Volume 2 (2011) Issue 2
pp. 267 - 285
Theory and Pedagogy of Reading
Phonological Recoding for L2 Reading
Lance R. Askildson (Notre Dame (IN), USA)
Abstract
The role of phonology in L2 reading was investigated via the Reading While Listening (RWL) technique for L2 reading instruction. The present study tested the reading gain of 43 L2 readers of French before and after a RWL treatment was administered and compared this gain with the performance of a control silent reading group consisting of an additional 34 L2 readers of French. The results support a significant gain for the RWL treatment group with a corresponding insignificant gain for the control group. Additionally, the difference in gain between the experimental and control groups was statistically significant. Pedagogical implications are discussed alongside recommendations for future research.
1 Introduction
The role of phonological access in reading has long been a salient focal point for linguistic research of lexical access and retention. Psycholinguistic insights into the reading process have provided a series of complex models wherein phonology clearly plays an integral part for both lexical retrieval and summative comprehension. The implications of a fundamental phonological component to reading—particularly among reading tasks requiring significant attentional commitment—are considerable for both L1 and L2 reading development. While L1 reading researchers and pedagogues have capitalized on the phonological component(s) of reading via phonics-based instructional approaches since the 1960’s, only very recently has phonology been considered as a potentially important component to L2 reading development. Recent investigations into phonological training via an audio accompaniment during L2 reading tasks (Reading While Listening or RWL) as a means to improve both reading rate and comprehension have produced results supportive of such phonological processing during L2 reading. Nonetheless, the longitudinal designs and classroom environments of these recent studies suggest the possibility of confounding variables affecting participants’ reading development and limit the interpretability of the results. Moreover, these studies fail to isolate the mechanism responsible for their reported gains in reading. The current pilot study is intended to provide a preliminary evaluation of the effectiveness of RWL within a more controlled experimental design that isolates the decoding of individual lexical items as evidence for the phonological training mechanism responsible for reading improvement. Specifically, the current research hypothesizes that RWL has an effect on the decodability and retention of individual lexical items and that this will be evidenced by an improvement in reading ability as measured by improved reading rate and comprehension of target lexical items.
2 Phonological Processing in Lexical Access
Contemporary linguistic scholarship has focused considerable attention upon cognitive processing models in order to elucidate the essential mechanics of reading. The processing route for lexical access during reading comprehension has been of particular interest. Traditional accounts often placed this processing burden wholly upon visual decoding of orthography and subsequent direct access to lexical semantics (Carver, 1990). As a result, a 1:1 model proposing a direct path from orthography to meaning has characterized much of previous research foci within linguistics. While such a 1:1 model of reading comprehension continues to inform some reading research and pedagogy, more recent investigations into reading processes have suggested an alternative, if not complimentary, model for lexical access. Specifically, many contemporary researchers have proposed a model of lexical access in reading that stipulates phonological mediation between the traditional route from orthography to lexical semantics. This proposal of phonological decoding of orthography before access to meaning has been most clearly and ardently articulated by Van Orde, Johnston, & Hale (1988) and is often identified as the ‘symbol to sound to sense’ model of lexical access.
Van Orde et al. (1988, 1997) have suggested that phonological mediation of orthography and semantics is the primary and perhaps only route for cognitive processing of reading. Their own claim is based on experiments investigating semantic categorization among target and non-target homophonic words—wherein a higher error rate was demonstrated for non-word homophones. Thus, when presented with a category like Fruits, participants were significantly more likely to produce false positive responses when shown a non-word homophonic distracter as in Pare (instead of Pear) than in trials that displayed a non-word distracter without phonological parallelism (such as Pork) to a member of that same category. In a similar manner, Lukatela & Turvey (1994) also found evidence for phonological processing via a homophonic priming task. In their own trials, they reported that participants showed a priming effect for a particular target word (i.e. orange) when presented with a homophonic non-word (i.e. apel) that paralleled the phonology of a word (i.e. apple) semantically similar to the target. These results thus strongly suggest a phonological processing route for lexical access. This particular phenomenon of phonological mediation and decoding of written language has led to the characterization of reading processes as an act of ‘silent speech’ or subvocalization.
Numerous additional studies have confirmed the significant role of subvocalization in reading using a diverse array of research methodologies and experimental foci. In an illuminating study of speech disorder by Bosshardt (1990), child and adult stutterers were compared to non-stutterers during two treatments of oral and silent reading. Not surprisingly, the production rate during oral reading among the stuttering group was considerably slower than the non-stuttering participants. Significantly, however, the stuttering group was also correlatively much slower during the silent reading treatment. Employing a different methodological approach, Liva & Pugh (1995) adapted a vowel-canceling technique for French speakers in order to investigate whether attention to phonology slows reading rate and/or comprehension. In one particularly telling treatment, participants were instructed to cross-out all instances of ‘a’ vowels pronounced as [æ] (as in valise; but not as in beau or aimer) during a silent reading activity. After the administration of comprehension tests, results showed that participants in the vowel-canceling treatment performed at nearly an identical rate and level of comprehension in their reading when compared with those in control groups—providing considerable support for subvocalization in reading. Additional advocation for phonological mediation and subvocalization comes from Abramson & Goldinger (1997). These researchers found that lexical decision tasks for monosyllabic words with lengths that were orthographically identical but phonologically different produced response times that varied in accordance with phonological length and not orthographic length. Thus, when presented with words like boar and tick in separate lexical decision tasks, and then asked to indicate in each case whether the word was inclusive of a particular category (i.e. insects), participants demonstrated a longer response time in the case of boar (a phonologically longer item) when compared to the response time for tick. These, and many other research findings (Jones, 1986; Liva, Fijalkow, & Fijalkow, 1994; De Guerro, 2004; et cetera), cast a considerable shadow over 1:1 visual-orthographic mapping accounts of reading, whilst also providing a convincing case for an alternative phonological processing route.
Despite the strong arguments and evidence supporting a phonological route for lexical access however, additionally persuasive findings have been used to advocate a compromising model describing a dual and parallel orthographic-phonological route for lexical access. Coltheart, Curtis, Atkins, & Haller (1993) and Coltheart & Coltheart (1997) have played prominent roles in the proposal for such a dual processing route in reading. In reaching this conclusion, these, as well as other researchers (Baddely, Eldridge, & Lewis, 1981; McGuigan, 1984; et cetera), have pointed to evidence of subvocalization in conjunction with apparently contradictory evidence of orthographic lexical processing. A cogent example of such evidence against an exclusively phonologically route to lexical access is provided by the reading abilities of two neurologically impaired individuals as described by Shelton & Weinrich (1997) and Hanley & McDonnell (1997) respectively. Both of these research participants displayed a unique disability that prevented them from deriving phonology from orthography. While reading was accomplished with normal comprehension and little difficulty, both subjects demonstrated, via a battery of experimental trials and tests, a near total lack of ability to decode non-words and non-word homophones (i.e. phocks in lieu of fox) in either explicit speech or silent reading. That the research participants were unable to access phonology for written non-words or non-word homophones demonstrates that phonological decoding was not requisite for either individual to access the lexicon during reading. These findings thus stand as a stark contradiction to any phonology-only processing model for lexical access. Moreover, one need only consider the pervasive ability of the congenitally deaf to read at rates near or identical to hearing individuals (Jensema & Burch, 1999), despite a lack of any significant phonological representation, in order to offer further evidence against an exclusively phonological route to meaning during the reading process. Nonetheless, the existence of alternative routes or strategies for lexical access does not negate the potential importance of phonology for the majority of readers who do indeed have access to such representations. In fact, dual route parallel processing does little to change phonological mediation models except to suggest the existence of alternative strategies among the minority of readers who lack access to phonology. Moreover, while reading rate has demonstrated equivalence between orthography-only and phonologically-aided readers, comprehension ability has not been sufficiently addressed among those lacking access to phonology.
In an additional complexification of phonological reading models, an alternative to both phonological mediation and dual route phono-visual models has been proposed in the form of post hoc phonological recoding. Within this model, phonology is accessed after visual orthography has been processed. The phonological information is then used to both retrieve meaning from long-term memory and hold onto meaning within working memory via the phonological loop (Baddeley, 1986; 1996) for later summative comprehension. This model has important implications for reading development. Indeed, it serves as the core mechanism within the self-teaching hypothesis (Jorm & Share 1983, 1987) of reading development. According to the hypothesis, decoding of orthography during the reading process is complimented by phonological decoding and subsequent phonemic awareness. Specifically, the hypothesis claims that reading develops as learners begin to match orthographic combinations to phonological sounds in order to access long-term memory and meaning. As learners are exposed to orthographic patterns repeatedly, their phonological recoding becomes more automatic and their improved reading speed and reduced processing burden results in better comprehension. According to Share (1995) and Kyte (2004), learners need only be exposed to a orthographic pattern “a few times” in order to automatize their phonological recoding and semantic retrieval. The self-teaching hypothesis via phonological recoding offers a potential mechanism for the RWL technique that would coincide well with its apparent success in developing reading rate and comprehension. Nonetheless, dual route parallel processing and phonological mediation would also appear to support the RWL technique in a similar manner of successive exposure to phonology and subsequent improvement in lexical retrieval. For the purposes of this study, all three models of lexical access and retention will be considered as potential mechanisms underlying the RWL technique.
3 Reading-While-Listening as a Pedagogical Tool
Pedagogical applications of phonology during lexical access have, perhaps surprisingly, been employed in the development of L1 reading fluency for quite some time (Van Bon, Boksebeld, Freide, & Van den Hurk, 1991; Jones, 1986; Kennell, 2002). In their instructional forms, the application of such techniques—despite a certain degree of diversity in manner and context of use—are all essentially designed to develop phonemic awareness and speed of phonological processing via listening to overt oral decoding [i.e. pronunciation] of a text’s orthographic constituents alongside concurrent and silent reading of the same text. This general technique is most often labeled Reading While Listening (RWL) (McMahon, 1983) and, at the very least, includes the essential characteristics as described above—though, as Van Bon et al. (1991) outline in their own study of RWL, a variety of distinct variations on this fundamental principle are also employed. Additionally, most curricular applications of RWL rely on recorded listening components to train phonological decoding over longer periods of time and with more clarity and consistency of presentation. This has often taken the form of prerecorded audio books played in conjunction with silent reading of the paper-based text. Medwell (1998), for example, reported significant gains in child L1 reading development during a ‘talking books’ project that employed RWL in a computer-based format. Evaluating RWL from a different vantage point, Van Bon et al. (1991) found RWL was not only effective at improving reading fluency among texts similar to those used in training subvocalization, but that this improvement also transferred to other genres of text. In another testament to RWL effectiveness, Reid (1971) found that students trained with an RWL treatment performed significantly better in levels of reading rate and comprehension. Additional and significant evidence for the pedagogical effectiveness of RWL in L1 reading development comes from Jones (1986) and his exhaustive review of experimental literature attesting to the effectiveness and efficiency of RWL.
While phonological approaches to L1 reading development have been common practice for more than 30 years, such techniques have received very little attention among researchers and instructors of L2 reading and RWL has only recently been applied to the development of L2 reading. In one early study of RWL in an L2 context, Amer (1997) found that EFL students given RWL treatments significantly outperformed their control group counterparts who were given only silent reading treatments in series of classroom treatments. In another example of RWL applied to L2 readers, Kennell (2002) studied the effects of RWL on a small group of ESL learners over the course of 8 weeks. When the treatment group was compared to a silent reading control group, significantly increased reading rates and comprehension gains were reported in the case of the former while only minimal and statistically insignificant improvements were made by the control group. These two studies certainly provide supportive evidence for RWL as an effective technique for improving L2 reading ability. However, their longitudinal designs and classroom settings provide the potential for confounding influence of other variables on the reading development of the test subjects. Moreover, and much more importantly, neither of these studies, alongside the vast majority of research investigating L1 RWL techniques, proposes nor isolates the underlying mechanism resulting in the apparent improvement in reading rate and comprehension of participants subjected to the phonological treatment. The present study is intended, therefore, to address these deficiencies in order to provide a more complete and accurate description of the RWL technique and the mechanism by which it improves L2 reading or a lack thereof.
4 Participants
A total of 77 freshman college students between the ages of 18 and 20 (M = 18.3) participated in the present study. All participants were enrolled in introductory French coursework, and course sections were divided into control (N=34) and experimental groups (N=43), respectively. All potential participants were given a biographical questionnaire that ensured a uniform L1 of English and no prior formal study of French; however, other formal [Romance] language study was not included as a control.
5 Materials
Two sets of five two-to-three syllable non-cognate lexical items (ten in toto) were chosen from the vocabulary list of a previously completed chapter within the Introductory French course textbook (see Appendix A). All items chosen were nouns presented in the singular form. Course instructors were consulted on the appropriateness of the chosen items, and great care was taken not to choose related or categorized items. These two sets of five vocabulary items each were then incorporated into one and four paragraphs, respectively. Thus, the first set of five words was incorporated into Reading 1, while the second set of five items was incorporated into Readings Readings
The paragraphs (referred to as Readings Reading Reading 1, while lexical set 2 was incorporated into Reading Reading 1 and Reading Readings
The treatment itself was rather straightforward. The control group simply read Readings Readings
6 Procedure
Participants were tested during their regular French class time and in their regular classroom. They were told that the study was investigating reading development, but no details were given regarding the treatment or materials themselves. Participants were administered Reading 1 in the form of a folded 8.5 x 11 paper with the blank side facing up. Each of the readings was prepared and displayed in a manner to prevent participants from starting the reading task too early or from continuing reading beyond the allotted time for the task. Thus, in the case of Readings Reading Reading
7 Results
A correlated groups t-test compared the mean number of correct comprehension items on the pretest (Reading 1) to the mean number of correct comprehension items on the posttest (Reading 5) for the experimental group. This test was found to be statistically significant at an alpha level of .05, t (42) = 4.234, p < .05, indicating that the RWL treatment had a positive effect on reading ability as measured by the target lexical items among participants in the experimental group. Specifically, the experimental group saw a mean gain of 1.45 (SD = 1.0) more questions correct on the posttest than on the pretest.
A second correlated groups t-test compared the mean number of correct comprehension items on the pretest (Reading 1) to the mean number of correct comprehension items on the posttest (Reading 5) for the control group. This test was found to be statistically insignificant at an alpha level of .05, t (33) = 1.439, p < .05, indicating that the control treatment of silent reading only had a negligible and insignificant positive effect on reading ability as measured by the target lexical items among participants in the control group. Specifically, the control group saw a gain of only 0.53 (SD = 2.03) more questions correct on the posttest than on the pretest.
Finally, the mean difference scores for the experimental group were compared to the mean difference scores for the control group. This test was found to be statistically significant at an alpha level of .05, t (75) = 2.61, p < .05, indicating that the scores by the control and experimental groups were significantly different from one another. Specifically, the mean difference between the control and experimental group was 0.92.
An analysis was also conducted on the reading rates of both the experimental and control groups as measured by the circled word on Reading
The results of the post-reading vocabulary test indicated that all but two of the 77 participants knew the meanings of the target lexical items. In the case of both of these two individuals, only one of the ten vocabulary items was unknown. Thus, their scores were deemed within the margin of error and included within the reported results.
8 Discussion and Conclusions
The results from the experimental group support the stated hypothesis of positive effect of the RWL technique on decodability and comprehension in reading as measured by the target lexical items. This would appear to further buttress previous research in support of RWL as an instructional technique for L2 reading. The results of the control group also support the hypothesis of this study by their lack of any statistically significant gain. Thus, these initial results seem to both confirm the hypothesized results of the present study as well as add further validity to previous studies supportive of RWL for L2 reading, but which failed to control for potentially confounding variables.
Moreover, as indicated in the statistical analysis above, the degree and consistency of differential gains in comprehension between the scores of the experimental and control groups also indicated a statistically significant difference. Thus, the fact that the experimental group made a significant gain in comprehension and the control group failed to make a significant gain as hypothesized suggests that the comprehension effects of the experimental group were due to the RWL treatment and not any other potentially confounding variables that would be unlikely to result in statistically significant findings. This result is therefore supportive of the stated hypothesis and buttresses the claim of effectiveness of RWL on reading as measured by lexical items in the present study.
While these findings are supportive of the claims made in the present study and suggestive of a broader efficacy for RWL in L2 reading development, these findings require additional validation in light of the small subject pool and the brief training and testing period. Additionally, future research should also attempt to control potential confounding variables such as prior exposure to the language and/or lexical test items as well as the durability of facilitation effects for RWL on reading fluency. In the present study, the control group had a M =0.53 and an SD = 2.03. This is an extremely high error term for a scale from 1 to 5 points. Indeed, the degree of variability in the control group scores suggests that the control group may have experienced additional confounding variables that were not addressed within the present experimental design. Such variability was used to calculate the significance of difference between the means of the control group and experimental group and may therefore be responsible for an unreliable skew in the obtained statistical value. In order to reduce the variability among scores in future research, it would very likely be effective to both increase the number of participants and provide a longer series of treatments. The latter is particularly important as some of the participants in the current study may have failed to receive enough of the treatment in order to produce significant and consistent results.
Future research, therefore, must undoubtedly increase the length of treatment if more reliable results are to be obtained. Moreover, such research would further benefit from a more controlled experimental design in which reading rate and comprehension could be measured with more precision—though instruments for such precise measures are both expensive to produce/procure and constrain the content of experimental treatment texts. Finally, the issue of attention and motivation should be addressed. One potential threat to the internal validity of the present study is the attention and motivation with which the participants completed the experimental tasks. Simple observation has led the current researcher to conclude that there were several participants in both groups who appeared to focus little attention on the tasks requested of them. Moreover, the unforeseen departure of the class instructor during the control trial may have contributed to a reduction in motivation and a subsequent reduction in the attention paid to the tasks. Both of these issues could be more effectively addressed in a controlled environment in which participants were encouraged to focus on the task at hand. These modifications may allow future research to shed a more illuminating light on the role and effectiveness of RWL in L2 reading.
Appendices
Appendix A
Vocabulary List 1
une chaussure
les lunettes
un pantelon
la confiture
le jambon
Vocabulary List 2
une pâtisserie
une promenade
une bouteille de
le santé
le visage
Appendix B
Reading 1
Chaque matin, Robert mange une baguette avec de la confiture pour son petit déjeuner. Ensuite, il prépare quelques sandwichs au jambon pour son déjeuner à midi. Après sa douche, Robert met un pantalon très chic. Il choisi ses chaussures noires, brunes ou blanches avant de regarder les info à la télévision. Il a toujours des problèmes pour trouver ses lunettes de soleil avant de partir au travail.
Reading 2
Marie travaille dans une belle pâtisserie à côte d’un parc. C’est une belle femme de vingt-cinq ans avec un visage très joli. Tous les matins, elle fait une longue promenade dans le parc avec son petit ami. Pour rester en bonne forme, Marie fait du sport tous les jours. Comme distraction, elle aime aller au café et commander une bouteille de vin avec son petit ami.
Reading 3
Daniel est en mauvaise santé pour beaucoup de raisons. Premièrement, il boit souvent une ou deux bouteilles d’alcool chaque nuit. Ensuite, chaque matin, il fait sa petite promenade de quinze mètres à la pâtisserie pour acheter des bonbons pour son déjeuner et son dîner. Finalement, Daniel n’aime pas être actif. C’est pourquoi il est gros et que son visage n’est pas mince du tout.
Reading 4
Sandrine vend des cosmétiques et des bouteilles de parfum. Bien sûr, tous les matins elle se maquille le visage. Quand elle achète un croissant à la pâtisserie, tous les hommes la regardent la bouche ouverte. En plus, elle est jeune et en très bonne santé. Trois fois par jour, elle fait une grande promenade au centre ville pour vendre ses produits dans tous les magasins.
Reading 5
Maurice va au supermarché tous les jours pour acheter une bouteille d’eau. Chaque matin, il fait une longue promenade sur une plage de Normandie. Bien sûr, son visage est toujours rouge à cause du vent froid sur la plage. Après ses exercises, il prends son petit déjeuner à la pâtisserie locale. Maurice est en bonne santé parce qu’il mange bien et fait du sport souvent.
Appendix C
Comprehension Questions for Reading
1. What does Robert put in his sandwiches for lunch?
2. What does Robert put on his baguette for breakfast ?
3. What does Robert always have trouble finding before leaving for work?
4. What does Robert put on after his shower?
5. What does Robert do before watching the news?
Comprehension Questions for Reading
1. Where does Maurice have his breakfast?
2. What does Maurice buy at the supermarket?
3. What does Maurice do in Normandy
4. What is the result of Maurice’s good eating and exercise habits?
5. How does the cold wind on the beach affect Maurice?
References
Abramson, M. & Goldinger, S. (1997). What the reader’s eye tells the mind’s ear: Silent reading activates inner speech. Perception and Psychophysics, 59, 1059-1068.
Amer, A. A. (1997). The effect of teacher’s reading aloud on the reading comprehension of EFL students. ELT Journal, 51, 43-47.
Baddeley A., Eldridge, M. & Lewis, V. (1981). The role of subvocalization in reading. Quarterly Journal of Experimental Psychology, 33, 439-454.
Bosshardt, H. G. (1990). Subvocalization and reading rate differences between stuttering and non-stuttering children and adults. Speech and Hearing Research, 33, 776-785.
Carver, R. P. (1990). Reading Rate: A Review of Research and Theory. San Diego , CA
Coltheart, M., Curtis, B., Atkins, P., & Haller, M. (1993). Models of reading aloud: Dual-route and parallel-distributed processing approaches. Psychological Review, 100, 589.
Coltheart, M. & Coltheart, V. (1997). Reading comprehension is not exclusively reliant on phonological representation. Cognitive Neuropsychology, 14, 167-175.
De Guerro, M. C. M. (2004). Early stages of L2 inner speech development: what verbal reports suggest. International Journal of Applied Linguistics, 14, 90-114.
Hanley, J. R. & McDonnell, V. (1997). Are reading and spelling phonologically mediated? Evidence from a patient with a speech production impairment. Cognitive Neuropsychology, 14, 3-33.
Higgins, J. & Wallace, R. (1989). Hopalong: A computer reading pacer. System, 17, 389-399.
Jensema, C. J. & Burch, R. (1999). Caption speed and viewer comprehension of television programs. U.S.
Jones, M. R. (1986). The effects of training with time compressed speech plus text on reading speed, vocabulary, and comprehension. (Eric Document Reproduction Service No. ED 345 214).
Jorm, A.F., & Share, D.L. (1983). Phonological recoding and reading acquisition. Applied Psycholinguistics, 4, 103-147.
Kyte, C.S. (2004). The Role of Phonological Recoding in Reading Development. UMI Dissertations: DANQ84856.
Liva, A., Fijalkow, E. & Fijalkow, J. (1994). Learning to use inner speech for improving reading and writing of poor readers. European Journal of Psychology of Education, 9, 321-330.
Liva, A. & Pugh, A. K. (1995). Inner speech in young French children’s reading. Journal of Research in Reading
Lukatela, G. & Turvey, M. T. (1994). Visual lexical access is initially phonological: Evidence from phonological priming by homophones and pseudohomophones. Journal of Experimental Psychology: General, 123, 331-353.
McGuigan, F. J. (1973). The function of covert oral behavior (silent speech) during silent reading. Linguistics, 112, 39-47.
McGuigan, F. J. (1984). How is linguistic memory accessed? A psychophysiological approach. Journal of the Biological Sciences, 19, 119-136.
Medwell, J. (1998). The talking books project: some further insights into the use of talking books to develop reading. Reading
McMahon, M.L. (1983). Development of reading-while-listening skills in the primary grades. Reading Research Quarterly, 19, 38-52.
Palmer, S. Working Memory: A Developmental Study of Phonological Recoding. Memory, 8, 179-193.
Paivio, A. (1986). Mental Representations. New York : Oxford University
Share, D.L. (1995). Phonological Recoding and Self-Teaching: The Sine Qua Non of Reading Acquisition. Cognition, 55, 151-218.
Slowiaczek, M. L. & Clifton
Van Bon, W. H. J., Boksebeld, L. M., Freide, T. A. M. & Van den Hurk, A. J. M. (2001). A comparison of three methods of reading while listening. Journal of Learning Disabilities, 24, 471- 476.
Van Orden, G.C. (1987). A rows is a rose: Spelling, sound, and reading. Memory and Cognition, 15, 181-198.
Van Orden, G.C., Johnston
Author:
Lance R. Askildson, PhD
Assistant Provost for Internationalization
Director of the Center for the Study of Languages & Cultures
Associate Professor of the Practice for Second Language Acquisition
University of Notre Dame
Associate Professor of the Practice for Second Language Acquisition
University of Notre Dame
Notre Dame , IN 46556