. Sometimes, it seemed to be based on English symbol/sound mappings or on inaccurate interpretations of the linguistic input. Three quarters of participants also used the analogy cluster, a sequence of strategies which involves using a particular written form (e.g. syllable or grapheme) within a known word to derive the pronunciation of the same written form in an unknown word. For example, participants used their knowledge of the word français to work out the pronunciation of the 5c 4 in caleçon. Whilst this cluster sometimes succeeds, a number of pitfalls also threaten to yield an incorrect outcome: for example, participants sometimes hold inaccurate phonological representations of the source word, are unable to segment it into its constituent phonemes successfully, or are unable to hold the segmented phoneme in working memory until it is transferred into the target word. Having worked out a pronunciation for a word or word-part, almost all participants were found to say these phonological forms aloud, apparently in order to evaluate them according to whether they sound right or sound French. There is also evidence that this evaluative strategy is more widespread than may appear in the verbal protocols, because participants may carry it out silently using their inner voice. However, participants often lack sufficient knowledge of correct L2 pronunciations in order to evaluate their own attempts successfully.
Language Learning Journal 15 To summarise, in contrast to the automatic, rapid processing of letter strings characteristic of fluent L1 reading, participants L2 decoding was often slow, effortful and conscious, resembling a problem-solving task. In some cases, this process was explicitly driven by knowledge that French decoding is different from English decoding. However, beyond this basic awareness of difference, participants lacked specific knowledge about what the French pronunciations of words should be. Further, when they tried to evaluate the pronunciations they generated, they did not know whether or not they were right. They can be compared to people struggling to solve a logic puzzle where the information provided is insufficient to complete the task, and where the criteria against which to measure successful completion of the puzzle are lacking. The interview-discussions in the current project constitute a somewhat artificial situation; the strategies participants reported using may not reflect what they usually do when reading French words. However, this does not necessarily undermine the value of the study. It was argued above that the first steps towards L2 decoding proficiency may be taken when learners become aware of the need to consciously modify automatic L1 decoding processes. For example, as described above, Erler s (2003) better decoders seemed to be those who engaged in strategic reasoning about their L2 decoding, rather than simply accepting L1-based pronunciations. The interview-discussions may represent a situation in which learners are more likely to make this kind of effort to produce a better L2 pronunciation (because of the presence of the researcher and the explicit focus on L2 decoding). We therefore gain insight into the strategies used by these participants when they are really trying to pronounce French words in the best way they can. The challenge then lies in motivating learners to put similar effort into L2 decoding in more routine settings (in class or at home). The current study also highlights the fact that, if learners are to use strategic reasoning to support their developing L2 decoding, then they need a secure and accurate knowledge base to underpin it. Participants in the current study often lacked this. One response to this problem might be to advocate explicit, systematic instruction in L2 decoding as part of MFL lessons (Siddons 2001; Erler 2004). However, we still lack evidence concerning the best approach to teaching L2 decoding (indeed concerning the possibility that teaching L2 decoding explicitly can succeed). The current study may help inform pilot programmes of instruction in French decoding, by giving some insight into strategies which learners already use, and which could be developed or built upon. For example, explicit knowledge of particular symbol/sound correspondences is widely used by participants in the current study, and clearly can help them decode words more accurately; but a challenge for teachers is to ensure that this knowledge is reliable and accurate. Similarly, there is evidence that the analogy cluster, already being spontaneously applied to L2 decoding by many participants, can provide an effective route to accurate pronunciations; however, teachers may need to provide instruction to ensure the strategy cluster is used effectively and that all learners have access to its benefits. Meanwhile, an emphasis on L2 decoding skills has been included in the KS2 Framework for languages teaching (DfES 2005), and the corresponding Schemes of Work for French (QCA 2007) contain a Phonics Focus in each unit, with suggested activities for teaching particular grapheme phoneme correspondences. It will be interesting to see what effects, if any, this explicit emphasis on decoding has on young linguists.
16 R. Woore Limitations Clearly, the limited sample in the current study prevents generalisation to the wider population, although there is no reason to believe that this sample is especially dissimilar to many other MFL learners in KS3. Further research is required to see whether the strategies described here are characteristic of learners in this context. Larger-scale research would also benefit from the use of an inter-rater to validate the analysis and classification of the data; this was not considered necessary in the current study given its exploratory nature. Finally, the inevitable limitations of the self-report procedure must be borne in mind when interpreting findings. It is possible that participants behaviour was influenced by the process of reflecting and reporting on it. However, the high degree of consistency found in the strategies reported by the 12 participants in the current study suggests that the self-reports do provide some insight into what is going on inside the learners minds. Acknowledgment I am grateful to Professor Ernesto Macaro for his helpful comments on an earlier draft of this article. Notes 1. The term is therefore used more narrowly than in some of the literature on reading, where it may also refer to the pronouncing or recognising of known words. 2. Throughout this article, phonemic representations are designated by forward slashes (/... /), written representations by angled brackets (5...4). Sounds are transcribed using the phonetic alphabet of the International Phonetic Association (1999). Capital letters indicate the use of letter names (e.g. H ¼ aitch, Z ¼ zed). References CILT (The National Centre for Languages), ALL (Association for Language Learning), ISMLA (Independent Schools Modern Languages Association). 2008. Language Trends Secondary Survey 2008: Data Report. www.cilt.org.uk/research/languagetrends/2008/ Language%20Trends%202008%20statistical%20report.pdf (accessed April 23, 2009). DfES (Department for Education and Skills). 2003. Key Stage 3 National Strategy. Framework for teaching modern foreign languages: Years 7, 8 and 9. London: DfES. DfES (Department for Education and Skills). 2005. The Key Stage 2 Framework for Languages. www.teachernet.gov.uk/publications (search reference 1721-2005DCD-EN) (accessed August 6, 2008). Ehri, L.C. 2005. Development of sight word reading: phases and findings. In The Science of Reading. A Handbook, ed. Margaret J. Snowling, and Charles Hulme, 135 54. Oxford, UK: Blackwell. Erler, L. 2003. Reading in a foreign language near-beginner adolescents experiences of French in English secondary schools. Unpublished D.Phil Thesis, Oxford University Department of Educational Studies. Erler, L. 2004. Near-beginner learners of French are reading at a disability level. Francophonie 30: 9 15. Erler, L. 2008. Rhyme identification at Key Stage 3 the results. Francophonie 37: 3 7. Goswami, U. 1995. Phonological development and reading by analogy: what is analogy, and what is not? Journal of Research in Reading 18, no. 2: 139 45. Hulstijn, J.H. 2005. Theoretical and empirical issues in the study of implicit and explicit second-language learning. Studies in Second Language Acquisition 27, no. 2: 129 40. International Phonetic Association. 1999. Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet. Cambridge, UK: Cambridge University Press.
Language Learning Journal 17 Klapper, J. 2003. Taking communication to task? A critical review of recent trends in language teaching. The Language Learning Journal 27, no. 1: 33 42. Kormos, J. 1998. Verbal reports in L2 speech production research. TESOL Quarterly 32, no. 2: 353 8. Macaro, E. 2001. Learner Strategies in Second and Foreign Language Classrooms. London: Continuum. Macaro, E. 2006. Strategies for language learning and for language use: revising the theoretical framework. Modern Language Journal 90, no. 3: 320 37. Macaro, E., and L. Erler. 2008. Basic literacy skills in L2 French and language learning motivation. Paper presented at AILA (Association Internationale de Linguistique Applique e) 15th World Congress of Applied Linguistics, August 24 29 in Essen, in Germany, Multilingualism: Challenges and Opportunities. Moustafa, M. 1995. Children s productive phonological recoding. Reading Research Quarterly 30, no. 3: 464 76. QCA (Qualifications and Curriculum Alliance). 2007. New Key Stage 2 Scheme of Work for French. http://www.qca.org.uk/qca_11752.aspx. Seliger, H.W. 1983. The language learner as linguist: of metaphors and realities. Applied Linguistics 4, no. 3: 179 91. Siddons, L. 2001. Practical reflections on the sound-spelling link. Francophonie 23: 10 4. Woore, R. 2006. Investigating beginner learners decoding of second language French: methodological issues and some preliminary substantive findings. Unpublished MSc dissertation, Oxford University Department of Educational Studies. Woore, R. 2007. Weisse Maus in meinem Haus : using poems and learner strategies to help learners decode the sounds of the L2. The Language Learning Journal 35, no. 2: 175 88. Woore, R. 2009. Beginners progress in decoding L2 French: some longitudinal evidence from English modern foreign languages classrooms. The Language Learning Journal 37, no. 1: 3 18. Appendix: Words selected for use in the self-report discussions te moin gicle e goinfrez gueulez caleçon museau houblon quignon Tre mie marteaux