Escolar Documentos
Profissional Documentos
Cultura Documentos
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.
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 lex-
ical development, the educationally relevant question remains: Which word
interventions (a) increase the rate of word learning through reading, (b) fos-
ter 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 form-
meaning 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 in-
volved in reading comprehension and lexical acquisition different, they might
even be in conflict. In particular, if we follow the assumption that WM is a
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 compre-
hension (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 vo-
cabulary 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 deci-
sion 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 moti-
vational 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 ex-
plicitly assigning a meaning to an unfamiliar word (Parry, 1993, 1997). Second,
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 unen-
hanced 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.).
& 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.
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: input-
only 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 un-
derstanding 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 in-
creased 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, gloss-
retrieval, and gloss-bolding) on text comprehension?
The level of word knowledge gain referred to one level of productive (ac-
tive recall) and two levels of receptive (passive recall, passive recognition)
word knowledge gain (Laufer, Elder, Hill, & Congdon, 2004). Productive word
knowledge was interpreted as a higher level of word gain than receptive knowl-
edge (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.
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 assump-
tion that encountering a TW four times stimulated word learn-
ing processes, such as rehearsal, elaboration, accessing exist-
ing 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’ at-
tention to comprehending propositions that contained glossed
TWs, as found in a study by Rott and Williams (2003). Addi-
tionally, the glossed words occurred in main ideas of the pas-
sages. 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’ atten-
tion 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. Partici-
pants were native speakers of English in four intact fourth-semester language
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 corre-
sponding 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 Köter 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
Table 1 The 24 target words of the three treatment passages by frequency (F1 and F4)
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 Köter 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
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 pro-
ductive word knowledge gain during reading. Three different levels were mea-
sured. First, participants received an active recall (AR; level 1) task, which
assessed productive word knowledge. The L1 translation of the TW was pre-
sented 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 diffi-
cult 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 dis-
tracter 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 com-
pleting 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 partici-
pants 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.
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 na-
tive 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 sup-
porting 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 be-
tween the two raters.
Sarahs Stein 19 17
Philadelphia 16 12
Ein unkonventionelles Leben 28 17
Total 63 46
Procedure
During the second week of the semester, learners signed a participation agree-
ment 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 com-
pleted 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 within-
subject design: 3 (gloss treatment condition) × 3 (word gain measure) × 2 (time)
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
Time AR PR R AR PR R AR PR R
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)
1 TW occurrence 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)
Delayed
4 TW occurrences 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)
1 TW occurrence 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).
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 vio-
lated, χ 2 (35) = 126.24, p < .05; therefore degrees of freedom were corrected us-
ing 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 compar-
isons, using least significant difference (LSD), of three levels of word knowledge gain
for the F4 treatment conditions
Treatment condition
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 treat-
ment, 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 knowl-
edge and benefited from the availability of multiple-choice options for word
knowledge retrieval.
These findings partially confirmed Hypothesis 1; that is, there was no dis-
tinct 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 recep-
tive 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 ini-
tial 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 com-
parisons, using least significant difference (LSD), of immediate and delayed word gain
scores in the F4 treatment conditions
Assessment task
Treatment AR PR R
∗ ∗
Gloss-retrieval 1.00 0.47 0.18
99% CI 0.73, 1.27 0.12, 0.83 −0.10, 0.47
Four-gloss 1.29∗ 0.68∗ 0.34
99% CI 0.80, 1.78 0.31, 1.06 0.00, 0.67
Gloss-bolding 0.42∗ 0.95∗ −0.11
99% CI 0.16, 0.69 0.55, 1.34 −0.51, 0 .30
Note. N = 38. ∗ p < .01, Bonferroni adjustment. Assessment tasks: active recall (AR),
passive recall (PR), passive recognition (R).
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 as-
sumption, 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 assess-
ment 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 cor-
rected 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 compar-
isons (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 ef-
fect 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 re-
trieval 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 signifi-
cant interaction, F(4, 148) = 23.02, p = .00, η2 = .38, indicating that readers
had gained different levels of receptive and productive word knowledge reten-
tion in the three different reading conditions. Pairwise comparisons (LSD) with
Table 8 Mean differences with effect sizes (Cohen’s d) of post hoc pairwise compar-
isons, using least significant difference (LSD), of the effect of treatment conditions on
active recall (AR), passive recall (PR), passive recognition (R)
Immediate Delayed
Active recall
GR vs. 4G 0.02 −0.02 0.26 0.42
GR vs. GB 2.29∗ 3.14 1.71∗ 3.05
4G vs. GB 2.32∗ 2.55 1.45∗ 2.85
Passive recall
GR vs. 4G 0.18 0.32 0.40 0.53
GR vs. GB 1.13∗ 1.50 1.61∗ 1.83
4G vs. GB .95∗ 1.33 1.21∗ 1.43
Passive recognition
GR vs. 4G 0.08 0.16 0.24 0.43
GR vs. GB 0.55∗ 0.91 0.26 0.49
4G vs. GB 0.47∗ 0.73 0.03 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 Hy-
pothesis 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,
frequency, and time (Table 5) were followed up with two 3 (treatment) × 2 (fre-
quency) 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 condi-
tion, 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 compar-
isons, using least significant difference (LSD), of word gain and retention scores of F1
and F4 TW occurrences
Assessment task
Treatment AR PR R
Immediate
Gloss-retrieval 2.58∗ 1.66∗ 0.47∗
d 4.31 2.43 0.79
Four-gloss 2.58∗ 1.47∗ 0.61∗
d 2.97 1.91 1.01
Gloss-bolding 0.32 0.50 0.11
d 0.52 0.55 0.20
Delayed
Gloss-retrieval 1.42∗ 0.91∗ 0.45∗
d 1.98 1.06 0.73
Four-gloss 1.18∗ 0.61∗ 0.02
d 1.74 0.72 0.05
Gloss-bolding −0.11 −0.66∗ 0.11
d 0.23 0.72 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).
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 signif-
icantly 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 devi-
ations 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 sup-
porting ideas, F(2, 111) = 1.51, p = .22, η2 = .03; that is, in all three reading
conditions, participants comprehended supporting ideas equally well. The sig-
nificant 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 en-
Table 10 Means and standard deviations of main and supporting ideas of L1 retell scores
Table 11 Mean differences with effect sizes (Cohen’s d) of post hoc pairwise compar-
isons, using least significant difference (LSD), of the effect of treatment conditions on
the retelling of main and supporting ideas
Main Supporting
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.
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 sup-
ports 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 re-
ceptive 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 differ-
ent 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 fa-
cilitated or hindered the comprehension of the text. Taking L2 learners’ lim-
ited 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.
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.
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.
2
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).
3
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.
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.
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.
Appendix
Example Passage
Target words are bolded.
Ein unkonventionelles Leben
“Ganz normal ist das nicht, was ich da mache”, meint Jürgen von sich
selbst. Im Winter wie im Sommer wandert er mit seinem Köter 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, trägt er selbst: zwei Schlafsäcke, ein Stück Plastik, einen Teller, ein
Messer, eine Gabel und einen Löffel.
Jürgen wäscht sich nicht oft. Seine Klamotten hat er immer an, tagsüber
und nachts. Wenn sie zu sehr stinken, wirft er sie weg. Eine andere Hose, ein
T-Shirt, und eine Jacke kann man immer bekommen.
Jürgen schläft oft unter einer Bückeroder auf der Strasse. Sein Köter Asda
schläft immer an seiner Seite und weckt ihn jeden Morgen. Dann packt Jürgen
seine Sachen zusammen und wandert in die nächste Stadt; heute nach Esslingen.
Zuerst geht er in ein Stehcafe. Die Verkäuferin dort macht schnell die Tür
weit auf, trotz der Kälte draussen. Wie gesagt: er stinkt, da er immer dieselben
Klamotten trägt. Danach geht Jürgen zum Büro für Obdachlose. Höflich 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 Städten, wo man keine
Knete bekommt.
Jürgen sucht sich einen Platz in der Einkaufspassage. Heute ist Samstag
und viele Leute sind beim Einkaufen. Die Leute, in ihren schicken Klamotten,
gehen an Jürgen vorbei. Er hält einen Karton vor sich hin, auf dem steht:
“Wir haben Hunger und bitten um eine Spende.” Nicht immer ist es Knete,
was Jürgen bekommt. Auf dem Karton steht ja etwas von Hunger, und daher
geben die Menschen ihm etwas zu Essen. Nach der sechsten Bratwurst muss er
aufhören. Auch der Köter, der neben ihm auf seinem Mantel liegt, will nichts
mehr fressen. Wenn die Leute ihm nichts geben, geht Jürgen in eine Küche
für Obdachlose. Manchmal kennt er ein paar Leute, die, wie er, auch auf der
Strasse leben.
In der nächsten Stadt, Stuttgart, arbeitet Jürgen ein paar Stunden und ver-
dient ein bisschen Knete. Er bekommt eine Säge und arbeitet mit einer Gruppe
von Männern im Wald. Nach zwei Stunden hat er 10 DM. Dafür kann er sich im
Second-Hand Laden ein paar warme Klamotten für den Winter kaufen. Für
Obdachlose ist der Winter immer problematisch. Es ist schwer einen Platz zum
Schlafen zu finden. Besonders, weil viele Leute keine Köter in ihre Wohnung
lassen wollen. Die Leute mögen 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.
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 m) Hunger
b) Büro n) Hemd
c) Hut o) Teller
d) Kälte p) Schlafsack
e) Knete q) Säge
f) Spende r) Löffel
g) Zahlung s) Wurst
h) Einkauf t) Köter
i) Obdachloser u) Gestank
j) Laden v) Knete
k) Brücke w) Strasse
l) Armut x) Reisepartner
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:
Köter Laden