Applied Linguistics 28/1: 66–86 doi:10.1093/applin/aml049 ß Oxford University Press 2007 The Effects of Topic Familiarity and Passage Sight Vocabulary on L2 Lexical Inferencing and Retention through Reading DIANA PULIDO Michigan State University The study examines the impact of topic familiarity and passage sight vocabulary on lexical inferencing and retention. Independent variables include (a) a topic familiarity questionnaire, and (b) a passage sight vocabulary test. A repeated-measures design was used with a cross-sectional sample of 35 adult L2 learners of Spanish. Ss read narratives (1 more and 1 less familiar) containing nonsense words. They guessed meanings of target words and rated degree of difficulty in guessing. After reading all stories, all participants completed an inference verification task to confirm or correct guesses, and to encourage deeper processing of target words. This was followed by two measures of retention: (a) translation production, and (b) translation recognition. Analyses reveal (a) robust effects of topic familiarity and passage sight vocabulary on lexical inferencing, (b) a significant interaction between topic familiarity and passage sight vocabulary on difficulty in lexical inferencing, and (c) robust effects of passage sight vocabulary on lexical retention and a significant effect of topic familiarity on lexical retention–translation recognition measure. The discussion concerns the significance of findings for lexical inferencing, processing, and retention through strategic reading tasks. L2 vocabulary development through reading is complex. It takes place through various component processes involved in text processing and comprehension. For example, it involves noticing that particular words are unfamiliar. Then, in the absence of dictionaries or human assistance, it requires inferring meaning (lexical inferencing), using context cues and linguistic and extra-linguistic knowledge (Faerch et al. 1984). Readers must also attend to the connections between new lexical forms and their meanings and integrate the new linguistic information into their developing language system. This involves some unspecified degree of elaborative rehearsal—forming connections between the new lexical form and meaning and associating these with previous knowledge (Baddeley 1998). If there are too many constraints on the individual’s processing capacity (McLaughlin 1987; Just and Carpenter 1992), characteristic of lower proficiency learners, or if unfamiliar words are not deemed important enough to warrant deeper processing (Craik and Tulving 1975), then these words may be processed more superficially, and are less likely to be retrieved from memory DIANA PULIDO 67 (see also Ellis 1994, 2001; Gass 1999; Hulstijn 2001, 2003; Laufer and Hulstijn 2001; Robinson 2003; Schmidt 2001). An attempt to operationalize the construct of elaboration during L2 incidental vocabulary acquisition is the Involvement Load Theory (Laufer and Hulstijn 2001), which proposes that retention of new words depends upon levels of ‘need’, ‘search’, and ‘evaluation’ imposed by reading tasks. ‘Need’, a motivational aspect, concerns the need to comply with the task. ‘Search’ entails making attempts to determine the meaning of an unknown word, and ‘evaluation’ involves making a decision about ‘semantic and formal appropriateness (fit) of the word and its context’ (Laufer and Hulstijn 2001: 15), the latter two being cognitive processing components. The theory predicts that the greater the involvement in a given task, the better the retention. Thus, to establish form–meaning connections for new words through reading, the new words must be noticed and sufficient attention allocated. The quality of processing during reading and on any tasks intended to facilitate integration of new words is crucial. Moreover, text processing and comprehension are influenced by reader-based factors such as background knowledge and L2 proficiency (see Pulido 2000, 2003, 2004b for further explanation). Scarce, however, is research investigating, in a controlled fashion, the impact and interaction between such readerbased factors in L2 lexical inferencing and retention. A study of this nature may not only contribute to interactive theories of reading and further refinement of models of lexical development through reading, but also help to explain the variability found in vocabulary learning outcomes through reading within instructed L2 learning environments. REVIEW OF RESEARCH Effects of background knowledge on lexical inferencing and retention Research reports that greater levels of background knowledge contribute to efficiency of attentional allocation to input during reading, enabling richer textual interpretations, and, in turn, superior memory performance (e.g. Ellis 2001; Graesser et al. 1994; Kintsch 1998; Nassaji 2002; Robinson 2003; Rumelhart 1980; Schank and Abelson 1977). During reading the syntactic, semantic, and pragmatic knowledge that is activated, held in working memory, and utilized online during text processing can constrain subsequent textual and lexical interpretations. The De Bot, Paribakht, and Wesche (1997) model of L2 vocabulary processing during reading describes subprocesses that serve to activate various knowledge when connecting a new word form (i.e. a lexeme, containing phonological and morphological specifications) to its semantic and syntactic specifications (i.e. its lemma). For example, upon recognizing words, readers can use the activated semantic information about meaning relations and semantic fields, and pragmatic information about conventional and contrastive uses of words and 68 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY expressions to interpret meanings of unfamiliar words. When processing the new words themselves, readers can use morphological information (e.g. bound roots, inflectional, or derivational morphemes) to obtain clues about word class, word meaning components, grammatical function, and semantic roles. During sentence parsing, readers can use syntactic information (e.g. word order and argument structure), combined with knowledge about event probabilities, to identify semantic information and roles of new lexical items. When a word’s meaning is not known, new form– meaning mapping can occur through the bootstrapping processes described above (e.g. for L1 fast-mapping see Clark 1993; for L2 accounts see De Bot et al. 1997; Nassaji 2003; Paribakht and Wesche 1999), in combination with cues in the local or global context (Chern 1993; Nagy 1997; Sternberg 1987). Thus, adult L2 learners may use whatever information available to interpret new words, to the extent that their proficiency level enables them to use such information. The significance of this explanation for understanding L2 lexical development through reading is that the role of background knowledge that is stored in long-term memory is largely determined by the quality of the textbase that is constructed during reading, which is affected by the individual’s text processing efficiency (i.e. ability in word recognition and syntactic processes, sentence parsing, and use of causal and rhetorical knowledge structures) and working memory. All of these elements function in tandem in generating initial form–meaning connections for new words. Research within a lexical inferencing paradigm1 has observed strategies and knowledge sources that L2 learners use to infer word meanings. Studies with advanced and intermediate ESL (Chern 1993; Nassaji 2003, respectively) and beginning EFL learners (Haastrup 1989) illustrated that learners of all levels relied on background knowledge to guess word meanings during think-aloud protocols. However, Rott (2000), also using a think-aloud task, found that few intermediate learners of German used background knowledge during inferencing. Similarly, De Bot et al. (1997) and Paribakht and Wesche (1999) reported that intermediate ESL learners appealed less frequently to background knowledge than to grammatical knowledge during a retrospective think-aloud task. In a cross-sectional study Lee and Wolf (1997) observed that native Spanish speakers used background knowledge the most to infer meaning, followed by the advanced, intermediate, and then beginning learners of Spanish during a retrospective think-aloud task. Finally, Parry’s (1993, 1997) longitudinal case studies described advanced ESL learners’ use of knowledge about anthropology to select which words to guess in their anthropology texts. These introspective accounts generally illustrated learners using background knowledge to some degree, whereas several studies demonstrated advanced learners using it more frequently. A few studies controlled for background knowledge, or measured inferencing in contexts varying in degree of semantic richness. Adams (1982) found that script activators (i.e. statements describing the topics) DIANA PULIDO 69 facilitated lexical inferencing for beginning learners of French when reading paragraphs, one sentence at a time, about everyday activities. Li (1988) reported more success and greater ease in guessing (determined by Likertscale ratings) when Chinese advanced learners of English processed semantically rich compared to semantically poor sentences. Likewise, Mondria and Wit-de Boer (1991) reported more successful inferencing in semantically rich sentences with Dutch intermediate learners of French. The studies above demonstrate that background knowledge and rich semantic contexts affect lexical inferencing in discrete sentence contexts. Few of the studies above assessed retention after the inferring phase. Retention in the present study refers to memory for the correct meaning of the word. Li (1988) reported superior retention of the words in the semantically rich sentences based on a cued recall. Rott (2000) also reported occasional successful recall of target word (TW) meanings when learners had used background knowledge to guess the meanings. However, Mondria and Wit-de Boer (1991), who assessed retention after a verification task to confirm or correct the guesses, reported low retention on an L2–L1 translation task, and weak and negative correlations between inferencing and retention. Similarly, Mondria (2003) also found low retention of correctly inferred words after a verification task, and high correct retention of incorrectly inferred words in a study with intermediate learners of Dutch. The latter two studies concluded that due to the strong association of context and meaning, and assumed ease in guessing, learners likely did not pay sufficient attention to the word form and its meaning during the learning tasks. A few studies investigated effects of background knowledge on retention of nonsense words, via an incidental research paradigm, where learners were not alerted to the vocabulary tests (see Hulstijn 2001, 2003 for methodological descriptions). Pulido (2000, 2003) measured retention after a cross-section of adult learners of Spanish read two stories describing more familiar scenarios (grocery shopping and doctor visit) and two describing less familiar scenarios (publishing and homebuying). Familiarity was determined by participants’ ratings on a questionnaire. Pulido found short-term facilitative effects of background knowledge on a multiple-choice translation recognition test administered two days after reading. In another study, with intermediate Spanish learners, Pulido (2004b) obtained facilitative effects of cultural background knowledge on immediate retention of nonsense words after learners read narratives depicting either culturally familiar or culturally unfamiliar versions of everyday scenarios (registering for classes and grocery shopping). Results from the adapted version of the Vocabulary Knowledge Scale (VKS), (Paribakht and Wesche 1993), reflected only self-reported memory for having seen the TWs within the stories, which is regarded as an initial stage in vocabulary acquisition. There are limitations to these studies. The majority of the inferencing studies did not measure retention of inferred words, and the retention-only studies did not measure inferencing. Most of the inferencing studies did not 70 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY assess background knowledge or ‘difficulty’ in guessing, which may provide insight on individual differences in strategies or outcomes. Most studies used expository texts, concerning difficult themes. Both factors combined may have curtailed opportunities to appeal to background knowledge. Moreover, most focused on one level of proficiency, prohibiting necessary crosssectional generalizations. Finally, most studies addressing inferencing and retention focused on discrete sentences, rather than text processing, which places additional constraints on memory because readers must construct, integrate, and store meaning across sentence boundaries. This study aimed to fill this gap by investigating the impact of background knowledge on both lexical inferencing and retention for narrative text processing, via a crosssectional, repeated measures design that accounts for background knowledge. Effects of passage sight vocabulary on lexical inferencing and retention Reading also entails the use of linguistic knowledge. One type of competence contributing to text processing, comprehension, and vocabulary acquisition is vocabulary knowledge associated with the texts, hereafter referred to as passage sight vocabulary. With efficient decoding skills and large sight vocabularies, readers can allocate attentional resources to the construction and integration of ideas from context, and the access and use of information from long-term memory. This results in a greater likelihood of successful lexical inferencing. Without such processing skills and proficiency, readers are more apt to experience a short-circuit of the lexical inferencing and integration process, and fewer chances for vocabulary development (e.g. Grabe and Stoller 2002; Koda 2005; Laufer 1997). Although studies report positive significant relationships between general breadth of vocabulary knowledge and vocabulary gains through reading (Haynes and Baker 1993; Horst et al. 1998), and general depth of vocabulary knowledge and lexical inferencing success (Nassaji 2004), such findings do not illustrate the precise relationship between vocabulary knowledge specific to a given text and lexical development associated with processing that text. Several studies reported in the previous section also observed more successful lexical inferencing when readers knew vocabulary in the passage (e.g. Haynes 1993; Haynes and Baker 1993; Lee and Wolf, 1997; Parry 1997; Rott 2000). Lower-proficiency learners experienced more difficulty in integrating multiple textual and extra-textual cues (background knowledge) than did higher proficiency learners, who appeared to know more words in the context. These results were determined by anecdotal observations, rather than quantitative measurements of passage sight vocabulary. Research concerning the impact of measured passage specific sight vocabulary is scarce. Within an incidental framework, Holley (1973) found no effects of new-word density on vocabulary retention after beginning German learners read a narrative text. Pulido’s (2000, 2003) study reported DIANA PULIDO 71 above also found that passage sight vocabulary, measured by a translation test, was a robust predictor of vocabulary retention measured by translation production and recognition tests 2 and 28 days after reading. As passage sight vocabulary increased, so did lexical retention. Moreover, on the difficult production measure, passage sight vocabulary was more robust than the variable of background knowledge. Pulido (2004b) also investigated passage sight vocabulary effects on immediate vocabulary retention by intermediate learners of Spanish, but found only one high significant positive correlation from the four narratives (i.e. a culturally familiar story about grocery shopping). The lack of more robust findings may have been an artifact of the homogenous group of participants, the small amount of passage sight vocabulary tested, or the gain measure (e.g. an adapted version of the VKS). A limitation of most of the research cited above is the absence of a measure of passage sight vocabulary and systematic study of its role in both lexical inferencing and retention. To the best of my knowledge there is no such study investigating the role of passage sight vocabulary on the lexical inferencing process itself. Based upon the review and limitations reported above, this study investigates the nature of the impact and interaction of the reader-based variables of passage sight vocabulary, and background knowledge, hereafter referred to as topic familiarity, when a cross-section of L2 learners are assigned strategic tasks involving lexical inferencing and processing of new vocabulary in narrative passages. Research questions The following questions guided this study: 1 Which of the following factors have a significant impact on L2 lexical inferencing when reading brief narratives: (a) topic familiarity, (b) passage sight vocabulary? 2 Which of the following factors have a significant impact on difficulty in lexical inferencing when reading brief narratives: (a) topic familiarity, (b) passage sight vocabulary? 3 Which of the following factors have a significant impact on L2 lexical retention from reading brief narratives, after an inference verification task to confirm or correct original guesses: (a) topic familiarity, (b) passage sight vocabulary? First, from previous research it was hypothesized that topic familiarity would significantly affect both lexical inferencing and retention, yielding greater outcomes when subjects are more familiar with a particular topic. Next, topic familiarity was also expected to affect degree of ease/difficulty in lexical inferencing given the richer context and greater likelihood that learners could utilize background knowledge successfully during reading. Last, passage sight vocabulary was expected to have a significant impact on 72 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY lexical inferencing, difficulty in lexical inferencing, and retention. As passage sight vocabulary increases, learners should not only have fewer words to learn and fewer constraints on memory, they should also have more available context to enable deeper processing of new words, contributing to ease and success in lexical inferencing, processing, and retention. METHOD Participants Thirty-five adult learners of Spanish as an L2 served as participants. They were recruited from five university courses: beginning (intensive for false beginners), n ¼ 8; intermediate (4th semester language), n ¼ 8; highintermediate (5th and 6th semester grammar and composition), n ¼ 6; advanced (7th and 8th semester literature), n ¼ 11; and graduate (Spanish literature), n ¼ 2. All participants were native speakers of English only. Passages The texts were two contrived narrative passages, one depicting a more familiar scenario and one a less familiar scenario (see Appendix A; available online at http://applij.oxfordjournals.org). The more familiar passage, ‘The Trip to the Supermarket’, was based on a scenario considered to be routine and very familiar to the participants. The less familiar passage, ‘Publishing an Article’, was based on a scenario regarded as less familiar to participants. That the scenarios were actually more or less familiar to participants was later confirmed by their self-reported ratings of familiarity on a topic familiarity questionnaire, described in the corresponding section below. Each passage conformed to a temporally ordered set of activities pertinent to the specific scenarios, where certain actions typically precede others. There were also particular roles and objects associated with the scenarios. Each story loosely conformed to a script purported to be stored in long-term memory (Graesser et al. 1994; Schank and Abelson 1977), if participants were familiar with the scenario. The contents of each story were constructed via the script norm procedure where an independent sample of 24 American university undergraduates wrote at least 20 actions and their order of occurrence corresponding to each scenario. The stories were constructed around the most frequently occurring actions and their order of occurrence. Both stories were of comparable length, however the Grocery story contained more semantic propositions (i.e. units of meaning, see Kintsch 1998) than the Publishing story (see Table 1). In terms of structural difficulty, the Publishing story contained one more que clause and more object pronouns than the Grocery story due to more personal interaction involved in the Publishing scenario compared to the Grocery scenario. DIANA PULIDO 73 Table 1: Comparison of stories Length (words) ASL Que clauses Object pronouns Reflexive pronouns Semantic propositions More familiar Grocery Less familiar Publishing 174 10.89 1 3 1 58 164 10.25 2 6 1 45 Note: ASL ¼ average sentence length. Target Words Sixteen lexical items, representing concepts frequently associated with the scenarios, were chosen evenly from among the two stories: eight per story (see Appendix B). They were then substituted with nonsense words to ensure that nobody had prior knowledge of the TWs. The nonsense words were invented words constructed according to the orthographic and morphological rules of Spanish. All derivational and inflectional morphemes were maintained. A ratio of nouns to verbs was kept constant from one story to the next. Each TW appeared only once in each story, and there were no definitional context clues provided for any of them. Independent Variables Topic familiarity was assessed by a 10-item Likert-scale questionnaire (see Appendix C: available online at http://applij.oxfordjournals.org). Participants rated on a scale of 1–5 their level of familiarity with specific routine activities and scenarios, the two portrayed in the passages and an additional eight distracters. Response options tapped participants’ selfreported knowledge of the activities involved in the scenarios and their order of occurrence. This task furnished information about participants’ background knowledge and confirmed the prior categorization of the passages. The results indicated that, overall, the Grocery scenario was more familiar than the Publishing scenario (see Table 2). To assess passage sight vocabulary of the non-TWs from each passage, a single measurement was administered, an open-ended translation (L2–L1) of the words that participants reported as being familiar (see Appendix D: available online at http://applij.oxfordjournals.org). All translations were scored as follows: 0 ¼ incorrect; .5 ¼ partially correct; 1 ¼ correct. A word designated as unfamiliar received a score of 0. Since it was not feasible to test every lexical item from each passage, I first counted the amount of unique 74 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY Table 2: Response frequency on topic familiarity questionnaire More familiar Grocery Less familiar Publishing Rating f % f % 1 2 3 4 5 0 0 1 1 33 0 0 3 3 94 14 12 9 0 0 40 34 26 0 0 Note: N ¼ 35 participants. Ratings on a 5-point scale. lexical items and multi-word units from each story. From this figure I estimated which function words and cognates were likely to be known by all learners, and excluded them (e.g. el, la, para, en, supermercado, publicación). The remaining 104 words were then randomized and included on one test. The total amount of words tested from each story and corresponding percentage of text coverage of all unique lexical items was as follows: more familiar Grocery story—64 words tested (77 per cent coverage); and less familiar Publishing story—59 words (81 per cent coverage).2 Dependent variables To measure lexical inferencing, participants were assigned the task of writing the meaning or translation of each underlined and boldfaced TW in the L1, English. Participants received the following instructions: ‘You are going to read 2 very brief stories in Spanish. Contained within each story are 8 bold-face underlined words (e.g. reading). Your first task is to guess the TRANSLATION of each of these 8 words based upon the context. Clearly print the English translation in the space ABOVE each of the boldface underlined words. If you cannot think of an exact translation, give an approximate one, or describe what you think the word means.’ All participants were first oriented to the reading and inferencing task by completing a practice exercise to guess the meaning of an unfamiliar boldfaced and underlined word within a sentence. Lexical inferencing was scored as follows: 0 ¼ incorrect;.5 ¼ partially correct; 1 ¼ correct (either the correct L1 translation, or a correct paraphrase or definition).3 Level of ease/difficulty in lexical inferencing was determined by responses on a Likert-scale questionnaire tapping level of difficulty in inferring the meaning of each TW. The following scale was used: 1 ¼ very difficult; 2 ¼ difficult; 3 ¼ moderately difficult; 4 ¼ moderately easy; 5 ¼ easy; 6 ¼ very easy. During the orienting task participants were also instructed DIANA PULIDO 75 to rate guessing ease/difficulty with the following instructions: ‘Immediately after guessing the translation of each word, please indicate on a scale of 1–6 the degree of ease or difficulty in guessing the translation of the bold-face underlined word.’ During the practice exercise participants also completed a difficulty rating to verify that they understood the task and scale. Two measures assessed lexical retention. Both contained all 16 TWs plus 6 additional nonsense distracters. The verbs were presented in their infinitival forms, and the nouns in their original text forms. To diminish effects of memory for order of presentation of the TWs within the stories, or of a common theme, the test items were first randomized. Two test formats were created for each measure by reversing the order of presentation of the items. A translation production test required participants to supply a translation, definition, or explanation of the nonsense words into their L1. Retention was scored as follows: 0 ¼ incorrect; .5 ¼ partially correct; 1 ¼ correct. A multiple-choice translation recognition test required participants to select the translation from a series of options emphasizing semantic, not syntactic, differences. Options included the correct translation of the TW, three distracters, and a fifth option, ‘I don’t know’. Each distracter met at least one of the following criteria: (a) contextually proximate to the TW; (b) schematically appropriate; and (c) orthographically or phonologically close to another word in the L1 or L2 and plausible for the context. There were no conceptually bizarre options. Procedure A repeated measures design was used. All participants read passages from each condition (i.e. one more and one less familiar passage). During session 1, in small groups, all participants completed the L2 passage sight vocabulary test, followed by the topic familiarity questionnaire. Session 2 was conducted approximately 1 week later in a laboratory, on an individual basis. After the training phase, each participant read the first story and completed the accompanying lexical inferencing and difficulty rating tasks before continuing to the next story. Order of presentation of the passages was counterbalanced. After reading both stories, all participants completed the same self-paced online TW verification task to promote elaborative processing of the TWs in an error-free manner by allowing participants to confirm or correct the guesses and in so doing make connections between the new word form, its meaning, and the context in which it had appeared in the story. Each TW sentence was presented individually on a computer screen, in the same order of presentation as in the story. The TW translation appeared in the right-hand margin. Participants were instructed to confirm or correct their guesses about each word and to be sure they understood the sentence before continuing to subsequent sentences. They were told to anticipate questions about the sentences, but were not explicitly alerted to the 76 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY vocabulary tests that followed in order to encourage integration of the TW meaning into the sentence and context, as opposed to looking only at the gloss or engaging in memorization routines. Afterwards, participants completed the lexical retention tests in the following order: L2–L1 translation production, L2–L1 multiple-choice translation recognition. Session 2 concluded with a debriefing phase to explain the purpose of the experiment and use of nonsense words. RESULTS One-way analyses of covariance (ANCOVAs) were conducted to answer the research questions. Topic familiarity was the within-subjects categorical variable with two levels, more familiar and less familiar. Passage sight vocabulary was the continuous between subjects variable (covariate) consisting of each participant’s sight vocabulary score.4 The impact of the within and between subject independent variables was tested with omnibus F-tests, observing Type III sum of squares. Because 4 ANCOVAs and 2 t-tests were completed, a Bonferroni adjustment required an alpha level of .0083 (.05/6) to be used for all statistical tests reported below. There was considerable variability across the participants of the group for the independent variable of passage sight vocabulary (combined average of both stories M ¼ .65, SD ¼ .16; more familiar M ¼ .60, SD ¼ .16; less familiar M ¼ .70, SD ¼ .16). The percentages also reflect to a minor extent any partial knowledge about the passage sight vocabulary that was tested.5 A paired t-test revealed higher scores from the less familiar story relative to the more familiar story (t (34) ¼
11.13, p 5 .0001). However, due to the high Pearson correlation between the passage sight vocabulary scores for the more and less familiar passage (r ¼ .944, p 5 .0001) I calculated the average of the two scores, and utilized the new score as the passage sight vocabulary continuous covariate. For research question 1, addressing the impact of the reader-based factors on lexical inferencing, the combined scores revealed that approximately half of the TWs were successfully inferred (M ¼ 4.51, SD ¼ 1.54, maximum per story ¼ 8). There were also nearly twice as many correct inferences from the more familiar passage (M ¼ 5.86, SD ¼ 2.10) compared to the less familiar passage (M ¼ 3.26, SD ¼ 1.40). In some cases all TWs from the more familiar passage were successfully guessed. There were strong Pearson correlations between the passage sight vocabulary and lexical inferencing scores (combined r ¼ .72, p 5 .0001). As sight vocabulary increased, so did TW inferencing ability. The correlation was slightly higher for the more familiar condition (r ¼ .67, p 5 .0001) compared to the less familiar condition (r ¼ .53, p 5 .01), suggesting a more robust contribution to inferencing when readers had the appropriate background knowledge. Results from the one-way ANCOVA revealed a consistent significant main effect for the within subject variable of topic familiarity (F (1, 33) ¼ 72.18, DIANA PULIDO 77 p 5 .0001, Z2 ¼ .69), and the between subject covariate of passage sight vocabulary (F (1, 33) ¼ 35.53, p 5 .0001, Z2 ¼ .52) on lexical inferencing. The interaction between passage sight vocabulary and topic familiarity was not significant (F (1, 33) ¼ 3.89, p ¼ .057, Z2 ¼ .12), thus confirming consistent effects of the variables on lexical inferencing. For the second question, concerning difficulty in lexical inferencing, on average participants reported moderate difficulty (combined M ¼ 2.72, SD ¼ 1.04, scale of 1–6). The TWs from the more familiar passage were considered easier to guess, and there was slightly more variability in ratings in this condition (more familiar M ¼ 2.99, SD ¼ 1.35; less familiar M ¼ 2.45, SD ¼ .97). The Pearson correlations between the passage sight vocabulary and lexical inferencing difficulty scores illustrated that perceived ease in lexical inferencing was directly proportionate to passage sight vocabulary knowledge (combined r ¼ .54, p 5 .01). However, there were differences between the two correlations (more familiar r ¼ .63, p 5 .0001; less familiar r ¼ .30, p ¼ n.s.) suggesting that regardless of degree of passage sight vocabulary knowledge, TWs within the less familiar topic were more difficult to guess. A comparison of the correlations revealed significant differences (t (32) ¼ 2.61, p ¼ .007). Results from the one-way ANCOVA revealed a significant effect of the variable of topic familiarity (F (1, 33) ¼ 11.10, p ¼ .002, Z2 ¼ .25), and also of the covariate of passage sight vocabulary (F (1, 33) ¼ 13.61, p ¼ .001, Z2 ¼ .29) on difficulty in lexical inferencing. However, these effects are qualified by a significant two-way interaction between topic familiarity and passage sight vocabulary (F (1, 33) ¼ 11.03, p ¼ .002, Z2 ¼ .25; more familiar condition, B ¼ 5.35, t ¼ 4.50, p 5 .0001, Z2 ¼ .38; less familiar condition, B ¼ 1.88, t ¼ 1.83, p ¼ .076, Z2 ¼ .09).6 The significant interaction revealed that the slopes of the two regression lines were significantly different, suggesting that the role of passage sight vocabulary in the perception of ease/difficulty in lexical inferencing depended upon the degree of familiarity with the topic (see Figure 1). To interpret the significant interaction it was necessary to identify regions where the two regression lines were significantly different for all possible points. The point of intersection of the two regression lines was at .50 on the horizontal axis, representing the range of passage sight vocabulary scores (see Figure 1). To determine at what value of passage sight vocabulary beyond 50 per cent the TWs from the more familiar scenario were significantly easier to guess compared to those of the less familiar scenario Potthoff’s extension of the Johnson–Neyman test was used. This analysis allows one to obtain cutoff values of the predictor variable (passage sight vocabulary), above and below the point of intersection, where significant differences are predicted to occur.7 Results revealed that when participants scored at least 67 per cent on the passage sight vocabulary measure (i.e. knowledge of approximately twothirds of the passage vocabulary) new TWs from the more familiar passage were significantly easier to guess than those from the less familiar passage. 78 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY For the third research question, concerning lexical retention, on average scores were better on the multiple-choice recognition measure (combined M ¼ 6.13, SD ¼ 1.45, maximum per story ¼ 8) compared to the production measure (combined M ¼ 3.83, SD ¼ 2.09, maximum possible ¼ 8) after the inference verification task. Scores were slightly higher for the TWs from the more familiar story compared to the less familiar story for both measures (production more familiar M ¼ 4.23, SD ¼ 2.39; less familiar M ¼ 3.43, SD ¼ 2.29; and recognition more familiar M ¼ 6.66, SD ¼ 1.47; less familiar M ¼ 5.60, SD ¼ 2.03). The Pearson correlations between the scores for passage sight vocabulary and the translation production and recognition measures are both positive, revealing that as passage sight vocabulary improved, so did memory for the L1 translation equivalents (production r ¼ .51, p 5 .01; recognition r ¼ .52, p 5 .01). However there are notable differences in the magnitudes across the different familiarity conditions. For both measures, in the less familiar condition there were moderate correlations between passage sight vocabulary and retention (production r ¼ .53, p 5 .01; recognition r ¼ .51, p 5 .01), whereas in the more familiar condition the correlations were considerably weaker (production r ¼ .35, p 5 .05; recognition r ¼ .27, p ¼ n.s.). A comparison of the correlations yielded no significant differences. Nonetheless, these patterns are the opposite from those observed for the 6 Average difficulty/ease rating 5 4 3 2 Less familiar rsq = 0.0924 1 0 0.0 More familiar rsq = 0.3801 0.2 0.4 0.6 0.8 1.0 Passage sight vocabulary (%) Figure 1: Significant interaction between topic familiarity and passage sight vocabulary for dependent variable of lexical inferencing difficulty DIANA PULIDO 79 lexical inferencing and difficulty dependent variables, where correlations were stronger in association with the more familiar story scenario. Results from the one-way ANCOVA for the translation production measure revealed a significant effect only for the covariate of passage sight vocabulary (F (1, 33) ¼ 11.46, p ¼ .002, Z2 ¼ .26). The within subject variable of topic familiarity did not reach the pre-established level of significance (F (1, 33) ¼ 4.99, p ¼ .032, Z2 ¼ .13). The interaction was not significant (F (1, 33) ¼ .77, p ¼ .39, Z2 ¼ .02). Results from the one-way ANCOVA for the translation recognition measure revealed a significant effect for the variables of topic familiarity (F (1, 33) ¼ 9.88, p ¼ .004, Z2 ¼ .23) and passage sight vocabulary (F (1, 33) ¼ 12.32, p ¼ .001, Z2 ¼ .27). The interaction was not significant (F (1, 33) ¼ 2.84, p ¼ .10, Z2 ¼ .08). These results indicate that on the recognition measure, more TWs were remembered from the more familiar story compared to the less familiar story, regardless of how much passage sight vocabulary learners knew. In addition, both analyses illustrate a similarly robust role of passage sight vocabulary on retention. DISCUSSION To determine the impact of two variables (passage sight vocabulary and topic familiarity) on lexical inferencing (success and ease of processing) and on retention of specific TWs that were encountered in two brief narrative passages, the following discussion focuses first on the impact of topic familiarity and then on that of passage vocabulary knowledge. Both are discussed in terms of the accompanying strategic tasks of lexical inferencing and difficulty rating during reading, followed by an online verification task to confirm or revise initial hypotheses. Topic familiarity had a significant impact on lexical inferencing. There were substantially more correct TW inferences when reading about a more familiar scenario compared to a less familiar scenario. Having appropriate background knowledge may have helped learners to more efficiently direct attention to input while reading the more familiar story. The local and global context cues that activated the readers’ syntactic, semantic, and pragmatic knowledge in constructing the meaning of the text were held in working memory and must have constrained ongoing textual interpretations, and in turn, form–meaning mappings for new words. These results expand upon previous research reported above that observed learners using background knowledge during think-aloud tasks in lexical inferencing research paradigms, or that studied effects of topic familiarity and semantic richness on discrete sentence processing. The results also increase our understanding of the strong role of background knowledge in vocabulary development that begins during reading by studying learners’ initial meaning assignments to new words. This broadens the conclusions from my previous studies (e.g. Pulido, 2000, 2003, 2004b) because in those studies only post-reading tasks were administered. In sum, from a cross-sectional perspective, the 80 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY results of the first question illustrated robust findings of topic familiarity on inferencing, regardless of how much passage sight vocabulary was known. However, topic familiarity had inconsistent effects on perceived ease or difficulty in guessing. For readers of the group with passage sight vocabulary scores at or beyond 67 per cent (i.e. 18 of the participants, including 2 graduate, 9 advanced, 3 high-intermediate, and 4 intermediate) the TWs from the more familiar passage were easier to guess, and were also guessed correctly more often. These readers likely experienced more success in constructing a textbase and situation model for this passage. In spite of knowing ample vocabulary from the less familiar, Publishing, passage, such knowledge did not compensate for lack of familiarity with the topic, insofar as ease and success of guessing was concerned. For the readers of the group with weaker passage sight vocabularies, they experienced equivalent degrees of difficulty in inferencing, regardless of the degree of familiarity with the topics. These readers were expected to have fewer processing resources given the predicted difficulties in word recognition. With incomplete and inaccurate access to meaning they also had less available context to support the use of lexical inferencing strategies. This finding refines our understanding of lexical inferencing and strategic processing during reading, going beyond anecdotal observations of beginning and intermediate learners’ frustration and ‘short-circuiting’ when encountering unfamiliar words during reading (e.g. Haynes 1993; Haynes and Baker 1993; Lee and Wolf 1997; Parry 1997; Rott 2000). Topic familiarity also had inconsistent effects on lexical retention after the inference verification task to promote revision, integration, and retention. Significant effects were only obtained on the translation recognition measure (a similar finding observed in Pulido 2000, 2003, where similar measures were used). First, there were on average fewer words to revise and learn from the more familiar condition compared to the less familiar condition, which could have freed up attentional resources during the verification task to better enable some learners to concentrate on correcting the words that were incorrectly guessed. Second, the translation recognition test itself contained retrieval cues (i.e. the L1 translation options) which likely facilitated access to rich episodic memory traces about the local or global story context that had been constructed for the more familiar story during the previous input processing cycles. These two combined factors may have contributed to superior memory for the TWs from the more familiar condition on the translation recognition measure. As for the translation production measure, although retention from the more familiar story was greater, there were no significant differences obtained between the two conditions. A similar finding was also obtained in Pulido (2000, 2003), which was conducted under different experimental conditions (i.e. participants read twice as many passages, did not complete lexical inferencing or verification tasks, and took retention measures 2 days after reading). However, there are some salient differences. In the previous DIANA PULIDO 81 study there was only 3 per cent correct retention on the translation production measure from both familiarity conditions, whereas in the present study there was on average 48 per cent, and a much wider gap in retention between the more and the less familiar conditions (53 per cent and 43 per cent, respectively). Still, with only the TW and no available cues, topic familiarity does not have as robust an effect on memory for meaning as tested in this manner. This conclusion is similar to that drawn in Pulido (2004b) concerning the effects of cultural familiarity on immediate retention. That study, using an adapted version of the VKS, found that topic familiarity only had an effect in terms of episodic memory, that is memory for having noticed the new words in certain stories. With regard to retention there were test effects: higher retention on the translation recognition measure compared to the translation production measure. This is understandable in light of transfer-appropriate processing (TAP) effects on human memory (Morris et al. 1977) proposing that memory improves when there is more similarity between the nature of the task at study and at testing. The inference verification task depicted the TW in the original sentence and its L1 translation in the margin, which was somewhat similar to the translation recognition task, which depicted the TW (in isolation) and L1 translation options. Finally, there were lower effect sizes obtained for the variable of topic familiarity over the duration of the experiment and strategic tasks. This may be attributed to other factors related to retention outcomes, such as repeated exposure to the TW sentences and TWs themselves, and the presence of glosses during the verification task. Passage sight vocabulary also played a robust role in the process of making form–meaning connections for new words during the reading tasks. The results consistently demonstrated that as passage sight vocabulary increased so did ability to correctly infer TW meanings. Recognizing words entails accessing meaning. When lexical access is laborious and/or inaccurate this strains the processing resources also needed for syntactic parsing, propositional generation, retention of information in working memory, and access to information in long-term memory. When learners know more vocabulary in a passage they also have more available context and clues from which to interpret specific relationships among ideas, and any new vocabulary contained therein. These findings expand upon anecdotal observations about the role of passage sight vocabulary during lexical inferencing (e.g. Haynes 1993; Haynes and Baker 1993; Lee and Wolf 1997; Rott 2000; Schouten-van Parreren 1989) by using an independent measure representing a range of vocabulary specifically connected with the texts that were read. Passage sight vocabulary also consistently predicted lexical retention, although the effect sizes were not as strong as on the lexical inferencing measure, a finding explained in the previous section. As stated above, those with greater passage sight vocabularies were expected to have had more available processing resources to engage in the verification task. Thus, to the extent that learners were efficient in word recognition, one would 82 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY expect correspondingly richer interpretations and more efficient cognitive involvement as they reprocessed the TWs and meanings, and reconstructed the textbase and situation model. This may have contributed to facilitating recall of the word meanings on the retention tasks. Although there were directly proportionate relationships between passage sight vocabulary and retention, there were differences in the strengths of the correlations across the different conditions. Such differences are due to the restriction in range of retention scores from the more familiar passage, perhaps a result of several factors, such as fewer words to learn, greater ease and success in learning and remembering the TWs across the different tasks, or even forgetting by some learners. Future studies will want to address the specific relationship between inferencing and retention, and learner involvement and attentional allocation during strategic tasks in order to furnish a more precise explanation. That passage sight vocabulary was more relevant to retention from the less familiar passage may be due to greater difficulty experienced by the weaker learners of the group in engaging in the elaborative processes required for word learning, namely, assigning initial meaning, making comparisons and revising the meanings in light of the TW translations, relating information to previous knowledge, and storing the new information. The stronger learners of the group, with larger sight vocabularies and fewer words to learn, likely had more attentional resources to successfully engage in the processes described above. In Pulido (2000, 2003) similar overall effects of passage sight vocabulary were obtained on the lexical retention measures, but there were also some differences. In that study the correlations were slightly greater for the more familiar conditions compared to the less familiar conditions. However, in the present study the opposite pattern was obtained, and with pronounced differences in magnitude. Second, in the earlier study there were stronger correlations on the translation recognition measure compared to the production measure, a pattern opposite from that of the current study. As suggested above, the characteristics of the strategic tasks, in combination with learner variables, likely facilitated more word learning from the more familiar condition, resulting in less variability compared to the less familiar condition. These results concerning passage sight vocabulary contribute to lexical threshold theories and the importance of vocabulary knowledge in reading (e.g. Grabe and Stoller 2002; Hazenberg and Hulstijn 1996; Horst et al. 1998; Koda 2005; Laufer 1997). In this study, readers with a wide range of passage sight vocabulary were able to correctly infer and retain some degree of semantic knowledge about the new words. Still, greater success was observed by those with stronger sight vocabularies, providing further support for the Matthew effects in language learning, described by Stanovich (1986) for L1 reading, and by Ellis (1994) for L2 vocabulary acquisition. These results also shed a new light on the impact of this variable on other stages involved in lexical development through reading (i.e. inferencing) as they are observed in connection with strategic tasks. DIANA PULIDO 83 The combined results of the effects of both independent variables also contribute to the Involvement Load theory (Laufer and Hulstijn 2001) for various stages of lexical development through reading tasks. They suggest that cognitive involvement (i.e. ‘search’—the attempt to find the meaning of the new word, and ‘evaluation’—decision-making to determine whether a word does or does not fit its context) in lexical inferencing and processing tasks is also affected by the interaction between reader-based variables, such as background knowledge and passage sight vocabulary. There are noteworthy pedagogical implications. First, teachers should consider whether the topic and vocabulary in the context are familiar to learners when assigning reading and inferencing tasks. If they are not, then learners may benefit from instruction to promote passage sight vocabulary development and to activate relevant background knowledge. Next, since weaker learners experience difficulty and little success in lexical inferencing, providing a verification task to encourage deeper processing of the TWs may provide additional benefits beyond a mere reading-inferring task, especially if learners are familiar with the topic (e.g. see Fraser 1999; Mondria 2003; Schouten-van Parreren 1989). Last, instructors should consider that one form of assessment compared to another (e.g. recognition vs. production) might provide greater opportunities for learners to demonstrate vocabulary gains. There are some limitations that could be addressed by future studies. It is possible that other intervening factors (e.g. those related to local textual cues, such as syntactic structure) affected outcomes. Also, lengthier texts with more TWs could be used to enhance the generalizability of the findings. This study did not systematically address the effectiveness of the online strategic verification task itself, compared to an inferring-only method. Studies could compare task effectiveness, in combination with how performance is affected by the interactions between various reader-based factors. Future research could also consider the nature of processing at each stage and determine the relationships between lexical inferencing, difficulty in inferencing, inference verification, episodic memory, retention, and strategy use. Finally, this study considered only two reader-based factors. Other reader-based factors would provide a more robust picture of lexical inferencing, processing, and retention during strategic reading tasks (e.g. reading proficiency, working memory, and syntactic knowledge). In sum, the present study sheds a new light on SLA research concerned with individual differences in lexical development through strategic reading tasks. An attempt was made to operationalize and measure variables that have received little attention, via a repeated-measures cross-sectional design. The results suggest the complex nature of the impact and interaction between passage sight vocabulary and background knowledge during lexical input processing, both of which work in tandem with other processes during reading. Final version accepted November 2005 84 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY APPENDIX B: TARGET WORDS More and less familiar passages Nouns Verbs ‘The Trip to the Supermarket’ ‘Publishing an Article’ las borrigas—aisles el resmo—grocery cart la olneda—checkout las pifantas—tabloids el lastredón—checkout counter las afeltas—bags dacar—to ring up sapirar—to compare las dostas—guidelines el tı́dero—letter of receipt los toleneros—reviewers las mascuas—proofs la segua—cover letter las acriciones—stipulations betenar—to delete regamir—to resubmit NOTES 1 Learners are told to guess meanings of unfamiliar words, often through a think-aloud task while reading. Typically, there is no test to measure retention of the words. 2 There were 19 words appearing in both stories and which were included only once on the passage sight vocabulary test. 3 For the TW concept proofs (mascuas) in the following context (translated into English), Two months before publication Sue received the proofs. She had three days to make the changes. She also had to sign the copyright agreement, incorrect responses included ‘book’, ‘royalties’, partially correct responses were ‘revisions’ and ‘drafts’, and correct responses were ‘final draft before its publication’, and ‘final copy before printing’. 4 I was also interested in determining if topic familiarity moderated the purportedly strong relationship between passage sight vocabulary and lexical input processing. This was accomplished by investigating the interaction between the independent variables. 5 All participants demonstrated partial knowledge of at least 1, but not more than 10, of the 104 tested passage sight vocabulary items. The mean percentage of words partially known was .05, SD ¼ .02. 6 B represents the parameter estimate slopes of the regression lines from the ANCOVA model. 7 The procedure to determine the cutoff values involves three steps: (a) running two separate regression analyses using the values of the continuous predictor variable and outcome variable for the two different familiarity conditions; (b) calculating three intermediate quantities using three different equations that require values obtained from the two separate regression analyses, such as F2, N–4 (the tabled F value), N, total SSresidual, SS associated with the predictor variables for each of the familiarity conditions, M of the predictor variables for each condition, and the slopes and intercepts obtained in each of the regression analyses; and (c) calculating the cutoff regions using two additional equations that contain the three intermediate quantities derived in (b) above (see Aiken and West 1991: 134–6 for equations). DIANA PULIDO 85 REFERENCES Adams, S. 1982. ‘Scripts and the recognition of unfamiliar vocabulary: Enhancing second language reading skills,’ Modern Language Journal 66: 155–9. Aiken, L. and S. West. 1991. Multiple Regression: Testing and Interpreting Interactions. Newbury Park, CA: Sage. Baddeley, A. 1998. Human Memory: Theory and Practice. Needham Heights, MA: Allyn & Bacon. Chern, C. 1993. ‘Chinese students’ word-solving strategies in reading in English’ in T. Huckin, M. Haynes, and J. Coady (eds): Second Language Reading and Vocabulary Learning, Norwood, NJ: Ablex, pp. 67–82. Clark, E. 1993. The Lexicon in Acquisition. Cambridge: Cambridge University Press. Craik, F. and E. Tulving. 1975. ‘Depth of processing and the retention of words in episodic memory,’ Journal of Experimental Psychology: General 104: 268–94. De Bot, K., T. Paribakht, and M. Wesche. 1997. ‘Toward a lexical processing model for the study of second language vocabulary acquisition,’ Studies in Second Language Acquisition 19: 309–29. Ellis, N. 1994. ‘Vocabulary acquisition: The implicit ins and outs of explicit cognitive mediation’ in N. Ellis (ed.): Implicit and Explicit Learning of Languages, London: Academic Press, pp. 211–82. Ellis, N. 2001. ‘Memory for language’ in P. Robinson (ed.): Cognition and Second Language Instruction, Cambridge: Cambridge University Press, pp. 33–68. Faerch, C., K. Haastrup, and R. Phillipson. 1984. Learner Language and Language Learning. Copenhagen: Clevedon Multilingual Matters. Fraser, C. 1999. ‘Lexical processing strategy use and vocabulary learning through reading,’ Studies in Second Language Acquisition 21: 225–41. Gass, S. 1999. ‘Incidental vocabulary learning,’ Studies in Second Language Acquisition 21: 319–33. Grabe. W. and F. Stoller. 2002. Teaching and Researching Reading. Essex: Longman. Graesser, A., M. Singer, and T. Trabasso. 1994. ‘Constructing inferences during narrative text comprehension,’ Psychological Review 101: 371–95. Haastrup, K. 1989. ‘The learner as word processor,’ AILA Review 6: 34–46. Haynes, M. 1993. ‘Patterns and perils of guessing in second language reading’ in T. Huckin, M. Haynes, and J. Coady (eds): Second Language Reading and Vocabulary Learning, Norwood, NJ: Ablex, pp. 46–64. Haynes, M. and I. Baker. 1993. ‘American and Chinese readers learning from lexical familiarizations in English text’ in T. Huckin, M. Haynes, and J. Coady (eds): Second Language Reading and Vocabulary Learning, Norwood, NJ: Ablex, pp. 130–50. Hazenberg, S. and J. Hulstijn. 1996. ‘Defining a minimal receptive second-language vocabulary for non-native university students: An empirical investigation,’ Applied Linguistics 17: 145–63. Holley, F. 1973. ‘A study of vocabulary learning in context: The effect of new-word density in German reading materials,’ Foreign Language Annals 6: 339–47. Horst, M., T. Cobb, and P. Meara. 1998. ‘Beyond A Clockwork Orange: Acquiring second language vocabulary through reading,’ Reading in a Foreign Language 11: 207–23. Hulstijn, J. 2001. ‘Intentional and incidental second language vocabulary learning: A reappraisal of elaboration, rehearsal and automaticity’ in P. Robinson (ed.): Cognition and Second Language Instruction, Cambridge: Cambridge University Press, pp. 258–86. Hulstijn, J. 2003. ‘Incidental and intentional learning’ in C. Doughty and M. Long (eds): Handbook of Second Language Acquisition, Oxford, UK: Blackwell, pp. 349–81. Just, M. and P. Carpenter. 1992. ‘A capacity theory of comprehension: Individual differences in working memory,’ Psychological Review 99: 122–49. Kintsch, W. 1998. Comprehension: A Paradigm for Cognition. Cambridge: Cambridge University Press. Koda, K. 2005. Insights into Second Language Reading: A Cross-linguistic Approach. Cambridge: Cambridge University Press. Laufer, B. 1997. ‘The lexical plight in second language reading; Words you don’t know, words you think you know, and words you can’t guess’ in J. Coady and T. Huckin (eds): Second Language Vocabulary Acquisition: A Rationale for Pedagogy, Cambridge: Cambridge University Press, pp. 20–33. Laufer, B. and J. Hulstijn. 2001. ‘Incidental vocabulary acquisition in a second language: The construct of task-induced involvement,’ Applied Linguistics 22/1: 1–26. 86 TOPIC FAMILIARITY AND PASSAGE SIGHT VOCABULARY Lee, J. F. and D. Wolf. 1997. ‘A quantitative and qualitative analysis of the word-meaning inferencing strategies of L1 and L2 readers,’ Spanish Applied Linguistics 1: 24–64. Li, X. 1988. ‘Effects of contextual cues on inferring and remembering meanings of new words,’ Applied Linguistics 9/4: 402–13. McLaughlin, B. 1987. Theories of Second Language Learning. London: Arnold. Mondria, J. 2003. ‘The effects of inferring, verifying, and memorizing on the retention of L2 word meanings: An experimental comparison of the ‘‘meaning-inferred method’’ and the ‘‘meaning-given method’’,’ Studies in Second Language Acquisition 25: 473–99. Mondria, J. and M. Wit–de Boer. 1991. ‘The effects of contextual richness on the guessability and retention of words in a foreign language,’ Applied Linguistics 12/3:249–67. Morris, C., J. Bransford, and J. Franks. 1977. ‘Levels of processing versus transfer appropriate processing,’ Journal of Verbal Learning and Verbal Behavior 16: 519–33. Nagy, W. 1997. ‘On the role of context in first- and second-language vocabulary learning’ in N. Schmitt and M. McCarthy (eds): Vocabulary: Description, Acquisition and Pedagogy, Cambridge: Cambridge University Press, pp. 64–83. Nassaji, H. 2002. ‘Schema theory and knowledgebased processes in second language reading comprehension: A need for alternative perspectives,’ Language Learning 52/2: 439–81. Nassaji, H. 2003. ‘L2 vocabulary learning from context: Strategies, knowledge sources, and their relationship with success in L2 lexical inferencing,’ TESOL Quarterly 37/4:645–70. Nassaji, H. 2004. ‘The relationship between depth of vocabulary knowledge and L2 learners’ lexical inferencing strategy use and success,’ The Canadian Modern Language Review 61/1: 107–34. Paribakht, T. and M. Wesche. 1999. ‘Reading and ‘‘incidental’’ L2 vocabulary acquisition. An introspective study of lexical inferencing,’ Studies in Second Language Acquisition 21: 195–224. Pulido, D. 2000. The impact of topic familiarity, L2 reading proficiency, and L2 passage sight vocabulary on incidental vocabulary gain through reading for adult learners of Spanish as a foreign language. Unpublished doctoral dissertation, University of Illinois at UrbanaChampaign. Pulido, D. 2003. ‘Modeling the role of second language proficiency and topic familiarity in second language incidental vocabulary acquisition through reading,’ Language Learning, 53/2: 233–84. Pulido, D. 2004a. ‘The relationship between text comprehension and second language incidental vocabulary acquisition: A matter of topic familiarity?’ Language Learning, 54/3: 469–523. Pulido, D. 2004b. ‘The effect of cultural familiarity on incidental vocabulary acquisition through reading,’ The Reading Matrix: An Online International Journal, 4/2: 20–53. Robinson, P. 2003. ‘Attention and memory’ in C. Doughty and M. Long (eds): Handbook of Second Language Acquisition, Oxford, UK: Blackwell. pp. 631–78. Rott, S. 2000. ‘Relationships between the process of reading, word inferencing, and incidental word acquisition’ in J. F. Lee and A. Valdman (eds): Form and Meaning: Multiple Perspectives, Boston, MA: Heinle & Heinle, pp. 255–82. Rumelhart, D. 1980. ‘Schemata: The building blocks of cognition’ in R. Spiro, B. Bruce, and W. Brewer (eds): Theoretical Issues in Reading Comprehension, Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 33–58. Schank, R. and R. Abelson. 1977. Scripts, Plans, Goals, and Understanding. Hillsdale, NJ: Lawrence Erlbaum Associates. Schmidt, R. 2001. ‘Attention’ in P. Robinson (ed.): Cognition and Second Language Instruction, Cambridge: Cambridge University Press, pp. 3–32. Schouten-van Parreren, C. 1989. ‘Vocabulary learning through reading: Which conditions should be met when presenting words in texts?’ AILA Review 6: 75–85. Stanovich, K. 1986. ‘Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy,’ Reading Research Quarterly 21: 360–400. Sternberg, R. 1987. ‘Most vocabulary is learned from context’ in M. McKeown and M Curtis (eds): The Nature of Vocabulary Acquisition, Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 89–105.