Language Learning

Language Learning

ISSN 0023-8333

The Effect of Frequency of InputEnhancements on Word Learning and Text Comprehension Susanne Rott University of Illinois at Chicago

Research on second language lexical development during reading has found positive effects for word frequency, the provision of glosses, and elaborative word processing. However, findings have been inconclusive regarding the effect of such intervention tasks on long-term retention. Likewise, few studies have looked at the cumulative effect of interventions on word learning or text comprehension. This investigation sought to assess the effect of increased frequency of target words (TWs) comparing lexical gain of words that occurred once (F1) or four times (F4) in the input passage. The study further investigated the combined effect of frequency (F4) and semantic or visual enhancements. It compared the following reading conditions: (a) TWs were glossed four times in the text (four-gloss: 4G); (b) TWs were first glossed, then retrieved in the first language, and bolded twice (gloss-retrieval: GR); and (c) TWs were first glossed and then bolded three times (gloss-bolding: GB). In addition, the study assessed the effect of these interventions on long-term retention (4–6 weeks) of lexical knowledge and on text comprehension. Findings revealed that the GR and 4G reading conditions resulted in more productive word gain than the GB condition or when readers encountered a TW only once. Repeated visual enhancements seemed to have no effect on strengthening word encoding. The comprehension of main ideas was highest when the TW was glossed four times followed by the gloss-bolding reading condition and the gloss-retrieval task. Keywords classroom research; lexical acquisition; text comprehension; input enhancement; frequency; lexical retention

Textual input is generally assumed to be an important source for second language (L2) lexical development because reading fosters L2 learners’ fluency in word recognition (e.g., Hulstijn, 2001), and texts provide readers with meaningbearing, semantically, syntactically, and pragmatically rich input (e.g., Krashen, Correspondence concerning this article should be addressed to Susanne Rott, Department of Germanic Studies 189, University of Illinois at Chicago, 601 S. Morgan, Chicago, IL 60607. Internet: [email protected] Language Learning 57:2, June 2007, pp. 165–199 ! C 2007 Language Learning Research Club, University of Michigan

165

Susanne Rott

The Effect of Frequency of Input-Enhancements

1993). Logically, L2 learners need to master such aspects of the lexicon to reach advanced and near-native-speakers’ language abilities (Judd, 1978; Nation, 2001; Richards, 1976). Given the undeniable advantage of reading for lexical development, the educationally relevant question remains: Which word interventions (a) increase the rate of word learning through reading, (b) foster target word retention, and (c) further or at least do not interfere with text comprehension? Research has repeatedly shown that encountering an unfamiliar word in a text or even comprehending it in its context might not lead to an initial assignment of meaning to the orthographical representation of the word (i.e., a lexical formmeaning connection [FMC], which is accessible after completing the reading task). Accordingly, it is well accepted that comprehension and learning are not the same phenomenon (e.g., Lee & VanPatten, 1995; Sharwood Smith, 1986). Whereas text comprehension aims at interpreting the message content, word learning aims at establishing FMCs to build a lexical system. In fact, reading a text for meaning (i.e., creating a mental representation of the text) requires the rapid integration of text-based and learner-based information. The reader constructs a textbase primarily via parsing or bottom-up processing of the textual input. The textbase consists of “those elements directly derived from the text itself. [. . .] In general, this procedure yields an impoverished and often even incoherent network” (Kintsch, 1998, p. 103). To obtain a coherent structure, the reader must interconnect the propositions in a network by complementing inferences and activating and integrating existing knowledge sources. Kintsch called the resulting reconstruction “the situation model of the text.” Encoding a new word in the mental lexicon, on the other hand, seems to require that readers first notice the gap in the mental lexicon. Next, readers need to isolate the word from its context (Prince, 1996) and allocate attentional resources to its orthographic and semantic properties (e.g., Schmidt, 2001). To consolidate the word in the mental lexicon, the reader must recognize the relationship between the lexical form and its meaning, which involves some form of elaboration by associating the word with existing knowledge sources or maintaining it in working memory (WM) for rehearsal (e.g., Ellis, 1994; Gass, 1997; Hulstijn, 2001). If no word meaning is directly provided with the text, readers have to experience the need to search for and infer meaning (e.g., Laufer & Hulstijn, 2001) by integrating semantic and syntactic aspects of the word using multiple strategic resources (Ellis). What becomes clear is that not only are the cognitive mechanisms involved in reading comprehension and lexical acquisition different, they might even be in conflict. In particular, if we follow the assumption that WM is a Language Learning 57:2, June 2007, pp. 165–199

166

Susanne Rott

The Effect of Frequency of Input-Enhancements

limited-capacity processing system (e.g., Barcroft, 2002; Ellis, 2001; VanPatten, 1996), learners can only attend to a limited amount of material at a time and naturally pay more attention to some aspects of language than to others. Therefore, constraints exist on the maximum amount of resources a L2 reader has available for comprehending a text and storing new words. In particular, this might be the case when texts are enhanced with word interventions that are favorable for promoting word gain. Readers’ attentional resources might be consumed by word encoding processes, not leaving sufficient resources for reconstructing the text. Likewise, it is possible that readers direct their attention to reconstructing propositions containing word interventions while passing over propositions without interventions (Rott & Williams, 2003). The goal of the present investigation was to shed more light on some of these issues by assessing and comparing the effect of three word intervention tasks on text comprehension and the learning and retention of vocabulary. Review of Research Cognitive Resources Many researchers agree with Schmidt’s (2001) position that the allocation of attentional resources is necessary for learning to take place. However, learners’ own attention-drawing mechanisms might not be sufficient. L2 readers might skip words because they do not perceive them as relevant for text comprehension (Hulstijn, 1993), thereby making word learning through reading rather unpredictable. Hence, lexical enhancements and intervention tasks, which guide the readers’ attention to specific words, increase the likelihood for word gain. Laufer and Hulstijn (2001) have proposed motivational (need) and cognitive (search and evaluation) factors that seem to be crucial for word learning. These factors are subsumed in the Involvement Load Hypothesis, which predicts that the higher the involvement load, the higher the possibility for incidental vocabulary learning. Learners reach the highest level of involvement when they (a) intrinsically perceive the need to learn rather than respond to an external stimulus, (b) search for rather than receive the meaning, and (c) make a decision as to how the new word will combine with additional words in an original sentence rather than simply recognize differences between words by choosing a correct meaning in a fill-in-the-blank task. This ideal combination of motivational and cognitive involvement might, however, not be pertinent for word learning during reading: First, L2 readers might not perceive the need to infer meaning to a particular word when they comprehend the context without explicitly assigning a meaning to an unfamiliar word (Parry, 1993, 1997). Second, 167

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

studies have repeatedly shown that when readers choose to search for meaning in the text, they often infer meanings incorrectly or are unable to guess at all (see review in Hulstijn, Hollander, & Greidanus, 1996). Finally, evaluating meaning usage in receptive and productive tasks has shown mixed results for learning. Readers who engaged in pedagogical activities after having read the text and who received the word meaning outperformed readers who received an unenhanced text (e.g., Laufer 2001; Wesche & Paribakht, 2000). Yet, output tasks were less effective for word learning when meaning was not provided explicitly and when readers engaged in output during reading (Rott & Williams, 2003; Rott, Williams, & Cameron, 2002). Additionally, prompting readers to engage in output tasks, such as text reconstruction, during reading seemed to interfere with text comprehension (Rott et al.). Word Intervention Through Glosses The Involvement Load Hypothesis predicts that providing readers with glosses might not be an efficient way to increase the rate of word learning because glosses only trigger the readers’ need extrinsically and there is no intrinsically motivated search or evaluation processes. Indeed, comparing the effect of reading a text with first language (L1) glosses with a composition task, Laufer and Hulstijn (2001) found that the composition group significantly outperformed the reading condition. However, the researchers were mainly concerned with word learning, rather than reading as a tool to simultaneously provide semantically and pragmatically rich lexical input and present content information. Other studies compared the effect of glosses with an unenhanced control condition. Watanabe (1997) found that readers in an L2 gloss condition gained significantly more words on the immediate as well as on the delayed posttest (2 weeks), as did Hulstijn et al. (1996) for L1 glosses. However, their delayed posttest took place after 1 hr. In turn, Ko (1995) observed that L1 glosses had a significant effect on the delayed posttest (2 weeks) but not on the immediate posttest. In contrast, Jacobs, Dufon, and Hong (1994) found a significant effect on an immediate posttest for L1 and L2 glosses. However, the effect disappeared 4 weeks later. These mixed results warrant future investigations comparing the effect of glosses with other input-based intervention tasks and developing further insights into the effect of glosses for long-term word gain. Word Enhancement Through Repeated Occurrences and Focused Attention In general, processing a new word repeatedly in one or multiple texts has been found to be conducive to incidental word learning (for a review, see Horst, Cobb, Language Learning 57:2, June 2007, pp. 165–199

168

Susanne Rott

The Effect of Frequency of Input-Enhancements

& Meara, 1998; Paribakht & Wesche, 1999; Rott, 1999; Waring & Takaki, 2003; Zahar, Cobb, & Spada, 2001). Studies found a relationship between frequency of exposure and acquisition, in particular for beginning learners (Zahar et al.), with sizable learning gains for words that appeared at least eight (Horst et al.) or more times in a text. Yet, even after encountering a word 15 times, Waring and Takaki found that word knowledge diminished drastically within 3 months, underscoring again the need for long-term studies. Ellis (e.g., 2004) explained why repeated encounters might fail to result in learning. Readers might not allocate sufficient attentional resources, might not isolate the word from the context, or might establish associations with the word during each encounter, thereby lacking processing mechanisms that strengthen and retain a form-meaning connection after it has been established. Therefore, instruction or enhanced materials might be crucial to direct L2 readers’ attention to words to increase the potential of processing. Even though focused attention to new words might not be as effective as for syntactic and morphosyntactic aspects of language, Gass, Svetics, and Lemelin (2003) found that guiding readers’ attention to new words, with an input and inferencing practice task, had a significantly stronger impact on word learning than a task that directed readers’ attention away from the target words (TWs). Furthermore, Hulstijn et al. (1996) manipulated attention and word frequency by comparing word learning of L2 readers whose text was enhanced with glosses and readers whose text was not enhanced. In both conditions, TWs occurred either one or three times. Findings revealed a significant cumulative effect for glosses and frequency on two tests that were administered within 1 hr. The study did not provide any information about the effect on long-term word retention. Text Comprehension and Word Interventions There is an inconclusive body of SLA research concerning (a) the relationship between word learning and text comprehension and (b) the effect of word interventions on text comprehension. In some studies, the texts were altered by increasing the occurrences of the TWs, others enhanced new words visually (e.g., bolding), yet others provided additional semantic clues for TWs (e.g., L1 or L2 glosses, appositives, dictionary access). In one study that manipulated the frequency of word occurrences, Rott (1997) found a moderate to strong positive correlation between text recall and immediate gain and retention of TWs; that is, participants who achieved greater levels of text comprehension were more likely to retain new words over an extended period of time. In a L1 study, Moravcsik and Healy (1998) measured the effect of visual word enhancements on text comprehension. They found 169

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

that placing words in boldface led to a worse performance on comprehension questions than highlighting all words or no words in a text. There was, however, only a trend with no significant negative effect of bolding individual words on a text summary task. The researchers did not provide an explanation for these findings. The effect of word interventions, which provided semantic clues about the TW meaning, also led to mixed results regarding the comprehension of the text. Based on open-ended questions, Watanabe (1997) found that only single glosses aided comprehension, whereas multiple-choice glosses and appositives did not. Similarly, Davis (1989) and Jacobs (1994) found a superior effect for glosses as compared to an unenhanced condition. Rott et al. (2002), on the other hand, found a significant increase in text recall for readers who received multiplechoice glosses as well as readers who simultaneously engaged in an input-output cycle task, compared to the control group, which read an unenhanced text. In contrast, Johnson (1982) and Pak (1986) did not find a positive effect of L1 glosses for text comprehension. Neither did Jacobs et al. (1994), who guided their readers’ attention to TWs by providing marginal L1 and L2 glosses. For both treatment conditions as well as the control group, which did not receive any enhancement, they found a significant correlation between text comprehension and word gain. However, unlike in the study by Rott (1997), they found no significant relationship 4 weeks later; that is, word retention was not related to text comprehension. The studies reviewed here revealed several issues that bear further investigation. First, concerning text comprehension, conclusive evidence is lacking for the effect of visually enhancing and simultaneously glossing words. One reason might be the inconsistency in measures used to assess text comprehension. Second, although there seems to be a strong relationship between word learning and text comprehension when L2 learners read unenhanced texts, the relationship between text comprehension and word retention remains unclear when texts are enhanced with glosses. Third, research to date has looked at word interventions separately, not paying attention to potential cumulative effects, particularly on long-term retention. Directing L2 readers’ attention to new words by providing glosses or increasing word occurrences in a text can improve the likelihood of establishing and strengthening lexical FMCs. Glosses ensure a correct word encoding in the mental lexicon, whereas increased occurrences seem to foster robustness. Fourth, the effect of frequency of word interventions has only been explored minimally. Enhancing a word in a text only once might not result in rehearsal of meaning or accessing existing knowledge sources and might

Language Learning 57:2, June 2007, pp. 165–199

170

Susanne Rott

The Effect of Frequency of Input-Enhancements

lead to activating only a partial set of learning mechanisms (e.g., rehearsal, activating existing knowledge sources). Repeated word enhancements increase the potential for establishing FMCs. Research Questions The current investigation sought to assess the combined effect of the frequency of word occurrences in the input passage and word intervention tasks: inputonly tasks designed to stimulate rehearsal and a word-retrieval task designed to stimulate access to knowledge sources. Additionally, the study examined the role of text comprehension in word learning. Thus, participants read texts that contained unfamiliar words, with the purpose of gaining an overall understanding of the content. Each participant read texts under three different treatment conditions: (a) TWs were glossed four times in the text (four-gloss: 4G); (b) TWs were first glossed, then retrieved in the L1, and bolded twice (gloss-retrieval: GR); and (c) TWs were first glossed and then bolded three times (gloss-bolding: GB). In addition, the study investigated the effect of increased frequency of word enhancements, comparing word gain of TWs that occurred once (F1) glossed or four times (F4, see above treatments) in the input passage. More specifically, the current study investigated the following research questions: 1. Which level of word knowledge, productive or receptive, do readers in the four-gloss, gloss-retrieval, and gloss-bolding interventions attain during reading? 2. Is the level of word knowledge retained over time? 3. Which of the three gloss interventions leads to more immediate word gain? Is the level of word knowledge gain, receptive or productive, the same in all three treatments? 4. Which of the three gloss interventions leads to more word retention? Is the level of word knowledge retention, receptive or productive, the same in all three treatments? 5. Which word frequency (F1 or F4) leads to more TW gain? Is the level of receptive and productive word knowledge gain the same? 6. What is the effect of the different reading treatments (four-gloss, glossretrieval, and gloss-bolding) on text comprehension? The level of word knowledge gain referred to one level of productive (active recall) and two levels of receptive (passive recall, passive recognition)

171

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

word knowledge gain (Laufer, Elder, Hill, & Congdon, 2004). Productive word knowledge was interpreted as a higher level of word gain than receptive knowledge (see assessment tasks below). Hypotheses Level of Word Gain (Within-Treatment) Hypothesis 1: Immediate measure. In the GR and 4G treatment conditions, participants will gain equal productive and receptive word knowledge. In the GB treatment condition, participants will gain less productive than receptive knowledge. Hypothesis 2: Delayed measure. In the GR condition, participants will retain the level of word knowledge over 5–8 weeks, whereas in the 4G and GB conditions, the level of word knowledge will decrease; that is, word gain in the GR condition will be more robust than in the two other treatment conditions. Number of Words Gained (Between-Treatment) Hypothesis 3: Immediate measure. Word learning will be the same for the GR and the 4G condition and lower for the GB condition (GR = 4G > GB). Hypothesis 4: Delayed measure. The GR condition will lead to better word retention scores than the 4G condition. Word retention scores will be lowest for the GB condition (GR > 4G > GB). These four hypotheses were based on the following considerations: A gloss obtrusively directs the readers’ attention to the TW. While readers look for the gloss in the margin of the text, they isolate the lexical form from the surrounding words in the text. As the readers process the word meaning provided in the gloss, the word meaning might stay long enough in WM to be rehearsed. These processes, isolation and rehearsal, might foster establishing an initial connection between the lexical form and its meaning. Repeatedly processing glosses for the same word might result in the long-term retention of word meaning. Likewise, GR is an obtrusive intervention that might strengthen word encoding. The prompt to provide the L1 translation of the TW requires readers to access the encoded TW meaning in the mental lexicon. In this study, this was the encoding established during the first glossed encounter. Retrieval reinforces the L2-L1 word association, which has been described to be a crucial step in lexical development (e.g., Jiang, 2002). If readers in the current investigation Language Learning 57:2, June 2007, pp. 165–199

172

Susanne Rott

The Effect of Frequency of Input-Enhancements

were not able to retrieve the word meaning, they could search for the gloss provided earlier in the text and establish a connection between the lexical form and its meaning. This search for meaning was also assumed to foster retention (Laufer & Hulstijn, 2001). In contrast, a bolded TW occurrence is less obtrusive than a gloss or a retrieval intervention. It directs the readers’ attention to the TW form but does not necessarily trigger or involve the processing of meaning (see Izumi, 2002, for an overview of input enhancement). Frequency of Encounter Hypothesis 5: The number of words learned and the level of receptive and productive word knowledge gain will be higher for words that occur four times (F4) in the text than for words that occur only once glossed (F1). This hypothesis was based on the assumption that encountering a TW four times stimulated word learning processes, such as rehearsal, elaboration, accessing existing knowledge sources, and search for meaning. One glossed encounter provided fewer opportunities to stimulate learning processes (F4 > F1). Text Comprehension Hypothesis 6: The comprehension of main ideas will be highest for the 4G condition, whereas the GR and the GB condition will lead to less comprehension of main ideas. This hypothesis was based on the consideration that glosses guided the participants’ attention to comprehending propositions that contained glossed TWs, as found in a study by Rott and Williams (2003). Additionally, the glossed words occurred in main ideas of the passages. Therefore, in the GR and the GB conditions, readers’ attention was guided only to the propositions that contained the first gloss of each TW. Bolding of words was expected to be less obtrusive and did not overtly direct the readers’ attention to the propositional content of the sentences containing the TWs, as compared to sentences containing glossed TWs (4G > GR > GB). Method Participants Fifty-four learners of German volunteered to participate in the study. Participants were native speakers of English in four intact fourth-semester language 173

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

classes (about 110 contact hours at the beginning of the study). Participants who had not taken the first three semesters of German in the program had tested into these sections by taking a placement test. Data provided by 16 of the learners had to be dropped from the final analysis because either they did not complete the three treatments or missed the posttest. Results of the study were based on 38 participants. Materials Treatment Passages Three input passages of varying length were loosely adapted from a version in intermediate German textbooks (Holschuh, 1989; Leblans, et al., 2000): Ein unkonventionelles Leben, Philadelphia, and Sarahs Stein. Two expository texts and one narrative text on different topics were chosen to accommodate different student interests and maintain their motivation to read three texts that were not part of the class materials. No special background or cultural knowledge was necessary to comprehend the texts. All three texts were shortened (Leben 411 words, Philadelphia 321 words, Stein 337 words) so that they could easily be read in a 50-min class session. Shortening and including four repetitions of four TWs required some changes in the plots. One near-native speaker and two native speakers of German who are experienced language instructors read and verified the texts for comprehensibility, coherence, and appropriateness for the language level of the participants. Each participant read all three texts under three different conditions (GR, 4G, GB; see Research Questions section). It is generally acknowledged that topic, text characteristics, text comprehension, and the TWs might influence which unknown words learners choose to deal with. Therefore, if participants’ learning behavior across three independent treatment passages and the corresponding TWs was the same, findings could be attributed to the treatment tasks, increasing the generalizability of the results. Target Words Each text contained eight TWs, all of which were nouns. To ensure that the TWs were unknown to the participants, low-frequency items and regional colloquialisms were used in the passages. Sociolinguistic, pragmatic, or other connotational aspects of individual TWs were not accounted for in the study.1 For example, the TW K¨oter has a negative connotation meaning “cur” or “stray dog.” In the passage, it was used for the nonjudgmental word Hund (dog). TWs occurred either one or four times in the treatment passage (Table 1). TWs that occurred four times (F4) in the text were presented in either the GR, the 4G, or the GB treatment condition. TWs that occurred only once (F1) were also Language Learning 57:2, June 2007, pp. 165–199

174

The Effect of Frequency of Input-Enhancements

Susanne Rott

Table 1 The 24 target words of the three treatment passages by frequency (F1 and F4) Text

F1 Words

F4 Words

Sarahs Stein

Krach Gedanken Str¨aucher Eichh¨ornchen

Noise Thoughts Bushes Squirrel

Geplapper Bude Karre Friedhof

Chatting Apartment Car Graveyard

Philadelphia

Metzger Getreide Besitz Lehrling

Butcher Crops Possession Apprentice

Vieh Acker Bauern Kaff

Livestock Field Farmers Village

Ein unkonventionelles Leben

Laden Spende Gestank S¨age

Store Donation Smell Saw

Klamotten Knete Obdachloser K¨oter

Clothes Money Homeless Dog

glossed. Previous investigations have shown that learning occurs when words are glossed once (see Review of Research section). However, it is generally accepted that incidental word acquisition is a cumulative process and requires repeated encounters to establish robust word encodings in the mental lexicon. Even though studies have found that 8–12 encounters might be necessary (for a comprehensive review, see Nation, 2001), more than four encounters would have been unnatural in the fairly short texts used for this study. All TWs were bolded in the text. The TW glosses were provided as L1 translations in the margin of the text. Each repetition of a gloss in the 4G condition looked the same, providing the same L1 meaning. In the GR condition, readers were prompted (e.g., “What does K¨oter mean in English?”) to provide the L1 translation of the TW in the margin of the text. Both F1 and F4 TWs were essential to the main ideas of each passage. English-German bilinguals were asked to read the texts in German and retell them in English. These baseline story text reconstructions (see below) showed that not all four TWs and all occurrences were of equal importance to passage comprehension. Yet, the use of each of the TWs was essential to reconstructing the passage. Assessment Vocabulary Pretest To ensure that TWs were unfamiliar to all participants, a vocabulary checklist test was administered prior to the reading treatment. Students received a list of 175

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

44 lexical items, including the 24 TWs and 20 distracters. Students were asked to explain, in English, what each one meant, even if they had only a vague idea, and to cross out only the words that they did not know at all. Vocabulary Posttests Word gain was assessed with three measures immediately after each reading task and then 4–6 weeks later.2 The first posttest measured word gain of the eight TWs readers had just encountered in the input passage. The delayed posttest assessed word knowledge retention of all 24 TWs from the three treatment passages. The order of TWs was randomized. Based on Laufer et al.’s (2004) Computer Adaptive Test of Size and Strength (CATSS),3 the current study assessed participants’ level of receptive and productive word knowledge gain during reading. Three different levels were measured. First, participants received an active recall (AR; level 1) task, which assessed productive word knowledge. The L1 translation of the TW was presented and participants were prompted to produce the TW in the L2. The next level assessed readers’ receptive word gain with a passive recall (PR; level 2) task. The L2 TW was presented and readers were prompted to provide the L1 translation. The third test, the passive recognition (R; level 3) task, also assessed the readers’ receptive word knowledge gain. This task was less difficult and required the least word knowledge because participants received three L1 meanings (one correct, two distracters, and a “don’t know” option) and were prompted to choose the correct meaning of the L2 TW. Even though the three measures are treated as independent measures, they have an implicational relationship. Therefore, in addition to the eight TWs, the test contained 22 distracter items that were either taken from the text or fitted the topic of the passage content. Distracters were added so that learners would not be able to rehearse word meanings from the level 1 test, which provided the L1 translation, and transfer it to levels 2 and 3, which required readers to produce (level 2) or choose (level 3) the L1 translation from multiple options (see the Appendix for assessment tasks). Text Comprehension Before starting the reading treatment, students were informed that after completing the entire text, they would have to retell, in English, the content of the passage as completely as possible. This was done to ensure that participants focused on meaning during the reading assignment. Text comprehension was assessed in their L1, English, so that their L2 production skills would not interfere with the demonstration of text comprehension. Language Learning 57:2, June 2007, pp. 165–199

176

Susanne Rott

The Effect of Frequency of Input-Enhancements

Scoring and Analysis Word Gain Two independent raters scored the level 1 (AR: L1 to L2 translation) and the level 2 (PR: L2 to L1 translation/explanation) task. The correct provision of the TW in the AR task received a score of 1; a correct translation or explanation of the TW in the PR task also received a score of 1. An incorrect translation (AR and PR) or explanation (PR) in either task received a score of 0. Interrater reliability was 94%. The disagreements (6%) were resolved between the two raters. Scoring the level 3 (R) task was a matter of counting correctly checked answers for the four possible choices. A correct answer received a score of 1; an incorrect choice received a score of 0. Scores for each level of receptive and productive knowledge were tallied separately.

Text Comprehension The goal was to determine whether students had understood the basic event structure of each text. Therefore, the texts were reduced to a set of chronologically ordered main and supporting ideas, based on five German native and near-native speaker retellings and the researcher’s notion of what was most important in the stories (Table 2). For Sarahs Stein (M = 19.2; SD = 1.92), 19 propositions were created; for Philadelphia (M = 16; SD = 1.14), there were 16; and for Ein unkonventionelles Leben 28 (M = 28; SD = 0.71). In addition, the researcher established a list of supporting ideas. For Sarahs Stein, 17 supporting ideas were counted; for Philadelphia, 12 were counted; and for Ein unkonventionelles Leben, 17 were counted. To calculate text comprehension, each proposition received one point. Main and supporting ideas were separately tallied. No partial credit was given. Two raters scored the participants’ retellings with interrater reliabilities of 94% (Sarahs Stein) and 96% (Philadelphia and Ein unkonventionelles Leben). Disagreements were discussed and resolved between the two raters. Table 2 Number of main and supporting ideas of the input passages Input passage

Main ideas

Supporting ideas

Sarahs Stein Philadelphia Ein unkonventionelles Leben

19 16 28

17 12 17

Total

63

46

177

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

Procedure During the second week of the semester, learners signed a participation agreement and completed the vocabulary checklist test. The reading treatments were administered in weeks 5, 6, and 7 by the regular instructor during their normal 50-min class session. Each treatment packet included instructions, the treatment text of one of the three reading conditions (4G, GR, GB), the L1 retelling, and the immediate vocabulary posttests (AR, PR, and R). Each participant completed all three treatments, one in each treatment session (weeks 5, 6, and 7). Treatments were counterbalanced and administered in different orders under different text/task combinations (Table 3). The TWs were considered a random factor because they occurred in three different passages and were encountered under different reading conditions. Students were told that they had 30 min to complete the reading task but that they should proceed through the packet at their own pace. Instructors were told to alert participants to start the L1 retelling at the latest after 30 min and the vocabulary tests after 40 min. In addition, instructors were asked to note the time when students handed in their treatment package. Moreover, students were informed that the packets varied and that they should not be alarmed if some students turned pages at different intervals. Four weeks after the last treatment, in week 11, lexical retention of all 24 TWs was assessed with the same three vocabulary tasks. Students had 20 min to complete them. Analysis For Research Questions 1–5, the statistical analyses were based on a withinsubject design: 3 (gloss treatment condition) × 3 (word gain measure) × 2 (time) Table 3 Setup of treatments Group 1 Vocabulary pretest–2 weeks prior to treatment Treatment 1–week 1 Text 1-GR Immediate vocabulary test Treatment 2–week 2 Text 2-GB Immediate vocabulary test Treatment 3–week 3 Text 3-4G Immediate vocabulary test Delayed vocabulary test–4 weeks later

Group 2

Group 3

Text 2-4G

Text 3-GB

Text 3-GR

Text 1-4G

Text 1-GB

Text 2-GR

Note. Treatment conditions: 4G: four-gloss; GR: gloss-retrieval; GB: one gloss. Language Learning 57:2, June 2007, pp. 165–199

178

The Effect of Frequency of Input-Enhancements

Susanne Rott

× 2 (frequency). The independent variables were the gloss treatment condition (GR, 4G, GB), word gain measure (AR, PR, and R), time (immediate and delayed vocabulary test), and frequency of TW occurrence (F1: once glossed; or F4: GR, 4G, GB). The dependent variables were scores of three levels of vocabulary learning: AR, PR, and R. Text was not included in the analysis because this variable was counterbalanced across participants. This created an N size of 38 for each cell. Effect size measures reported as eta squared (η2 ) were computed as partial η2 because of four independent variables. For Research Question 6, the statistical analysis was based on the following design: 2 (type of idea unit) × 3 (type of treatment condition). Independent variables were the three different treatment conditions (GR, FG, GB) and type of idea unit (main and supporting ideas). The dependent variables were the L1 retell scores for main and supporting idea units. Results Means and standard deviations for the three levels of TW gain in the three F4 and the F1 treatment conditions are reported in Table 4. In order to answer Research Questions 1–5, a 3 × 3 × 2 × 2 MANOVA was conducted. The analysis showed significant main effects for treatment condition, word measure, time of word knowledge measure, and word frequency (Table 5). These main effects were Table 4 Means and standard deviations of word gain measured immediately and 5–8 weeks after the treatment Treatment Condition Gloss-retrieval R

AR

PR

Gloss-bolding

Time

AR

4 TW occurrences

2.82 3.66 3.71 2.84 3.47 3.63 .53 2.53 3.18 (0.73) (0.48) (0.46) (1.05) (0.69) (0.54) (0.73) (0.95) (0.68) 0.24 2.00 3.24 0.26 2.00 3.03 0.21 2.03 3.05 (0.43) (0.84) (0.71) (0.64) (0.84) (0.64) (0.48) (0.85) (0.61)

1 TW occurrence

PR

Four-gloss R

AR

PR

R

Delayed 4 TW occurrences 1 TW occurrence

1.82 3.18 3.53 1.55 2.79 3.29 0.11 1.57 3.26 (0.69) (0.77) (0.60) (0.60) (0.70) (0.52) (0.39) (0.98) (0.50) 0.39 2.26 3.08 0.37 2.18 3.26 0.21 2.24 3.16 (0.75) (0.95) (0.63) (0.75) (0.98) (0.65) (0.47) (0.88) (0.56)

Note. N = 38. Maximum word gain score was four in each cell. Assessment tasks: active recall (AR), passive recall (PR), passive recognition (R). 179

Language Learning 57:2, June 2007, pp. 165–199

The Effect of Frequency of Input-Enhancements

Susanne Rott

qualified by significant two-way, three-way and four-way interactions (Table 5). Overall effect sizes (η2 ) were high with the exception of the main effect for time and interactions involving time; that is, the amount of word gain varied across the three different measures for words that learners encountered one or four times in the three different treatment conditions. In addition, the amount of word knowledge varied across time (although a comparatively small proportion of variance in word gain scores was attributable to the time word gain was measured). To directly address research questions, main effects and interactions were followed up with ANOVAs and pairwise comparisons of least significant difference (LSD) of least square means. In order to avoid Type I errors, a Bonferroni correction was used to adjust the confidence interval. Research Question 1 determined the highest level of word knowledge gain in each F4 gloss treatment (4G, GR, GB). AR was considered the highest level, followed by PR and R (see assessment tasks in the Methods section). The significant main effects for word learning measure and time and the significant interaction between measure and time (Table 5) were followed up with two repeated-measures ANOVAs: one with word gain scores assessed immediately after the treatment and one with word retention scores. For the immediate scores, Table 5 Multivariate test (MANOVA) of 3 (Gloss treatment condition) × 3 (Word gain measure) × 2 (Time) × 2 (Frequency) Design Effect

F

Hypothesis df

Error df

Sig.

Eta squared

Treatment condition (TC) Measure (M) Time (T) Frequency (F) TC × M TC × T M×T TC × M × T TC × F M×F TC × M × F T×F TC × T × F M×T×F TC × M × T × F

70.33 1,567.30 20.61 220.77 40.09 0.98 19.41 3.31 51.42 107.15 25.53 10.13 2.02 14.68 5.07

2 2 1 1 4 2 2 4 2 2 4 1 2 2 4

36 36 37 37 34 36 36 34 36 36 34 37 36 36 34

.00 .00 .00 .00 .00 .39 .00 .00 .00 .00 .00 .00 .15 .00 .00

.80 .99 .36 .86 .83 .05 .52 .28 .74 .85 .75 .62 .10 .45 .37

Note. N = 38; p < .05 Language Learning 57:2, June 2007, pp. 165–199

180

The Effect of Frequency of Input-Enhancements

Susanne Rott

Mauchly’s test indicated that the assumption of sphericity had been violated, χ 2 (35) = 213.27, p < .05; therefore degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = .36). The ANOVA showed that readers had gained different levels of receptive and productive word knowledge, F(2.87, 106.08) = 81.79, p = .00, η2 = .69. Likewise, for the word retention scores, Mauchly’s test indicated that the assumption of sphericity had been violated, χ 2 (35) = 126.24, p < .05; therefore degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = .56). The ANOVA showed that readers had also retained different levels of receptive and productive word knowledge, F(4.51, 166.96) = 118.69, p = .00, η2 = .76. Table 6 presents the results of pairwise comparisons (LSD) of least square means of the three assessment tasks (AR, PR, R). The results showed the same word gain pattern when readers encountered the TWs in the GR and the 4G treatments immediately after reading the input passage. Readers’ word gain scores yielded significantly more receptive (PR and R) than productive word knowledge (AR). Both receptive measures yielded the same amount of word Table 6 Mean differences with confidence intervals (CI) of post hoc pairwise comparisons, using least significant difference (LSD), of three levels of word knowledge gain for the F4 treatment conditions Treatment condition Gloss-retrieval Immediate vocabulary test AR vs. PR −0.84∗ 99% CI −1.14, −0.53 AR vs. R −0.90∗ 99% CI −1.20, −0.59 PR vs. R −.05 99% CI −0.15, 0.04 Delayed vocabulary test AR vs. PR −1.37∗ 99% CI −1.76, −0.98 AR vs. R −1.71∗ 99% CI −2.08, −1.34 PR vs. R −0.34∗ 99% CI −0.60, −0.11

Four-gloss

Gloss-bolding

−0.63∗ −0.95, −0.32 −0.78∗ −1.18, −0.40 −0.16 −0.35, 0.02

−2.00∗ −2.32, −1.68 −2.63∗ −2.96, −2.30 −0.63∗ −1.05, −0.22

−1.24∗ −1.61, −.86 −1.74∗ −2.10, −1.37 −0.50∗ −0.77, −0.23

−1.47∗ −1.85, −1.09 −3.16∗ −3.44, −2.88 −1.68∗ −2.11, −1.26

Note. N = 38 in each cell. ∗ p < .01, Bonferroni adjustment; CI = confidence interval upper and lower. Assessment tasks: active recall (AR), passive recall (PR), passive recognition (R). 181

Language Learning 57:2, June 2007, pp. 165–199

The Effect of Frequency of Input-Enhancements

Susanne Rott

gain, indicating that multiple-choice options (R) did not aid word knowledge retrieval from the mental lexicon. Yet, on the delayed posttest, 4–6 weeks after the reading treatment, the multiple-choice task (R) yielded more word knowledge than the passive recall (PR) and the active recall assessment task (AR), indicating that multiple-choice options aided word knowledge retrieval significantly. In contrast, when readers encountered the TWs four times in the GB treatment, word gain scores were highest for the R task and significantly lower for the PR and the AR tasks. This was the case on the immediate and on the delayed measure. Readers gained significantly more receptive than productive knowledge and benefited from the availability of multiple-choice options for word knowledge retrieval. These findings partially confirmed Hypothesis 1; that is, there was no distinct word learning advantage to develop productive word knowledge during reading for either of two ways: encountering a word four times glossed or glossed once plus a retrieval prompt. In both treatments, readers developed more receptive than productive word knowledge. As predicted, the GB treatment resulted in less productive than receptive word gain. The data analysis for Research Question 2, which assessed whether the initial level of word knowledge (AR, PR, R) was retained across 4–6 weeks, further explored these results (Table 7). Pairwise comparisons (LSD) of immediate and posttest scores showed that after all three treatments (GR, 4G, and GB), word knowledge gain was only robust on the R vocabulary measure, the lowest level of word gain. In fact, word knowledge of the highest level, AR (GR condition Table 7 Mean differences with confidence intervals (CI) of post hoc pairwise comparisons, using least significant difference (LSD), of immediate and delayed word gain scores in the F4 treatment conditions Assessment task Treatment Gloss-retrieval 99% CI Four-gloss 99% CI Gloss-bolding 99% CI

AR

PR ∗

1.00 0.73, 1.27 1.29∗ 0.80, 1.78 0.42∗ 0.16, 0.69

R ∗

0.47 0.12, 0.83 0.68∗ 0.31, 1.06 0.95∗ 0.55, 1.34

0.18 −0.10, 0.47 0.34 0.00, 0.67 −0.11 −0.51, 0 .30

Note. N = 38. ∗ p < .01, Bonferroni adjustment. Assessment tasks: active recall (AR), passive recall (PR), passive recognition (R). Language Learning 57:2, June 2007, pp. 165–199

182

Susanne Rott

The Effect of Frequency of Input-Enhancements

from 71% to 46%, 4G condition from 71% to 39%, and GB condition from 13% to 2%, respectively) and the middle level, PR (GR condition from 92% to 80%, 4G condition from 87% to 70%, and GB condition from 63% to 39%, respectively) dropped significantly across 4–6 weeks. These results partially confirmed Hypothesis 2. Contrary to the initial assumption, the GR treatment did not result in an advantage on long-term word encoding in the mental lexicon compared to 4G TW encounters. As predicted, productive word knowledge decreased within 6 weeks for the 4G treatment and PR word knowledge decreased for the GB treatment. Research Question 3 assessed which of the F4 gloss interventions were most effective for receptive and productive word knowledge development. The main effects for treatment condition, type of word measure, and time of assessment and the significant triple interaction were followed up with a 3 (treatment) × 3 (measure) repeated-measures ANOVA for immediate word gain scores. Mauchly’s test indicated that the assumption of sphericity had been violated, χ 2 (2) = 19.20, p < .05, for treatment; therefore, degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = .71). The ANOVA showed a main effect for treatment, F(1.42, 52.36) = 57, p = .00, η2 = .61, and word measure, F(2, 74) = 342.95, p = .00, η2 = .90. These findings were qualified by a significant interaction, F(4, 148) = 39.53, p = .00, η2 = .52, indicating that readers had gained different levels of receptive and productive word knowledge in the three different reading conditions. Pairwise comparisons (LSD) with associated effect sizes (Cohen’s d) compared cell means of AR, PR, and R scores among the three treatments of the immediate posttest. Results showed that when readers encountered the TWs in the GR and the 4G treatment, they gained the same amount of words on all three levels of word knowledge (AR, PR, R), outperforming the GB treatment (Table 8). All effect sizes of significant differences were medium and large on the Cohen’s d index. These findings confirm Hypothesis 3, indicating a learning advantage for reading texts enhanced with four glosses or a gloss and an additional retrieval task compared to encountering a word once glossed plus three times Delayed posttest scores, measured 4–6 weeks after the treatment (Research bolded. Question 4), showed a slightly different pattern. A 3 (treatment) × 3 (measure) repeated-measures ANOVA for delayed word gain scores showed main effects for treatment, F(2, 74) = 59.12, p = .00, η2 = .62, and word measure, F(2, 74) = 534.52, p = .00, η2 = .94. These findings were qualified by a significant interaction, F(4, 148) = 23.02, p = .00, η2 = .38, indicating that readers had gained different levels of receptive and productive word knowledge retention in the three different reading conditions. Pairwise comparisons (LSD) with 183

Language Learning 57:2, June 2007, pp. 165–199

The Effect of Frequency of Input-Enhancements

Susanne Rott

Table 8 Mean differences with effect sizes (Cohen’s d) of post hoc pairwise comparisons, using least significant difference (LSD), of the effect of treatment conditions on active recall (AR), passive recall (PR), passive recognition (R) Time of Vocabulary Test Immediate Comparison

Mean difference

Active recall GR vs. 4G 0.02 GR vs. GB 2.29∗ 4G vs. GB 2.32∗ Passive recall GR vs. 4G 0.18 GR vs. GB 1.13∗ 4G vs. GB .95∗ Passive recognition GR vs. 4G 0.08 GR vs. GB 0.55∗ 4G vs. GB 0.47∗

Delayed Cohen’s d

Mean difference

Cohen’s d

−0.02 3.14 2.55

0.26 1.71∗ 1.45∗

0.42 3.05 2.85

0.32 1.50 1.33

0.40 1.61∗ 1.21∗

0.53 1.83 1.43

0.16 0.91 0.73

0.24 0.26 0.03

0.43 0.49 0.06

Note. N = 38 in each cell; ∗ p < .01, Bonferroni adjustment; d = Cohen’s d. Treatment condition: gloss-retrieval (GR), four-gloss (4G), gloss-bolding (GB).

associated effect sizes (Cohen’s d) compared cell means of AR, PR, and R scores among the three treatments of the immediate posttest. Table 8 shows that, as on the immediate posttest, reading texts with the GR and the 4G tasks resulted in significantly more word knowledge retention on the AR and the PR measures than reading a text with the GB treatment. However, retention scores on the R measure indicate that all three treatments had a similar effect for the development of R word knowledge. All effect sizes of significant differences were large on the Cohen’s d index. These results only partially confirmed Hypothesis 4. Even though results corroborated the hypothesized inferior word gain through the GB task, the GR task did not result in more word knowledge after 4–6 weeks than the 4G intervention task. Nevertheless, findings indicate a distinct word learning advantage for 4G and GR treatments. Research Question 5 compared lexical development after processing TWs once or four times in the input passage. Table 4 displays the means and standard deviations for F1 (one gloss) and F4 (GR, 4G, GB) word gain scores. The main effect for frequency and the significant interaction among treatment, measure, Language Learning 57:2, June 2007, pp. 165–199

184

The Effect of Frequency of Input-Enhancements

Susanne Rott

frequency, and time (Table 5) were followed up with two 3 (treatment) × 2 (frequency) repeated-measures ANOVAs: one for immediate and one for delayed word gain scores. The ANOVA on the immediate word measure showed main effects for treatment, F(6, 32) = 26.49, p = .00, η2 = .83, and frequency, F(3, 35) = 174.26, p = .00, η2 = .94, and a significant interaction, F(6, 32) = 25.67, p = .00, η2 = .83. Likewise, the ANOVA on the delayed word measure resulted in main effects for treatment, F(6, 32) = 38.37, p = .00, η2 = .88, and frequency, F(3, 35) = 25.27, p = .00, η2 = .68, and a significant interaction, F(6, 32) = 23.40, p = .00, η2 = .81. Pairwise comparisons (LSD) with associated effect sizes (Cohen’s d) of F1 and F4 scores in each treatment condition revealed that when L2 readers encountered the TWs four times in the GR or the 4G condition, they gained significantly more receptive and productive word knowledge than when they encountered a TW only glossed once in the text (Table 9). This was the case on the immediate posttest as well as on the delayed posttest. One exception was the R measure in the 4G condition, which resulted in the same word knowledge score on the delayed posttest. Yet, results further revealed that

Table 9 Mean differences with effect sizes (Cohen’s d ) of post hoc pairwise comparisons, using least significant difference (LSD), of word gain and retention scores of F1 and F4 TW occurrences Assessment task Treatment Immediate Gloss-retrieval d Four-gloss d Gloss-bolding d Delayed Gloss-retrieval d Four-gloss d Gloss-bolding d

AR

PR

R

2.58∗ 4.31 2.58∗ 2.97 0.32 0.52

1.66∗ 2.43 1.47∗ 1.91 0.50 0.55

0.47∗ 0.79 0.61∗ 1.01 0.11 0.20

1.42∗ 1.98 1.18∗ 1.74 −0.11 0.23

0.91∗ 1.06 0.61∗ 0.72 −0.66∗ 0.72

0.45∗ 0.73 0.02 0.05 0.11 0.19

Note. N = 38 in each cell; ∗ p < .01, Bonferroni adjustment; d = Cohen’s d. Assessment tasks: active recall (AR), passive recall (PR), passive recognition (R). 185

Language Learning 57:2, June 2007, pp. 165–199

The Effect of Frequency of Input-Enhancements

Susanne Rott

when readers encountered a TW once glossed plus three additional times bolded (F4), they did not develop more word knowledge than when they encountered a TW glossed only once (F1). This was the case on the immediate posttest and 4–6 weeks later. In fact, encountering a word glossed once (F1) resulted in a significantly higher word gain score on the PR measure on the delayed posttest than encountering the word glossed and bolded three times. All effect sizes of significant differences were large on the Cohen’s d index. These findings only partially supported Hypothesis 5. Apparently, bolding TWs did not significantly affect word learning compared to more obtrusive interventions, such as additional glossing or word retrieval. When readers’ attention was obtrusively directed twice (GR) or four times (4G) to the TWs, they gained significantly more word knowledge than when their attention was less obtrusively (bolded) directed to the TW or occurred glossed only once. Research Question 6 assessed the effect of the individual gloss treatment conditions on text comprehension. Table 10 reports means and standard deviations of L1 retell scores for the three different treatment conditions. For the statistical analysis a 2 (type of idea unit) × 3 (type of treatment condition) MANOVA was conducted. The analysis revealed a significant effect for main ideas, F(2, 111) = 5.39, p = .01, η2 = .88, but no significant effect for supporting ideas, F(2, 111) = 1.51, p = .22, η2 = .03; that is, in all three reading conditions, participants comprehended supporting ideas equally well. The significant effect for main ideas was followed up with pairwise comparisons of LSD among the three treatment conditions (Table 11). Overall, after reading texts enhanced with four glosses (4G) participants comprehended more main ideas than when they read texts enhanced with a gloss and a retrieval task (GR) or a gloss and subsequent bolding of the words (GB). However, comprehension of main ideas was only significantly higher in the 4G than in the GR condition (Table 10). Effect sizes of significant differences were large on the Cohen’s d index. These findings partially confirmed Hypothesis 6. As predicted, a distinct positive effect for text comprehension was found when readers read texts enTable 10 Means and standard deviations of main and supporting ideas of L1 retell scores Gloss-Retrieval

Four-Gloss

Gloss-Bolding

Main

Support

Main

Support

Main

Support

12.42 (4.98)

11.87 (1.93)

16.55 (5.68)

11.21 (1.49)

14.34 (5.78)

11.36 (1.72)

Note. N = 38. Maximum score of main ideas is 63 and 46 for supporting ideas. Language Learning 57:2, June 2007, pp. 165–199

186

The Effect of Frequency of Input-Enhancements

Susanne Rott

Table 11 Mean differences with effect sizes (Cohen’s d) of post hoc pairwise comparisons, using least significant difference (LSD), of the effect of treatment conditions on the retelling of main and supporting ideas Type of Idea Unit Main

Supporting

Comparison

Mean difference

Cohen’s d

Mean difference

Cohen’s d

GR vs. 4G GR vs. GB 4G vs. GB

−4.13∗ −1.92 2.21

−0.77 −0.36 0.38

0.66 0.50 −0.16

0.10 0.29 −0.31

Note. N = 38 in each cell; ∗ p < .01, Bonferroni adjustment; d = Cohen’s d. Treatment condition: gloss-retrieval (GR), four-gloss (4G), gloss-bolding (GB).

hanced with four glosses than when they read a text enhanced with a retrieval task. Contrary to the initial assumption, readers comprehended the main ideas of a text equally well when words were first glossed and subsequently bolded three times. Discussion and Conclusion The current investigation expanded previous lexical research by assessing the combined facilitative effect of word interventions and frequency on vocabulary learning and retention. This study was based on the assumption that the quality of attentional resources and the frequency of occurrences of new lexical forms in a text influence the learning of word meanings during reading. Three treatment conditions were set up to vary the stimulation of attentional resources spent on individual target words: four glosses, one gloss plus a meaning retrieval task and two visual enhancements of the target words, and one gloss plus three additional visual enhancements of the target words. Frequency was accounted for by exposing readers to a TW either once or four times. One major goal of this study was to compare the effect of obtrusively directing L2 readers’ attention to word meaning once (both GB and F1conditions), twice (GR), or four (4G) times. Previous investigations that assessed the effect of frequency of unenhanced words have shown that the repeated occurrence of a new word in a text is a strong predictor for word learning. These studies showed clear differences in meaning gain between one and three encounters (Hulstijn et al., 1996) or two and six encounters (Rott, 1999). Regarding establishing an initial FMC, no frequency effects were found in the current investigation where repetitions included semantic information about word meanings (glosses) or prompted the reader to retrieve the target word meaning. One glossed encounter 187

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

seemed to be as useful as four encounters in establishing a FMC and leaving a memory trace that readers were able to access 4–6 weeks after the encounter. This memory trace was assessed with a receptive multiple-choice measure, the assessment task that required the least word knowledge. Therefore, it can be said that frequency of obtrusive encounters seems to be less important to establish an initial FMC when the word meaning is provided through glosses than when a reader encounters the word in an unenhanced reading situation and has to infer its meaning. Yet, frequency of more obtrusive word interventions, which stimulated the processing of the meaning of the word, had a significant effect on the quality of word encoding (i.e., the level of receptive and productive word knowledge gain). After processing the target word glossed and encountering it three additional times visually enhanced (GB), readers were able to retrieve the word receptively but in only few instances productively. Readers benefited significantly from multiple choices when they were prompted to access the target word knowledge in their mental lexicons. Likewise, when readers processed the word meaning four times through a gloss or when they received a gloss and subsequently were prompted to provide the L1 translation of the word, they equally benefited from multiple meaning choices to demonstrate word knowledge, in particular 4–6 weeks after the treatment. Nevertheless, unlike in the one gloss plus bolding condition, these two intervention tasks resulted in word encoding that allowed readers to retrieve a significant number of words productively. This means that repeatedly directing L2 readers’ attention to semantic word enhancements within the same text has clear advantages for the development of productive word knowledge. Consequently, these findings imply that the visual enhancement of bolding activates qualitatively different attentional resources than semantic enhancements (the provision of glosses or the prompt to retrieve the word meaning), which resulted in higher levels of word knowledge encoding. Current results suggest that repeated visual enhancements (bolding), after establishing an initial FMC, did not strengthen word encoding in the mental lexicon. In fact, readers did not retain more word knowledge of words that were glossed and visually enhanced in three additional contexts than words they encountered only once glossed. A clear explanation for the noneffect of bolding cannot be drawn from the current study. Although assuming that retention of words requires the isolation of the lexical form from its context and some type of elaboration of meaning or rehearsal in working memory (e.g., Ellis, 1994; Gass, 1997; Hulstijn, 2001), it might be that when readers encountered the bolded target words, they merely

Language Learning 57:2, June 2007, pp. 165–199

188

Susanne Rott

The Effect of Frequency of Input-Enhancements

focused on comprehending the text without temporarily shifting their attention to the lexical form and its unique meaning. Bolding individual words in a text without indicating the purpose might cause readers to attend to the word aspect they consider most important (e.g., spelling) or to not attend to any particular aspect at all (see Izumi, 2002, for a review on input enhancement of grammatical forms). Moreover, in contrast to the initial assumption, the data revealed that a gloss followed by a retrieval prompt did not lead to more productive word knowledge than processing a word glossed four times. In fact, readers established and retained the same level of word knowledge in both of these treatment conditions. Accessing and retrieving the stored target word meaning apparently did not lead to qualitatively different word encoding than shifting attentional resources to the glossed word meaning in the margin of the text. This finding was surprising because it was assumed that word retrieval would lead to a deeper level of word processing and the reinforcement of the L1-L2 word association. Yet, current findings suggest that the quantity of attentional resources allocated during the repeated processing of glosses might compensate for the quality of attentional resources spent during one word retrieval. Whereas the retrieval task resulted in word processing that led to more productive word gain than one gloss alone, the repeated focus on glosses resulted in word processing that led to equally productive word gain. It can be assumed that shifting attentional resources repeatedly to the target word in the margin provided repeated opportunities to rehearse word meaning in WM. Nevertheless, this study does not provide conclusive evidence for the effect of frequency because it cannot be deduced from the current data collection how word encoding in these two conditions developed across four word encounters: Whether it occurred gradually or during one particular word encounter. However, what becomes clear is that the combined effect of frequency and attention to semantic meaning increased the rate of productive word gain. These findings can be further explained through the Involvement Load Hypothesis (Laufer & Hulstijn, 2001). According to the Involvement Load Hypothesis, a gloss prompts readers to perceive the need for the target word meaning, earning a processing load of one. The retrieval task receives a processing load of three because in addition to stimulating the need for the target word, the reader has to search for the meaning in the mental lexicon and evaluate the correct meaning retrieval by writing it down. This higher level of processing load is said to result in better learning. Nonetheless, the repeated processing of glosses in the current study resulted in a processing load of four (a load of 1 × 189

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

4 occurrences), the GR condition also had a processing load of four (one for the gloss plus three for the word retrieval task), whereas the one-gloss condition earned a processing load of one. Thereby, the current investigation further supports the Involvement Load Hypothesis and suggests that the factor frequency of word intervention might mediate and interact with the three dimensions of word learning; need, search, and evaluation. Current results provided further insights into the effect of glosses on word meaning retention. As in studies by Ko (1995) and Watanabe (1997), the present findings confirmed that the processing of glosses has a long-term effect on word learning. The current study expanded previous investigations, which mainly used L1 to L2 translation tasks to assess word learning, by using more graded assessment tasks to measure long-term word gain (one productive and two receptive tasks). Thereby, this study revealed a variation in retention patterns for target words that were encountered four times as compared to target words encountered only once. Word encoding after one glossed encounter remained relatively robust across 4–6 weeks on all three levels of word measures. A different pattern was observed for receptive and productive word knowledge retention after four glossed target word encounters. Receptive word knowledge, measured with the multiple-choice recognition task, was retained for all words over 4–6 weeks when readers encountered the target words four times. Yet, the ability to recall the word meaning receptively on a L2 to L1 translation task and to retrieve word meaning productively (L1 to L2) was not robust over time; that is, word interventions that led to the highest level of immediate word knowledge gain also decreased the most over 4–6 weeks. Readers were still able to retrieve more words for productive use after four glossed or glossed and retrieved encounters in the text than after one glossed encounter. Therefore, it can be said that the increased frequency of obtrusive semantic word interventions resulted in the increase of word retention. A second goal of the current investigation was to assess whether repeated word interventions that directed the readers’ attention to individual words facilitated or hindered the comprehension of the text. Taking L2 learners’ limited ability to attend to multiple information sources at the same time as a starting point, this study assessed interventions that varied in frequency and the level of obtrusiveness with which the readers’ attention was directed to the TWs. As reported earlier, even though the instructions of the reading tasks directed the participants’ attention to comprehending the input passage, in all conditions the readers engaged in word processing that led to word learning.

Language Learning 57:2, June 2007, pp. 165–199

190

Susanne Rott

The Effect of Frequency of Input-Enhancements

Previous studies found that exposing L2 readers to glosses led to superior text comprehension as compared to an unenhanced reading condition (Davis, 1989; Jacob, 1994) or the use of appositives (Watanabe, 1997). The present investigation extended this line of investigation by further comparing text comprehension when words were glossed once and bolded thereafter, glossed once and subsequently retrieved, or glossed four times. Results revealed that when texts were enhanced with four glosses for each target word, readers comprehended main ideas better than when texts were enhanced with one gloss, a word retrieval prompt, plus two visual enhancements. In addition, there was a trend (yet no statistically significant difference) indicating that readers comprehended main ideas better when they read texts with four glosses than with only one gloss and additional visual enhancements. Findings suggest that the repeated provision of meaning for the same word triggered not only the comprehension of the target word but also the entire proposition, a finding also reported in Rott and Williams (2003). Therefore, it can be said that the obtrusive intervention of glosses directly influenced text comprehension by making propositional content easier to access while raising the readers’ awareness of the importance of the propositional content. Visual enhancements of individual words did apparently also stimulate the readers’ motivation to comprehend the propositional content. These results contrast with observations made by Moravcsik and Healy (1998). They showed that L1 readers comprehended text less well when words were highlighted compared to when the entire text or no words were enhanced. One explanation might be that, in the current study, the initial gloss stimulated the readers’ motivation to comprehend additional textual propositions containing the same word bolded. Future research needs to directly compare the effect of a gloss plus bolding and bolding alone on text comprehension. In turn, the data revealed a trend indicating that even though the production task (GR) might foster word learning, it might interfere with text comprehension. Overall, participants comprehended the least main ideas when the reading process was interrupted to recall and note down the meaning of individual words. However, the current data do not reveal whether the retrieval task drew on attentional resources necessary for text comprehension or whether the retrieval task prompted learners to perceive the task more of a word learning task than a reading task. Additionally, in all three intervention conditions, readers comprehended supporting ideas equally well. The explanation might be that all three conditions were alike in that supporting ideas were not enhanced. Consequently, these findings lend support to the assumption that the provision of glosses serves

191

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

two purposes. They foster word learning mechanisms and trigger L2 readers to focus their attention to the textual propositions enhanced with glosses without depleting cognitive resources for either process.

Limitations and Future Research Although the current investigation further expanded our understanding of the importance of glosses for text comprehension, it failed to clearly demonstrate how the number of glosses in a text is related to the number of ideas recalled from a text. Neither did this study show the clear relationship between glosses and individual readers’ situation model of that text. Future research should manipulate the frequency of glosses in a text as well as the distribution of glosses across main and supporting ideas. Likewise, it would be of interest to assess whether there is an ideal ratio of glossed and nonglossed words in a text that stimulates word learning as well as text comprehension. Moreover, the current investigation was not designed to determine why the word retrieval task did not lead to the anticipated superior word retention. It would be interesting to see whether four retrieval prompts would enhance word learning or text comprehension, or whether this would lead to readers’ boredom. In addition, this study tapped learners’ partial word knowledge gain and only assessed selected word aspects. Future research should further expand studies on word knowledge gain by assessing a broader spectrum of word aspects, such as syntactic and pragmatic aspects in more contextualized production tasks. Finally, this study set out to investigate incidental vocabulary acquisition directing the participants’ attention to comprehend the content of each reading treatment. Nevertheless, participants might have directed more attention to the TWs when they read the second or the third treatment passage because they knew from the first treatment that a vocabulary measure would follow. Therefore, the study cannot make an unqualified claim about incidental learning because some of the words might have been learned intentionally. Revised version accepted 24 August 2006

Notes 1

Based on the textbook used for the first three semesters of the language program, fourth-semester language learners should know already between 1,500-2000 words for day-to-day topics. Therefore, it is not easy to identify target words central to the meaning of the passage that can be repeated four times. The researcher wanted to refrain from making students learn nonsense words.

Language Learning 57:2, June 2007, pp. 165–199

192

Susanne Rott

2

3

The Effect of Frequency of Input-Enhancements

Retention of all TWs from the different treatment conditions was measured on the same day for the following reasons: a) to reduce the possibility of missing students; b) so that students would not be forewarned during the first retention test that two more tests would be following; c) to avoid repeated interruption of class time. It was also assumed that word retention was fairly stable after three weeks (Schmitt, 2001). Changes were made to the original CATSS (Laufer et al., 2004) because the current test was administered with paper and pencil and not on the computer. In addition fewer TWs were tested in the current study. Therefore, the first letter of the word was not provided and distracters were added. Finally, the active recognition level was omitted because it requires learners to choose the correct TW meaning from multiple L2 choices. Because the current student population has a very limited lexicon, it would have been too difficult to find appropriate distracters.

References Barcroft, J. (2002). Semantic and structural elaboration in L2 lexical acquisition. Language Learning, 52, 323–363. Davis, J. (1989). Facilitating effects of marginal glosses on foreign language reading. Modern Language Journal, 73, 41–48. Ellis, N. (1994). Implicit and explicit language learning: An overview. In N. Ellis (Ed.), Implicit and explicit learning of languages (pp. 1–31). London: Academic Press. Ellis, N. (2001). Memory for language. In P. Robinson (Ed.), Cognition and second language instruction (pp. 33–68). Cambridge: Cambridge University Press. Ellis, N. (2004). The processes of second language acquisition. In B. VanPatten, J. Williams, S. Rott, & M. Overstreet (Eds.), Form-meaning connections in second language acquisition (pp. 49–76). New York: Erlbaum. Gass, S. M. (1997). Input, interaction and the second language learner. Mahwah: Erlbaum. Gass, S. M., Svetics, L., & Lemelin, S. (2003). Differential effects of attention. Language Learning, 53, 497–545. Holschuh, A. (1989). Leutebuch. Ein leichtes Lesebuch. New York: Hartcourt Brace. Horst, M., Cobb, T., & Meara, P. (1998). Beyond A Clockwork Orange: Acquiring second language vocabulary through reading. Reading in a Foreign Language, 11, 207–223. Hulstijn, J. (1993). When do foreign language readers look up the meaning of unfamiliar words? The influence of task and learner variables. Modern Language Journal, 77, 139–147. Hulstijn, J. (2001). Intention and incidental second language vocabulary learning: A reappraisal of elaboration, rehearsal, and automaticity. In P. Robinson (Ed.), Cognition and second language instruction (pp. 258–286). Cambridge: Cambridge University Press. 193

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

Hulstijn, J., Hollander, M., & Greidanus, T. (1996). Incidental vocabulary learning by advanced foreign language students: The influence of marginal glosses, dictionary use, and reoccurrence of unknown words. Modern Language Journal, 80, 327–339. Izumi, S. (2002). Output, input enhancement, and the noticing hypothesis. An experimental study on ESL relativization. Studies in Second Language Acquisition, 24, 541–577. Jacobs, G. (1994). What lurks in the margin: Use of vocabulary glosses as a strategy in second language reading. Issues in Applied Linguistics, 5, 115–137. Jacobs, G., Dufon, P., & Hong, C. F. (1994). L1 and L2 glosses in L2 reading passages: Their effectiveness for increasing comprehension and vocabulary knowledge. Journal of Research in Reading, 17, 19–28. Johnson, P. (1982). Effects of reading comprehension of building background knowledge. TESOL Quarterly, 16, 503–516. Jiang, N. (2002). Form-meaning mapping in vocabulary acquisition in a second language. Studies in Second Language Acquisition, 24, 617–637. Judd, E. L. (1978). Vocabulary teaching and TESOL: A need for re-evaluation of existing assumptions. TESOL Quarterly, 16, 503–516. Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge: Cambridge University Press. Ko, M. (1995). Glossing in incidental and intentional learning of foreign language vocabulary and reading. University of Hawaii Working Papers in ESL, 13, 49–94. Krashen, S. D. (1993). The power of reading: Insights from research. Englewood, CO: Libraries Unlimited. Laufer, B. (2001). Reading, word-focused activities and incidental vocabulary acquisition in a second language. Prospect, 16, 44–54. Laufer, B., Elder, C., Hill, K., & Congdon, P. (2004). Size and strength: Do we need both to measure vocabulary knowledge? Language Testing, 21, 202–227. Laufer, B., & Hulstijn, J. (2001). Incidental vocabulary acquisition in a second language: The construct of task-induced involvement. Applied Linguistics, 22, 1–26. Leblans, A., Mifflin, D., Mullens, D. M., Paskow, J., Poser, Y., & Strauch, G. (2000). Was ist deutsch? Boston: Houghton Mifflin. Lee, J., & VanPatten, B. (1995). Making communicative language teaching happen. New York: McGraw Hill. Moravcsik, J. E., & Healy, A. F. (1998). Highlighting important words leads to poorer comprehension. In A. F. Healy & L. E. Bourne (Eds.), Foreign language learning: Psycholinguistic studies in training and retention (pp. 259–272). Mahwah, NJ: Erlbaum. Nation, I. S. P. (2001). Learning vocabulary in another language. Cambridge: Cambridge University Press. Pak, J. (1986). The effect of vocabulary glossing on ESL reading comprehension. Unpublished doctoral dissertation, University of Hawai’i at Manoa, Hawai’i. Paribakht, T., & Wesche, M. (1999). Reading and “incidental” L2 vocabulary acquisition. Studies in Second Language Acquisition, 21, 195–224. Language Learning 57:2, June 2007, pp. 165–199

194

Susanne Rott

The Effect of Frequency of Input-Enhancements

Parry, K. (1993). Too many words: Learning the vocabulary of an academic subject. In T. Huckin, M. Haynes, & J. Coady (Eds.), Second language reading and vocabulary acquisition (pp. 109–129). Norwood, NJ: Ablex. Parry, K. (1997). Vocabulary and comprehension: Two portraits. In T. Huckin & J. Coady (Eds.), Second language vocabulary acquisition (pp. 55–68). Cambridge: Cambridge University Press. Prince, P. (1996). Second language vocabulary learning: The role of context versus translation as a function of proficiency. Modern Language Journal, 80, 478– 493. Rott, S. (1997). The effect of exposure frequency and reading comprehension on incidental vocabulary acquisition and retention through reading for learners of German as a foreign language. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign. Rott, S. (1999). The effect of exposure frequency on intermediate language learners’ incidental vocabulary acquisition and retention through reading. Studies in Second Language Acquisition, 21, 589–619. Rott, S., & Williams, J. (2003). Making form-meaning connections while reading: A qualitative analysis of the effect of input and output tasks on word learning. Reading in a Foreign Language 15, 45–74. Rott, S., Williams, J., & Cameron, R. (2002). The effect of multiple-choice L1 glosses and input-output cycles on lexical acquisition and retention. Language Teaching Research, 6, 183–222. Richards, J. C. (1976). The role of vocabulary teaching. TESOL Quarterly, 10, 77–89. Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language instruction (pp. 3–32). Cambridge: Cambridge University Press. Sharwood Smith, M. (1986). Comprehension vs. acquisition: Two ways of processing input. Second Language Research, 7, 118–132. VanPatten, B. (1996). Input processing and grammar instruction: Theory and research. Norwood, NJ: Ablex. Waring, R., & Takaki, M. (2003). At what rate do learners learn and retain new vocabulary from reading a graded reader? Reading in a Foreign Language, 15, 130–163. Watanabe, Y. (1997). Input, intake, and retention: Effects of increased processing on incidental learning of foreign language vocabulary. Studies in Second Language Acquisition, 19, 287–307. Wesche, M., & Paribakht, T. S. (2000). Reading-based exercises in second language vocabulary learning: An introspective study. Modern Language Journal, 84, 196–213. Zahar, R. Cobb, T., & Spada, N. (2001). Acquiring vocabulary through reading: Effects of frequency and contextual richness. Canadian Modern Language Review, 57, 541–572.

195

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

Appendix Example Passage Target words are bolded. Ein unkonventionelles Leben “Ganz normal ist das nicht, was ich da mache”, meint J¨urgen von sich selbst. Im Winter wie im Sommer wandert er mit seinem K¨oter Asda durch Deutschland. Er ist Obdachloser. Er hat keine Arbeit und verdient keine Knete. Er ist einer von hunderttausenden, die auf der Strasse leben. Alles, was er braucht, tr¨agt er selbst: zwei Schlafs¨acke, ein St¨uck Plastik, einen Teller, ein Messer, eine Gabel und einen L¨offel. J¨urgen w¨ascht sich nicht oft. Seine Klamotten hat er immer an, tags¨uber und nachts. Wenn sie zu sehr stinken, wirft er sie weg. Eine andere Hose, ein T-Shirt, und eine Jacke kann man immer bekommen. J¨urgen schl¨aft oft unter einer B¨uckeroder auf der Strasse. Sein K¨oter Asda schl¨aft immer an seiner Seite und weckt ihn jeden Morgen. Dann packt J¨urgen seine Sachen zusammen und wandert in die n¨achste Stadt; heute nach Esslingen. Zuerst geht er in ein Stehcafe. Die Verk¨auferin dort macht schnell die T¨ur weit auf, trotz der K¨alte draussen. Wie gesagt: er stinkt, da er immer dieselben Klamotten tr¨agt. Danach geht J¨urgen zum B¨uro f¨ur Obdachlose. H¨oflich bittet er den Sozialarbeiter um ein Hemd und Schuhe. Das Hemd bekommt er, Schuhe nicht, dazu aber 12 DM. Das ist mehr als in anderen St¨adten, wo man keine Knete bekommt. J¨urgen sucht sich einen Platz in der Einkaufspassage. Heute ist Samstag und viele Leute sind beim Einkaufen. Die Leute, in ihren schicken Klamotten, gehen an J¨urgen vorbei. Er h¨alt einen Karton vor sich hin, auf dem steht: “Wir haben Hunger und bitten um eine Spende.” Nicht immer ist es Knete, was J¨urgen bekommt. Auf dem Karton steht ja etwas von Hunger, und daher geben die Menschen ihm etwas zu Essen. Nach der sechsten Bratwurst muss er aufh¨oren. Auch der K¨oter, der neben ihm auf seinem Mantel liegt, will nichts mehr fressen. Wenn die Leute ihm nichts geben, geht J¨urgen in eine K¨uche f¨ur Obdachlose. Manchmal kennt er ein paar Leute, die, wie er, auch auf der Strasse leben. In der n¨achsten Stadt, Stuttgart, arbeitet J¨urgen ein paar Stunden und verdient ein bisschen Knete. Er bekommt eine S¨age und arbeitet mit einer Gruppe von M¨annern im Wald. Nach zwei Stunden hat er 10 DM. Daf¨ur kann er sich im Second-Hand Laden ein paar warme Klamotten f¨ur den Winter kaufen. F¨ur Obdachlose ist der Winter immer problematisch. Es ist schwer einen Platz zum

Language Learning 57:2, June 2007, pp. 165–199

196

Susanne Rott

The Effect of Frequency of Input-Enhancements

Schlafen zu finden. Besonders, weil viele Leute keine K¨oter in ihre Wohnung lassen wollen. Die Leute m¨ogen den Gestank nicht und haben Angst, wenn Asda bellt. Immediate Posttest Page 1 Instructions: Please provide the German meaning of the words. Even if you have only a vague idea about a word, please write it down. Scratch out those words you cannot make sense of at all. 1. sleeping bag 2. office 3. saw 4. dog 5. plate 6. hat 7. cold 8. donation 9. payment 10. homeless person 11. shopping 12. spoon

13. travel companion 14. smell 15. money 16. street 17. store 18. bridge 19. poverty 20. clothes 21. hunger 22. shirt 23. peddler 24. sausage

Page 2 Instructions: Please provide the English meaning of the words. Even if you have only a vague idea about a word, please write it down. Scratch out those words you cannot make sense of at all. a) Klamotten b) B¨uro c) Hut d) K¨alte e) Knete f) Spende g) Zahlung h) Einkauf i) Obdachloser j) Laden k) Br¨ucke l) Armut

197

m) n) o) p) q) r) s) t) u) v) w) x)

Hunger Hemd Teller Schlafsack S¨age L¨offel Wurst K¨oter Gestank Knete Strasse Reisepartner

Language Learning 57:2, June 2007, pp. 165–199

Susanne Rott

The Effect of Frequency of Input-Enhancements

Page 3 (This page includes the target words only. The actual test contained 14 additional distracter items.) Please circle the correct meaning from the following words: K¨oter a) travel companion b) friend c) dog d) don’t know

Laden a) store b) office c) organization d) don’t know

Gestank a) peddler b) smell c) lodging d) don’t know

Spende a) agency b) donation c) payment d) don’t know

Obdachloser a) homeless person b) travel agent c) peasant d) don’t know

Knete a) security b) money c) poverty d) don’t know

Klamotten a) boots b) hats c) clothes d) don’t know

S¨age a) saw b) forest c) ax d) don’t know

Main Propositions of the Input Passage Ein unkonventionelles Leben (28 Propositions) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

J¨urgen ist Obdachloser. Er hat einen K¨oter. Er wandert durch Deutschland. Er hat keine Arbeit. Er hat kein Geld. Er tr¨agt seinen ganzen Bestiz bei sich. Er w¨ascht sich nicht. Er wirft seine Klamotten weg. J¨urgen schl¨aft unter der Br¨ucke. Er geht nach Esslingen. Er geht in ein Cafe. Er stinkt.

Language Learning 57:2, June 2007, pp. 165–199

198

Susanne Rott

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

199

The Effect of Frequency of Input-Enhancements

Im Cafe machen sie die T¨ur auf. Er geht zum B¨uro f¨ur Obdachlose. Er bittet um Kleidung. Er bekommt 12 DM. In anderen St¨adten bekommt man keine Knete. J¨urgen ist in einer Passage. Viele Leute sind beim Einkaufen. Er bettelt. Er bekommt nicht nur Knete sondern auch Essen. Er kann nicht so viel Wurst essen. K¨oter kann nicht so viel Wurst fressen. Er schl¨aft in einem Heim f¨ur Obdachlose. Er arbeitet mit einer S¨age im Wald. Er verdient 10 DM. Er kauft Winterklamotten. Er hat Probleme mit Schlafplatz, weil Leute keine K¨oter m¨ogen.

Language Learning 57:2, June 2007, pp. 165–199