Implementation of assisted repeated reading techniques for the incidental acquisition of novel foreign vocabulary

Abstract

Although repeated reading – a pedagogical practice often involving phonological support in which learners revisit novel forms in context – has been extensively studied in terms of reading fluency and comprehension, little research has explored whether it promotes language acquisition. The current study, an attempt to fill this gap, found that without explicit instruction 80 Mandarin-speaking intermediate learners of Japanese as a foreign language were capable of recognizing nearly one novel foreign orthographic form in every three seeded in passages after one hour of repeated-reading-based practice. The study also shows that the efficacy of repeated reading was modulated by how the phonological practice was implemented (e.g. shadowing, time-lapse imitation, subvocalization) and whether the target vocabulary was encountered in the same or different contexts. Additionally, repeated reading was found to be more effective with target vocabulary that shares etymological roots with the learners’ L1. Pedagogical implications based on these findings are discussed in terms of how repeated-reading-based techniques should be best implemented for the most positive outcomes in terms of incidental foreign vocabulary acquisition.

Keywords

Foreign language acquisition incidental vocabulary acquisition Japanese L2 reading repeated reading

I Introduction

The study described in this article set out to uncover the optimal cognitive and implementation conditions of repeated reading, a pedagogical intervention designed originally to improve reading fluency in first-language (L1) learners with reading difficulties (Therrien, 2004). More recently, practitioners and researchers have used repeated reading in early reading instruction for non-struggling readers (e.g. Taguchi & Gorsuch, 2002; Webb & Chang, 2012).

Unassisted repeated reading involves multiple successive readings of a passage, silently or aloud, until a gain in reading speed is observed and/or ‘a performance criterion is reached’ (Therrien, 2004, p. 259; compare Therrien & Watt, 2012). Assisted repeated reading, on the other hand, involves repeated reading whilst or after listening to either a teacher reading the same text or a recorded version. This is also called audio-supported repeated reading. Proponents of assisted repeated reading believe that phonological support allows learners to draw on the more mature oral speech system to crack the code of a written text.

A key component of repeated reading is the repetition of to-be-learned items in context, giving learners multiple opportunities to familiarize themselves with micro-level lexical constraints such as typeface, collocation, and orthographic features, as well as macro-level constraints such as contextual cues. Learners’ episodic memory of the micro- and macro-level constraints of novel forms from repeated reading sessions makes consequent recognition easier and faster.

Researchers have empirically established the efficacy of repeated reading with regard to comprehension and fluency. For instance, research with both L1 and second-language (L2) learners has indicated that it leads to faster reading rates (e.g. Hindin & Paratore, 2007; Millis & Simon, 1994; Taguchi, 1997; Stahl & Heubach, 2005) and improved comprehension (e.g. O’Shea, Sindelar, & O’Shea, 1985; Taguchi & Gorsuch, 2002; Taguchi, Takayasu-Maass, & Gorsuch, 2004).

However, such findings shed little light on whether repeated reading practices are capable of promoting language acquisition. As a matter of theoretical adequacy and logic, comprehension is not acquisition and cannot be taken as evidence that language acquisition has occurred. Accordingly, extant studies do not provide sufficient empirical evidence to gauge the efficacy of repeated reading in terms of language acquisition (Han & Chen, 2010). This underscores the need for further systematic research. To this end, the present study is designed to gauge the efficacy of repeated reading with regard to incidental acquisition of form–meaning information of novel L2 vocabulary. Following Han and Chen (2010), incidental vocabulary acquisition is defined as acquiring new words without direct study whilst engaging in meaning-oriented reading or listening.

II Literature review

1 Repetition and incidental L2 vocabulary learning

Incidental vocabulary acquisition through reading is a slow process involving multiple exposures to vocabulary in comprehensible contexts and is more likely in first language (L1) acquisition. In contrast, L2 learners rarely read extensively enough to incidentally acquire even the most frequent 3,000 words (Cobb, 2007).

Furthermore, even when incidental learning occurs through reading in L2, the gains are usually very limited. For instance, Horst, Cobb and Meara (1998) shows that L2 learners only incidentally acquired an average of five words after reading a comprehensible text of 21,000 words. However, a slightly different picture emerges in repeated reading.

Testing incidental vocabulary learning through repeated reading, Horst and Meara (2000) asked an L2 learner of Dutch to read a 6,000-word illustrated text once a week for eight weeks and tested the learner’s knowledge of 300 novel words via a four-point self-assessment scale four days after each reading. During the above procedure, the learner encountered the target words eight times. At the end of the treatment, the learner was able to identify the meaning and form of 223 words. Because the learner had no exposure to related Dutch input whilst reading and had no access to a dictionary or other learning support, his incidental vocabulary gains could be mainly attributed to the treatment.

Although, as noted by the researchers, some intentional learning may have been invoked due to the learner’s increasing awareness of the weekly tests, the study indicates that repeated reading may be a promising acquisitional tool. Indeed, as noted by Webb and Chang (2012), repeated reading might provide a way to help L2 learners overcome the limited number of incidental word gains before they are able to read extensively.

2 How should ‘repetition’ be operationalized for optimal acquisition?

Although ‘repetition in context’ is a core component of repeated reading, there is no consensus on whether repeated encounters with novel forms in the same context or different contexts maximizes repeated reading’s acquisitional value. Drawing on insights obtained from research into repeated reading and on read-alouds, some researchers hold that repeated encounters with novel forms in different contexts make repeated reading a powerful tool for enhancing language acquisition (Blachowicz & Obrochta, 2007; Therrien, 2004). In Blachowicz and Obrochta’s (2007) study, language learners were asked to read text sets in which the target vocabulary was encountered at least twice in multiple different contexts. Recognition of the vocabulary’s orthographic forms significantly increased after the rereading.

At first glance, ‘repeated encounters with novel forms in different contexts’ in repeated reading appears to overlap with the pedagogical practice of narrow reading, which also stresses repetition of to-be-learned items. However, a closer look at the two approaches reveals subtle differences between the two pedagogical practices. Specifically, whilst repeated reading greatly emphasizes the speed and fluency with which texts are read (Webb & Chang, 2012), reading speed is never an issue in narrow reading. Second, while repeated reading deliberately cues students’ attention to the same set of target vocabulary or forms via speeded, meaning-oriented reading (task-generated attention), narrow reading relies mainly on self-motivated (student-generated) attention to a set of thematically related words or forms, which do not all appear in each reading (see Han & Chen, 2010). Third, while narrow reading focuses on subconscious implicit language learning, repeated reading mainly depends on explicit learning (Hunt & Beglar, 2005). Martinez-Fernandez (2008, p. 210) notes that ‘whilst implicit learning can be incidental only, explicit learning can be both intentional and incidental’ (see Han & Chen, 2010, for an example). Due to the above differences in task demand and attentional investment involved, the cognitive mindset engaged in revisiting novel forms in repeated-reading-based settings should differ from that in narrow reading.

With a focus on reading speed and task-generated attention, the repetition in repeated reading can take the form of either repeated encounters with novel forms in the same or different contexts. The issue therefore is which form of repetition leads to the most desirable environment for language development? Whilst some scholars champion repeated reading of the same text (e.g. Kennison, Sieck, & Briesch, 2003), others contend that motivation and attention diminishes with each rereading (Nichols, Rupley & Rasinski, 2008). This point of contention is partially attributable to the fact that empirical evidence ‘in an ecologically valid context’ is lacking (Webb & Chang, 2012, p. 268). To date, the only methodologically appropriate evidence regarding the most desirable form of repetition comes from Therrien (2004), who found that the repeated reading of novel forms in the same text led to larger gains in reading fluency and comprehension than the repeated reading of novel forms in different texts. Furthermore, it was also found that such gains could be transferred to the reading of new articles.

Therrien’s findings, however, are not pertinent to language acquisition. More research is warranted to provide recommendations to teachers about which repetition context best facilitates language acquisition.

3 Does assisted repeated reading promote L2 acquisition?

Existing research, addressed below, has not provided unequivocal evidence to establish the acquisitional value of repeated reading. Han and Chen (2010) explored whether home-based repeated reading with phonological support provided a supportive environment for both incidental and intentional novel vocabulary acquisition. To this end, Anna, the participant, read self-selected articles from a Chinese newspaper. After 20 repeated-reading sessions, Han and Chen investigated Anna’s incidental and intentional vocabulary learning. Analysis of Anna’s performance in contextualized and decontextualized word recognition and production tasks showed that repeated reading had led to both intentional and incidental vocabulary acquisition. However, an important caveat is in order. Han and Chen’s study involved repeated reading with corrective feedback, explicit instruction and did not involve a control group. Anna’s vocabulary gains therefore cannot be solely attributed to the efficacy of repeated reading. In other words, the study does not establish that a reading lexicon in a second or foreign language can be developed through repeated reading without explicit teaching.

Zahar et al.’s (2001) study provides clearer evidence: 144 ESL learners, placed into five groups, ranging from ‘beginning to bilingual’, read a text in English, The golden fleece, chosen from a graded reader series. The learners read whilst listening to an audio recording of the story. Analysis of the participants’ performance on a multiple-choice word definition posttest showed that vocabulary learning, defined as the ability to map the form of novel words to contextually-appropriate meanings, occurred after an average of seven encounters.

This finding is confirmed by Webb and Chang (2012). Their study compared non-audio-supported and audio-supported repeated reading environments in which 82 adolescent EFL students were asked to, respectively, ‘sight read’ or ‘read and listen to’ 14 short stories over two seven-week periods. 92 novel words and eight unfamiliar words were seeded in the stories; of the 100 words, 85 were encountered once and 15 were encountered twice or more. A modified self-assessment test was administered to gauge the participants’ form–meaning mapping knowledge through repeated reading. Overall, the participants assigned to the audio-supported group showed better vocabulary gains. In particular, learners in this audio-supported group showed 12.03% and 14.14% vocabulary gains in the first and second seven-week periods, respectively. The better vocabulary gains during the second period was, allegedly, due to the learners’ familiarity with the repeated reading learning environment. Despite these encouraging results, Webb and Chang’s findings also cannot be solely attributed to repeated reading as their participants had access to dictionaries and were allowed to consult their peers during the repeated reading sessions. This might have promoted the learning of the target words through other external variables.

To date, a conclusive view regarding the acquisitional value of repeated reading is not possible due to the methodological differences among the limited number of relevant studies (e.g. audio-supported vs. non-audio-supported; production-based vs. non-production-based audio support). Specifically, whilst Han and Chen (2010) and Zahar et al. (2001) explored the acquisitional value of repeated reading with phonological practice, a phonological component was not included in Raney and Rayner’s (1995) study. Among the studies in which phonological practice was employed, some involved phonological production practice (Han & Chen, 2010) and others required learners to simply listen to phonological modeling (e.g. Webb & Chang, 2012). To shine more light on the role and nature of phonological practice in assisted repeated reading, a systematic study of the various forms of phonological practice is warranted.

4 Which form of phonological practice is most beneficial for L2 development?

Evidence from repeated reading research has generally supported the importance of directing learners’ attention to the aural aspect of a text (e.g. Blum, Koskinen, Tennant, Parker, Straub, & Curry, 1995; Brown, Waring, & Donkaewbua, 2008; Han & Chen, 2010). Nonetheless, researchers have not arrived at a consensus on which form of phonological practice is most beneficial for either fostering reading L2 literacy or promoting L2 acquisition. In implementing assisted repeated reading, some L2 instructors or researchers have learners read aloud whilst listening to phonological modeling of the text, students shadowing what they hear in real time (Commander & de Guerrero, 2013), whilst others use time-lapse imitation (Han & Chen, 2010). Still others require L2 learners to subvocalize whilst reading and listening to the teacher’s model (Samuels, 1979; Webb & Chang, 2012; Zahar et al., 2001). These three types of phonological practice require different degrees of attentional investment on the part of L2 learners. It is likely that shadowing requires the highest level of attention, whilst subvocalization probably requires the lowest.

The different degrees of attention invoked by the different phonological practices probably determined the different rates of vocabulary learning in Zahar et al. (2001) and Han and Chen (2010). Specifically, these two studies differ in terms of how phonological practice was implemented. Unlike the participant in Han and Chen who performed production-based time-lapse phonological practice, the participants in Zahar et al. passively listened, reading the words in their minds (subvocalization). The higher degree of attention required by the participant in Han and Chen (2010) could have enabled her to recognize 1.8 of every two words tested (55% success rate; Task 4), whilst the participants in Zahar et al’s (2001) study were only able to recognize one from every fourteen words tested (success rate: 7.1%). Further research is warranted on how attentional investment in phonological practice affects the acquisitional value of assisted repeated reading in L2 acquisition.

5 Is the efficacy of assisted repeated reading affected by L1–L2 formal overlap?

Research has indicated that phonological practice in assisted repeated reading helps L2 readers decode typologically-distant L2 script (e.g. Japanese learners of English; see Taguchi, 1997; Taguchi & Gorsuch, 2002). However, issues regarding the impact of L1–L2 formal overlap have not been systematically explored in existing assisted repeated reading research conducted with non-struggling L2 learners. Insights could be obtained by: (1) simultaneously studying groups of L2 learners with varying degrees of L1–L2 overlap (e.g. French learners of English vs. Chinese learners of English), and (2) studying a given group of L2 learners whose L1 differs from their L2 (e.g. script) to varying degrees in the same formal aspect (e.g. having some L1–L2 etymologically-distant words and some L1–L2 etymologically-close words). However, because no repeated reading studies have employed such methodological designs, the role of L1–L2 formal overlap has not been established.

Nevertheless, findings from bilingual word recognition research do shed some light on this issue. It has been shown that L2 learners of formally-distant languages (e.g. Chinese and English) attempt to extract phonological information while comprehending a text and that L2 learners’ knowledge or awareness of the structural similarities and/or differences between their L1 and L2 reliably predicts their fluency in reading aloud in both languages (Cheung, Chan, & Chong, 2007). This suggests that L1–L2 formal overlap in bilinguals could affect the efficacy of the phonological practice in the audio-supported repeated reading practice. Specifically, learners may fare better in audio-supported repeated reading when learning L2 words that are formally- or etymologically-close to their L1. However, to establish this, more systematic research is necessary.

III Research questions

In light of gaps in repeated reading research, four questions were explored:

Do the assisted repeated-reading-based techniques used in this study promote the incidental acquisition of novel foreign vocabulary?

Which of the two repeated reading environments, i.e. repeatedly encountering words in the same or in different texts, best enables the incidental acquisition of novel foreign vocabulary?

Is phonological practice in assisted repeated reading key to enhancing the incidental acquisition of novel foreign vocabulary? And, if so, what are the relative contributions of shadowing, time-lapse imitation, and subvocalization? And, do the relative contributions change according to the reading context (i.e. same vs. different texts)?

Is the efficacy of assisted repeated reading modulated when the foreign vocabulary is (or is not) etymologically related to the learners’ L1? And, if so, is this affected by the repeated reading environment (i.e. same vs. different texts)?

IV The study

1 Participants

Eighty (40 male and 40 female) high-intermediate Chinese learners of Japanese participated in this study, with a mean age of 19.8 years. All were Japanese majors and had received their Japanese language education in a university setting over an average of 3.9 years. The participants exhibited similar (high-intermediate) reading skills in Japanese, as attested to by their performance in the N3-level Japanese Language Proficiency Test (JLPT; The Japan Foundation and Association of International Education, 2004). With this level of vocabulary they were proficient at the 5,000-word reading level, as verified by the results of annual vocabulary tests.

2 Materials

According to Hu and Nation’s (2000) study on known word density and comprehension, learners need to know at least 98% of the running words in a text to gain adequate comprehension and learn new words. Therefore, to create appropriate texts for vocabulary acquisition, control of the unknown word density is essential. To this end, texts used for repeated reading can be easily obtained from language learning texts whose known word density and content have been carefully examined. Accordingly, the present study used passages from a language textbook.

Seven passages on the same theme were selected from the textbook. The passages were written by an experienced native instructor of Japanese as a foreign language (JFL), using level-appropriate vocabulary drawn from the 5,000 most frequent Japanese words listed in the Balanced Corpus of Contemporary Written Japanese. This vocabulary is required to pass the N3-level JLPT, which was the participants’ current proficiency level. The known word density of the seven passages used in this study was carefully examined by 10 language professionals. In the passages, the JFL instructor wrote on his observations of different aspects of education in Taiwan, a popular theme according to opinions expressed in a survey administered to prospective participants of the study. The seven passages were comparable in terms of both length (Mean = 1,208 words) and vocabulary, as verified by the JFL instructor. An oral reading of the seven passages, at normal reading speed and with appropriate pauses, was recorded by the same instructor.

Thirty-six target words were evenly seeded throughout each passage, with comparable contextual cues, as examined by 10 other JFL instructors. These words were selected from the vocabulary required to pass the N1-level and N2-level Japanese Language Proficiency Test and were hence slightly above the participants’ lexical knowledge. The words were tested, along with filler words, in a vocabulary pretest to ensure that the target words were indeed novel to the participants (for details, see Section 3).

Although the 36 target words were comparable in terms of difficulty,¹ they might have imposed different learning challenges to the participants due to their etymological attributes. To illustrate this, some linguistic information about Japanese script is in order.

Learning to read Japanese involves learning three different scripts (see Table 1 for examples):

Kanji, a writing system directly adopted from the logographic Chinese writing system and hence orthographically identical to the participants’ L1.

Hiragana, a traditional Japanese script with no Chinese logographic root used to write native words for which there are no Kanji.

Katakana, a writing script developed from more complex Kanji and which therefore only shares some orthographic features with the Chinese writing system.

Table 1.

Examples of the three Japanese scripts.

Hiragana	Katakana	Kanji

It is important to note that due to Japanese etymological development the lexical meaning of any particular character in Kanji may or may not correspond to the meaning of its original Chinese root.

As a result of the above features, written Japanese can be categorized into four types of lexical items, each type posing a different learning challenge to the participants of the study.

[–O, –S] lexical items: Orthographically unfamiliar and semantically opaque vocabulary (e.g. いらいら ‘nervous’);

[–O, +S] lexical items: Orthographically unfamiliar but semantically transparent vocabulary (e.g. 痺れる ‘numb’);

[+O, –S] lexical items: Orthographically familiar but semantically opaque vocabulary (e.g. 御馳走 ‘treating someone; banquet’);

[+O, +S] lexical items: Orthographically familiar and semantically transparent vocabulary (e.g. 応援, ‘assistance’).

The [–O, –S] lexical items (e.g.いらいら), often composed of Hiragana sub-lexical units, are neither orthographically familiar nor semantically transparent to Chinese learners of Japanese. Faced with novel [–O, –S] lexical items, Chinese learners of Japanese need to learn both the forms and corresponding semantic information from scratch.

When confronted with novel [–O, +S] lexical items (e.g. 痺れる), Chinese learners of Japanese face a different challenge. Specifically, [–O, +S] lexical items are often composed of two sub-lexical units: (1) Kanji (e.g. 痺), which originated from Chinese ideographs; and (2) Hiragana (e.g. れる), which have no Chinese etymological root. Comprehending and learning the semantic aspects of novel [–O, +S] lexical items is usually not a big issue to Chinese learners of Japanese, because their sub-lexical Kanji components usually convey semantic cues (see Han & Liu, 2013). Despite this, Chinese learners of Japanese still need to familiarize themselves with the orthographical features of [–O, +S] lexical items, because their sub-lexical Hiragana components bear very limited orthographic resemblance to Chinese characters.

On the other hand, novel [+O, –S] lexical items (e.g. 御馳走) present a totally different challenge. Specifically, Chinese learners of Japanese find mastering the writing of [+O, –S] lexical items easier because they are mainly composed of Kanji, which are orthographically identical to the learners’ L1. But learners still have to grasp the meanings of [+O, –S] items as they are not consistent with their original Chinese etymological roots.

In the study, the learning of the [–O, –S] vocabulary required acquisition of both orthographic and semantic aspects and was not facilitated by the participants’ L1. However, faced with the [–O, +S] and [+O, –S] vocabulary, the participants only had to master one aspect of lexical knowledge, either orthographic or semantic. The learning of the [–O, +S] and [+O, –S] vocabulary could thus be facilitated by the participants’ existing L1 knowledge.

In the present study the 36 target words, 12 [–O, –S] items, 12 [–O, +S] items, and 12 [+O, –S] items, were embedded in each of the seven reading passages. [+O, +S] items (e.g. 応援) were not used as target items because they are directly borrowed from Chinese characters and their meanings converge with their etymological roots.

3 Instruments

a A vocabulary pretest

The pretest assessed the participants’ knowledge of the target vocabulary prior to the repeated reading treatment. It was a 46-item multiple-choice definition test (the 36 target words plus 10 filler items). For each item, the participants had to match an L1 definition (e.g. 聚集, which means “flocking together”) with the appropriate target word. The formal distractor choices only differ from the correct target word by a morpheme. Below is an example:

聚集 (Flocking together):

群力る (formal distractor)

群がる (key)

わずか (filler)

裙がり (formal distractor)

助ける (filler)

b Repeated reading treatment

Following the typical repeated reading standard, the emphasis being on reading speed, the treatment was administered a week after the vocabulary pretest. The treatment aimed at promoting the participants’ incidental vocabulary learning.

c An oral summary of the read passage(s)

During the repeated reading treatment orientation, all the participants – irrespective of their group assignment – were informed that they were to complete a set of structured reading activities and would be asked at the very end to provide one oral summary of one passage. The purpose of the oral summary was to ensure that the passages were read for meaning and that any learning of the target vocabulary would be as a by-product. However, the summaries were not the focus of later analysis (for a similar design, see Goldman & Saul, 1990; Webb & Chang, 2012).

d A delayed vocabulary posttest

The delayed vocabulary posttest assessed whether the repeated-reading treatment had led to any incidental vocabulary gain. The items were the same as the pretest items, although the order was shuffled to avoid any memory effect. Each participant was awarded one point for a correct answer. The highest possible score was 36.

4 Design

The study followed a pretest–treatment–posttest experimental design. The following paragraphs describe the design of the study vis-à-vis the (a) vocabulary pretest and delayed posttest and (b) repeated reading treatment.

a Vocabulary pretest and delayed posttests

During the study, the participants’ vocabulary was tested twice: once before the experiment (vocabulary pretest) and once 24 hours after the repeated reading treatment (delayed posttest). In both tests, the participants were specifically told to answer each item only if and when they could recognize the word that they thought constituted the best match for the definition provided. This was to ensure that the participants would choose words they knew rather than opt for the unknown target word through random guessing. Although the participants’ attention was oriented to known words in the vocabulary pretest, there remains a possibility that exposure to the target words in the pretest primed them to give special attention to these words as they were encountered in the reading text(s). This possibility was reduced, however, by limiting the time allowed to answer each item to 10 seconds.

One hundred and twenty people participated in the vocabulary pretest; however, only those (n = 80) who left all the 36 target items unanswered – indicating a lack of form–meaning knowledge of the target words – participated in the repeated reading treatment. This ensured that all the target words were novel to the 80 participants. Given the participants’ pivotal point (zero knowledge of the form–meaning relationship of the target vocabulary), the instruction to orient their attention to known words, and the seven-day interval between the pretest and repeated reading treatment, memories of the target vocabulary, if any, would be very fragmented. Therefore, any significant vocabulary gains could mainly be attributed to the repeated reading treatment itself.

b Repeated reading treatment

The ensuing paragraphs describe the repeated reading treatment vis-à-vis its two major components: repetition and phonological practice.

Repetition

The repeated reading treatment, which took one hour to complete, was administered individually, with each participant randomly assigned to one of the two reading conditions: repeated encounters with the target words in the same text (RR-same) or repeated encounters with the target words in different texts (RR-different). The participants assigned to the RR-different group read all seven passages, reading each passage once. In doing so, they revisited the 36 target words seven times in seven different contexts. However, the participants assigned to the RR-same group read only one of the seven passages, but seven times. Therefore, the RR-same group also revisited the 36 target words seven times, but in the same context. Comparison of the two groups’ vocabulary learning – including the learning of the [–O, –S], [–O, +S], and [+O, –S] lexical items under the RR-same vs. RR-different conditions – would illuminate the comparative efficacy of the two repeated-reading environments. The number of encounters with the target vocabulary, seven, was mainly based on the findings of Zahar et al. (2001), reviewed earlier.

Guided phonological practice

All the participants – both the RR-same vs. RR-different groups – read passage(s) for meaning under one of the following conditions: (1) shadowing, (2) time-lapse imitation, (3) subvocalization, and (4) null phonological practice (baseline).

The participants who were shadowing were instructed to simultaneously listen and read aloud as accurately as possible. Whether they were able to perfectly shadow every word was not an issue; what really mattered was whether they attended to the oral input with focal attention. The participants under the time-lapse imitation condition were told to repeat as many words as possible at the end of a sentence. In subvocalization, shadowing without reading aloud, the participants were told ‘to imagine the sound of each word as it is read whilst listening to the recording’.

The participants involved in ‘null’ phonological practice read without listening to a recording and were instructed to read the passage with an ‘articulation suppression technique’. In performing this technique the participants had to make an irrelevant sound (e.g. [d]) approximately every three seconds whilst reading. This technique ensures that readers rely primarily on the visual modality. It originated in psycholinguistic studies but has been used in L2 research (e.g. Hamada & Koda, 2011) to eliminate the possibility of reading aloud or subvocalizing the text content. Three measures were used to reduce any possible negative effect arising from performing this technique. First, the participants were told to repeat the sound ‘in their mouth’ to avoid overt articulatory production. Second, instead of repeating a ‘real’ word, which would consume the same pool of attentional resources used to read the repeated text, the participants uttered a short non-word sound ([d]). Third, the participants did not have to intensively repeat the sound, only doing so every three to four seconds.

Comparing the participants’ vocabulary learning under the null phonological practice condition to the other three conditions allowed this study to explore the role of phonological practice in promoting incidental acquisition of form–meaning knowledge of novel foreign vocabulary and the optimal implementation of phonological practice. The various repetition and phonological practice conditions led to eight implementation conditions in the repeated reading treatment (see Figure 1). The presentation of the eight experimental conditions was counterbalanced across the participants using the Excel Randomization function.

Figure 1.

The eight implementation conditions in the repeated reading treatment.

5 Procedure

The study was comprised of the following stages: (1) recruitment of participants; (2) vocabulary pretest: (2) repeated reading treatment; (3) oral summary task; and (4) delayed vocabulary posttest. The study began with participant recruitment, and only those who had obtained the N3-level JLPT certificate and scored zero points on the vocabulary pretest were invited to participate in the repeated reading treatment administered a week later (n = 80).

When the participants returned for the one-hour repeated reading treatment, none were aware that a vocabulary posttest would be administered. They were only asked to read the passages for meaning and to provide an oral summary.

During the repeated reading treatment, the participants followed these steps:

Each participant was instructed in Japanese regarding the nature of the task and was given 3–5 minutes to familiarize himself or herself with it.

Each participant performed repeated reading under one of the following conditions: shadowing, time-lapse imitation, subvocalization, null phonological practice.

To reduce fatigue, each participant, regardless of group assignment, took a three-minute break after each round of repeated reading.

Upon finishing the repeated reading treatment, each participant, regardless of group assignment, provided one oral summary of one passage they had just read.

The repeated reading treatment took approximately one hour. After the repeated reading treatment, the participants were invited back for an additional meeting a day later. They only realized they were to take part in a delayed vocabulary posttest when returning for the meeting. This increased the possibility of any learning being incidental.

V Data analysis

The participants’ performance on the pretest and delayed posttest was the focus of the current study. The t-test was employed to detect if two sets of data (e.g. vocabulary pretest vs. posttest data) were significantly different from each other and to yield a preliminary picture for the queries of this study.

Furthermore, a multiple linear regression analysis was employed to yield more fine-grained data for the four research questions. Two regression models were proposed as explanatory accounts for examination of the effects of possible variables on the participants’ vocabulary learning outcomes.

Model 1: This model considers the effects of the RR-same environment, the phonological practice condition, the nature of the vocabulary, and the interaction among the aforementioned variables.

Model 2: This model considers the effects of the RR-different environment, the phonological practice condition, the nature of the vocabulary, and the interaction among the aforementioned variables.

VI Results

It is worth noting that on the vocabulary posttest, all the participants chose the correct keys for the items they elected to answer. No wrong answers were recorded.

Overall, the t-test results, in Tables 2 and 3, showed that the participants’ mean performance on the vocabulary posttest was significantly higher than their performance on the vocabulary pretest (Mean = 10.47 points; t = 38.65, p < .005).

Table 2.

Overview of descriptive data.

	RR-same	RR-different
Vocabulary gains under the RR-same and RR-different conditions:
Group mean	10.34*	10.59*
Effect of vocabulary type on vocabulary gains:
[–O, –S]	2.74	2.96
[–O, +S]	3.78	3.79
[+O, –S]	3.81	3.84

Note. * average = 10.47

Table 3.

Relative contributions of different forms of phonological practices to vocabulary gains.

	Mean
Shadowing	10.62
Time-lapse imitation	10.44
Subvocalization	10.42
Null	10.38

A significant incidental vocabulary gain was also observed in both the RR-same and RR-different groups. The RR-same and RR-different group participants obtained an average of 10.34 and 10.59 points on the vocabulary posttest, respectively. These scores were both significantly different from these on the vocabulary pretests (RR-same: t = 58.45, p < .005; RR-different: t = 59.84, p <.005). Furthermore, the t-test also revealed that the vocabulary gain in the RR-different group was significantly higher than that of the RR-same group (t = −1.95, p < .05).

In terms of the effect of vocabulary type on incidental vocabulary gains, the t-test indicated that both RR-same and RR-different groups had the lowest vocabulary gains with [–O, –S] items and that the participants generally had significantly better learning outcomes when faced with [–O, +S] and [+O, –S] vocabulary (see Tables 2 and 3).

Regarding the relative contributions of the four phonological practice conditions to incidental vocabulary gains, the average raw score of the participants’ vocabulary learning performance under the shadowing condition appears highest (Mean = 10.62). However, the t-test showed that when the other variables such as group assignment were not comprehensively considered, the participants’ vocabulary learning performance did not significantly differ among the four practice conditions. However, whether this holds true when other variables are considered (e.g. repeated reading environments, RR-same vs. RR-different) requires more fine-grained validation from a multiple linear regression test. The results of the multiple linear regression analysis will now be addressed.

As attested to by the R-squared values and F-values in Table 4, Models 1 and 2 satisfactorily explain the participants’ vocabulary learning outcomes as they explain nearly 52% of the vocabulary learning behavior (F = 22.25, p < .005); the eta-squared values (ŋ²) of both models are both .52, which is indicative of a large effect size regarding the impact of group difference (RR-same vs. RR-different) on the participants’ performance data. Exploring how well Models 1 and 2 explained the observed vocabulary learning outcomes, the multiple linear regression analysis showed that the β value for the RR-same condition is significant (β = –.94, p < .05), as is the β value for the RR-different condition (β = .85, p < .05). Based on the negative β value of Model 1 (i.e. ‘–’), it can be inferred that the vocabulary learning scores tended to be negatively associated with the RR-same condition (lower vocabulary gains). The positive β value of Model 2, on the other hand, suggests that the vocabulary learning scores tended to be positively associated with the RR-different condition (higher vocabulary gains). It thus follows that the vocabulary gains under the RR-different condition were generally better and higher than that under the RR-same condition. This significant cross-group difference is further confirmed by the testing power of Models 1 and 2, being 88.2 and 87.1%, respectively. These two testing power values indicate that the regression analysis result regarding the cross-group difference is very reliable. Furthermore, an interaction between the repeated reading context (RR-same vs. RR-different) and the phonological practice condition was observed in the examination of Models 1 and 2.

Table 4.

Multiple linear regression analysis results for Models 1 and 2.

	Model 1	Model 2
Constant	10.98*** (.308)	10.98*** (.308)
RR-same	−.94* (.436)	−
RR-different	−	.85* (.436)
Shadowing	−.88* (.356)	.88* (.356)
Time-lapse imitation	−.72* (.356)	.72* (.356)
Sub-vocalization	.03 (.356)	−.03 (.356)
RR-same * shadowing	1.73*** (.503)	−
RR-same * Time-lapse imitation	1.32* (.503)	−
RR-same * Sub-vocalization	.30 (.503)	−
RR-different * shadowing	−	−1.73*** (.503)
RR-different * Time-lapse imitation	−	−1.32* (.503)
RR-different * Sub-vocalization	−	−.30 (.503)
(–O, –S)	−2.63*** (.308)	−3.14* (.308)
(–O, +S)	−.16 (.308)	.10 (.308)
RR-same* (–O, –S)	−.50 (.436)	−
RR-same* (–O, +S)	.06 (.436)	−
RR-different* (–O, –S)	−	.50 (.436)
RR-different* (–O, +S)	−	−.06 (.436)
R-squared	51.8%	51.8%
F-value	22.25***	22.25***

Notes. *p < .05; **p < .01; ***p < .005.

Specifically, in Model 1, the β values of the RR-same * shadowing interaction (β = 1.73, p < .01) and RR-same * time-lapse imitation interaction were both significant and positive (β = 1.32, p < .05). On the other hand, the β values for the RR-same * sub-vocalization interaction and RR-same * null condition interaction were not significant. Collectively, in the RR-same reading environment regarding shadowing and time-lapse imitation, the β values tended to be positively associated with better vocabulary gains. However, the subvocalization and null conditions did not result in a significant impact on the participants’ vocabulary learning outcomes (for visual schematization, see Figure 2). It can thus be seen that the RR-same group participants assigned to the shadowing and time-lapse imitation conditions tended to outperform their counterparts under the subvocalization and null conditions.

Figure 2.

Relative contributions of the four practice conditions for the repeated encounters with the target words in the same text (RR-same group).

In the examination of Model 2, a significant interaction between the repeated reading context (RR-different) and phonological practice condition was also obtained. Specifically, the β values for the RR-different * shadowing interaction and RR-different * time-lapse imitation interaction were −1.73 (p < .01) and −1.32 (p < .05), respectively. The significant negative β values indicate that in the RR-different reading environment, shadowing and time-lapse imitation conditions tended to be negatively associated with better vocabulary gains. On the other hand, the β values under the sub-vocalization and null conditions were not significant. Accordingly, the shadowing and time-lapse imitation conditions did not significantly promote the RR-different group participants’ vocabulary learning. In fact, the two conditions were associated with lower vocabulary learning performance (in comparison with subvocalization and null conditions) for the RR-different group: the opposite of what was observed in the RR-same reading environment (see Figure 3).

Figure 3.

Relative contributions of the four practice conditions for the repeated encounters with the target words in different texts (RR-different group).

Despite the between-group differences concerning the relative contributions of the four phonological conditions, a similarity between the RR-same and RR-different groups emerges. The analysis showed that the β values in Models 1–2 were both significant and negative when the vocabulary was [–O, –S], i.e. –2.63 (p < .005) and −3.14 (p < .005), respectively. This suggests that in both Models 1 and 2 lower vocabulary gains were associated with the [–O, –S] vocabulary. In turn this suggests that the efficacy of repeated reading was somewhat disrupted by the [–O, –S] vocabulary both in the RR-same and RR-different environments.

VII Discussion and pedagogical implications

This study addressed four questions. The first asked whether repeated-reading promotes the incidental learning of foreign vocabulary. As attested to by the significant score differences between the vocabulary pretest and posttests, the participants were able to retain the forms of novel vocabulary even when specifically told to read for meaning and unaware that they would be tested on vocabulary. The participants scored 10.47 out of 36 on the vocabulary posttest, identifying one word of every 3.44 tested. In comparison to Webb and Chang’s (2012) study, the incidental vocabulary gains through audio-supported repeated reading in the present study were markedly higher. Furthermore, whilst most of the participants of the present study were able to identify the correct form–meaning knowledge of the target words they incidentally acquired, the participants in Webb and Chang (2012) showed differential success and failure rates on the same task. These greater vocabulary gains may be attributed to the fact that the participants in this study were given more opportunities (i.e. seven) to revisit the target vocabulary, whilst the reading materials used by Webb and Chang did not have such a high encounter rates. This possibility is consistent with Zahar et al.’s (2001) and Horst and Meara’s (2000) studies in which words encountered more than seven times were likely to be incidentally acquired. When passages are only repeated three or four times or less, gains would mostly be manifested in reading fluency and comprehension, but not in acquisition (see Therrien, 2004; Therrien & Hughes, 2008).

Research questions 2–3 were pursued to address the relative contributions of the RR-same and RR-different environments to the incidental acquisition of novel foreign vocabulary. Analyses of the participants’ posttest performance indicate that the acquisitional potential of repeated reading is enhanced when novel vocabulary is repeatedly encountered in different thematically-related passages (RR-different), although both repeated reading contexts led to vocabulary gains. This could be because, as some researchers (e.g. Zahar et al., 2001) contend, variable contextual cues in the RR-different environment actually incur a more facilitative cognitive state of mind for language acquisition. Specifically, when reading unknown or partially acquired words in different passages, learners inevitably encounter contextually neutral and/or opaque contexts. Such contexts ‘open up a learning need, or conceptual gap’ (Zahar et al., 2001, p. 556) and hence trigger learners’ internal drive to perform form–meaning mapping when eventually encountering the words in rich, clear contexts.

In contrast, when repeatedly reading unknown or partially acquired words in the same passage (RR-same), the participants were not allowed to ‘flounder in the variable contexts’ (Zahar et al., p. 545), which vary in degrees of contextual constraint (e.g. clear, neutral, opaque). Furthermore, each rereading of a passage enhances a language learner’s familiarity with, and global comprehension of, the passage (Therrien, 2008). In this case, given the observed universal input processing predilection – that learners process input primarily for meaning (VanPatten, 2008) – learners may or may not see the need for further exploration of the micro-level context which can enhance the form–meaning mapping of novel words. In such situations, repeated encounters with novel words in the same passage, analysis of the formal features of novel words will be less active. In contrast, leading learners to revisit target foreign vocabulary in different texts may lead to better vocabulary gains.

The third question probed the role of phonological practice in repeated reading for promoting incidental vocabulary acquisition and the relative contributions of shadowing, time-lapse imitation, and subvocalization, as compared with the baseline (null). Overall, as shown in the regression analysis, when the RR-same and RR-different groups’ data were collapsed, the vocabulary learning outcome under the null phonological practice condition was not significantly different from any of the other three phonological practice conditions. This appears to suggest that phonological practice did not enhance incidental vocabulary acquisition. However, when the differences in the RR-same and RR-different reading environments were considered in Models 1 and 2, subtle differences emerged. The regression analysis of these models showed that under the RR-same and RR-different reading environments shadowing and time-lapse imitation both exerted a significant although opposite impact on the participants’ vocabulary learning. Specifically, as displayed in Figure 4, under the RR-same environment shadowing and time-lapse imitation had a more pronounced facilitative impact on the participants’ vocabulary learning compared with the sub-vocalization and null conditions. Accordingly, when RR-same repeated reading is used as a pedagogical practice, directing learners’ focal attention to the phonology of the text via shadowing or time-lapse imitation would probably maximize incidental vocabulary acquisition. In these two types of audio-supported repeated reading, the participants showed greater gains than when repeated reading without phonological output.

Figure 4.

Relative contributions of the four practice conditions in environments with the repeated encounters with the target words in the same text (RR-same) and repeated encounters with the target words in different texts (RR-different).

In stark contrast, in the RR-different reading environment, the participants’ incidental vocabulary gains were more pronounced under the sub-vocalization and null conditions. This indicates that with RR-different repeated reading, directing students’ focal attention to the phonology of the texts being read is best avoided.

The above findings help build understanding of the most efficacious phonological practice in different (RR-same/RR-different) repeated reading environments. Specifically, for the purpose of promoting incidental vocabulary development, phonological practice in assisted repeated reading seems beneficial, although different phonological practices should be considered depending on whether the target vocabulary is revisited in the same or different contexts. Explicit attention to the phonology of a text whilst reading leads to enhanced vocabulary acquisition only in an RR-same reading environment where learners with each re-reading become incrementally more familiar with the meaning of the passage.

The fourth question explored whether the efficacy of repeated reading techniques is modulated by the nature of the novel foreign vocabulary, i.e. with words having (or having no) etymological derivation from the learners’ L1. As previously noted, the efficacy of repeated-reading-based techniques seems to be dampened by the [–O, –S] vocabulary, but facilitated by the [–O, +S] and [+O, –S] vocabulary, words with etymological traces in the participants’ L1. This was observed both when the target vocabulary was revisited in the same and different texts. It may be that the participants – irrespective of repeated reading environment (RR-same or RR-different) – already had some knowledge required for learning the [–O, +S] and [+O, –S] vocabulary and that this made learning such words arguably comparable to the learning of partially-learned lexical items. In contrast, the [–O, –S] vocabulary, with no etymological roots in the learners’ L1, is entirely novel. If the above argument is valid, it may be that repeated reading works better for foreign vocabulary with which the learners already have some etymological knowledge, and is less effective with novel words that are etymologically novel.

VIII Conclusions

The evidence provided by this study shows significant gains, in seven encounters and without any form of instruction, in developing form–meaning knowledge of unknown foreign vocabulary through a one-hour repeated-reading-based activity. The participants, Mandarin-speaking learners of Japanese, were able to recognize 33.7% of the target words in a vocabulary test administered 24 hours after the initial exposure without confusing these words with formally similar lexical entries. These results are perhaps particularly significant for foreign- or second-language teachers seeking a useful pedagogical practice for vocabulary learning activities.

1 Limitations

Despite the positive results of this study, several limitations should be noted. First, the study is mainly based on the results of a form–meaning mapping task. Consequently, it remains unclear whether such techniques promote other aspects of vocabulary knowledge. Second, due to the cross-sectional nature of the study design and interval between the repeated-reading treatment and vocabulary posttest, this study tells us little about whether the acquired form–meaning knowledge persists over time. The third limitation concerns the lack of a control group. Although the participants exhibited significant vocabulary growth, a stronger case could be made about the acquisitional value of repeated reading if a control group were included. Fourth, in the current study there were eight repeated-reading environments (2 Group conditions * 4 Practice conditions), with each environment populated by only 10 participants. It must be recognized that a larger sample size would greatly enhance the validity of the claims. Finally, although the oral reporting of passage content was intended to ensure that the participants read the passages for meaning, such a task might have occasionally encouraged them to deploy more attentional resources on the form of novel words than they would usually do when reading. Thus, alternative assessment methods are probably necessary to gauge the amount of occasional intentional learning in assisted repeated reading.

2 Future research

It would be useful to explore how repeated reading affects different aspects of language knowledge (e.g. productive knowledge) and language acquisition of a different nature (incidental vs. intentional) using multiple tests.

Furthermore, because this research indicates that assisted repeated reading may lead to incidental vocabulary acquisition, further research in this area may want to explore whether the efficacy of such a pedagogical practice is modulated by individual learning styles (e.g. auditory vs. visual). Finally, it would be interesting to investigate the effectiveness of a hybrid form of repeated reading. Learners could read several texts (e.g. three) twice, the first reading focusing on understanding the content and the second on the formal aspects of the target vocabulary. With meaning clarified by the first reading, more cognitive space would be available during the second reading to attend to the formal aspects of the target vocabulary. The same routine could be repeated for each text in which the same target vocabulary is seeded. This may embrace the advantages of both the RR-same and RR-different environments and lead to better vocabulary learning outcomes.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Blachowicz

C.L.Z.

Obrochta

(2007). Tweaking practice: Modifying read-alouds to enhance content vocabulary learning in grade 1. In Rowe

D.W.

Jimenez

Compton

Dickson

Kim

Leander

. (Eds.), Fifty-seventh Yearbook of the National Reading Conference (pp. 140–150). Oak Creek, WI: National Reading Conference.

Blum

Koskinen

P.A.

Tennant

Parker

E.M.

Straub

Curry

(1995). Using audiotaped books to extend classroom literacy instruction into the homes of second-language learners. Journal of Reading Behavior, 27, 535–563.

Brown

Waring

Donkaewbua

(2008). Incidental vocabulary acquisition from reading, reading-whilst-listening, and listening to stories. Reading in a Foreign Language, 20, 136–163.

Cheung

Chan

Chong

(2007). Use of orthographic knowledge in reading by Chinese-English bi-scriptal children. Language Learning, 57, 469–505.

Chujo

Utiyama

Nishigaki

(2007). Towards building a usable corpus collection for the ELT classroom. In Hidalgo

Quereda

Santana

(Eds.), Corpora in the foreign language classroom (pp. 47–69). Amsterdam: Rodopi.

Cobb

(2007). Computing the vocabulary demands of L2 reading. Language Learning and Technology, 11, 38–63.

Commander

de Guerrero

M.C.M.

(2013). Reading as a social interactive process: The impact of shadow-reading in L2 classrooms. Reading in a Foreign Language, 25, 170–191.

Goldman

S.R.

Saul

E. U.

(1990). Flexibility in text processing: A strategy competition model. Learning and Individual Differences, 2, 181–219.

Hamada

Koda

(2011). The role of the phonological loop in English word learning: A comparison of Chinese ESL learners and native speakers. Journal of Psycholinguistic Research, 40, 75–92.

10.

Han

Z-H.

Chen

C.L.

(2010). Repeated-reading-based instructional strategy and vocabulary acquisition: A case study of a heritage speaker of Chinese. Reading in a Foreign Language, 22, 242–262.

11.

Han

Z.H.

Liu

Z.H.

(2013). Input processing of Chinese by ab initio learners. Second Language Research, 29, 145–164.

12.

Hindin

Paratore

J.R.

(2007). Supporting young children’s literacy learning through home-school partnerships: The effectiveness of a home repeated-reading intervention. Journal of Literacy Research, 39, 307–333.

13.

Horst

Meara

P.M.

(2000). Test of a model for predicting second language lexical growth through reading. Canadian Modern Language Review, 56, 308–328.

14.

Horst

Cobb

Meara

(1998). Beyond a Clockwork Orange: Acquiring second language vocabulary through reading. Reading in a Foreign Language, 11, 207–223.

15.

Nation

I.S.P.

(2000). Vocabulary density and reading comprehension. Reading in a Foreign Language, 13, 403–430.

16.

Hunt

Beglar

(2005). A framework for developing EFL reading vocabulary. Reading in a Foreign Language, 17, 23–59.

17.

The Japan Foundation and Association of International Education. (2004). Japanese Language Proficiency Test: Test content specifications. Tokyo: Japan Bonjinsha Books.

18.

Kennison

S.M.

Sieck

J.P.

Briesch

K.A.

(2003). Evidence for a late-occurring effect of phoneme repetition during silent reading. Journal of Psycholinguistic Research, 32, 297–312.

19.

Martinez-Fernandez

(2008). Revisiting the involvement load hypothesis: Awareness, type of task, and type of item. Unpublished paper presented at the 2007 Second Language Acquisition Forum, Somerville, MA, USA.

20.

Millis

Simon

(1994). Rereading scientific texts: Changes in resource allocation. Advances in Discourse Processes, 53, 115–134.

21.

Nichols

W.D.

Rupley

W.H.

Rasinski

(2008). Fluency in Learning to read for meaning: Going beyond repeated readings. Literacy Research and Instruction, 48, 1–13.

22.

O’Shea

L.J.

Sindelar

P.T.

O’Shea

D.J.

(1985). The effects of repeated readings and attentional cues on reading fluency and comprehension. Journal of Literacy Research, 17, 129–142.

23.

Raney

G.E.

Rayner

(1995). Word frequency effects and eye movements during two readings of a text. Canadian Journal of Experimental Psychology, 49, 151–172.

24.

Samuels

S.J.

(1979). The method of repeated readings. The Reading Teacher, 50, 376–381.

25.

Stahl

S.A.

Heubach

K.M.

(2005). Fluency-oriented reading instruction. Journal of Literacy Research, 37, 25–60.

26.

Taguchi

(1997). The effects of repeated readings on the development of lower identification skills of FL readers. Reading in a Foreign Language, 11, 97–119.

27.

Taguchi

Gorsuch

G.J.

(2002). Transfer effects of repeated EFL reading on reading new passages: A preliminary investigation. Reading in a Foreign Language, 14, 43–65.

28.

Taguchi

Takayasu-Maass

Gorsuch

G.J.

(2004). Developing reading fluency in EFL: How assisted repeated reading and extensive reading affect fluency development. Reading in a Foreign Language, 16, 70–96.

29.

Therrien

W.J.

(2004). Fluency and comprehension gains as a result of repeated reading: A metaanalysis. Remedial and Special Education, 25, 252–261.

30.

Therrien

W.J.

(2008). Fluency and comprehension gains as a result of repeated reading: A meta-analysis. Remedial and Special Education, 25, 252–261.

31.

Therrien

W.J.

Hughes

(2008). Comparison of repeated reading and question generation on students’ reading fluency and comprehension. Learning Disabilities: A Contemporary Journal, 6, 1–16.

32.

Therrien

W.J.

Watt

(2012). Repeated reading. In Hattie

Anderman

(Eds.), International handbook of student achievement (pp. 320–321). London: Routledge.

33.

VanPatten

(2008). Processing matters in input enhancement. In: Piske

Young-Scholten

(Eds.), Input matters (pp. 47–61). Clevedon: Multilingual Matters.

34.

Webb

S.A.

Chang

C-S . (2012). Vocabulary learning through assisted and unassisted repeated reading. Canadian Modern Language Review, 68, 267–290.

35.

Zahar

Cobb

Spada

(2001). Acquiring vocabulary through reading: Effects of frequency and contextual richness. The Canadian Modern Language Review, 57, 541–572.