This chapter reviews theoretical support for the design of the lexical
interventions employed in this study. This section will first introduce glossing, a
technique that can enhance vocabulary learning from reading. Then, there is a
review of involvement load hypothesis to point out the weakness of glossing in
triggering search and evaluation processes, which paves the way for the subsequent
discussions of lexical inferencing and retrieval. Based on the review, six hypotheses
corresponding to the research questions are posed at the end of this chapter.
Glossing
Glosses explicitly direct learners’ attention to lexical items (Schmitt, 2008) and
help them to notice more target vocabulary (Bowles, 2004; Hulstijn, 2001; Miyasako,
2002; Rott, 2005; Yanguas, 2009), “supporting the notion of ‘consciousness-raising’
and ‘input enhancement’” (Nagata, 1999, p. 469). In an effort to induce learners to
notice words, it does not matter whether glosses are presented electronically or in the
traditional paper-and-pen format (Bowles, 2004), combined with pictures, displayed
in the form of texts or pictures alone (Yanguas, 2009). Although noticing is essential
in learning language (deBot, Lowie, & Verspoor, 2005), it cannot be taken for granted
that a new word will always come to learners’ notice (Laufer, 2003, 2005, 2006). In
fact, some kind of Focus on Form such as attention to glosses has been found to be
superior to comprehension-based learning (Laufer, 2006). If the learners read
without any meaning support, they often ignored the unknown words or inferred their
meanings incorrectly, which led to the limited reoccurrence effect, but if the
form-meaning connection was established through external meaning references,
reappearance of a word would reinforce the connection in the reader’s mental lexicon
(Hulstijn et al., 1996). Therefore, compared with the no gloss condition, learners are
more likely to acquire and retain more words from glossed reading (Abraham, 2008;
Bowles, 2004; Hulstijn et al., 1996; Watanabe, 1997; Yanguas, 2009).
What makes glossing effective in vocabulary learning can lie in its more accurate
provision of word meaning than guessing and its deliberate, intentional element that
involves thoughtful processing (Nation, 2009). Glossing was found to be as
effective as text elaboration (O’Donnell, 2009), but Kim (2006) found that explicit
lexical elaboration devices were found to enhance vocabulary learning more than
implicit ones, which suggests that using a more explicit lexical support device such as
glossing is very likely to facilitate vocabulary learning better than employing less
explicit devices such as lexical elaboration. In fact, Watanabe (1997) showed that
the group provided with marginal glosses significantly recalled more word meaning
than the group with implicit lexical elaboration in the form of appositions on both
short-term and long-term bases. It is probably because elaboration may be so
confusing that learners are likely to feel disoriented in the redundant information, and
without clear signals present in the explanation, learner at elementary levels may
mistake the intended elaborations for new information (Lynch, 1996). In avoidance
of such confusion, using glosses as lexical support, especially for those not proficient
enough to recognize lexical elaboration, is more advisable.
Holley and King (1971) also explained the value of glosses from two
perspectives: prompting in paired-associate learning and rehearsal in verbal learning.
In paired-associate studies with prompting, individuals are given a stimulus
immediately followed by a correct response so that they can repeat the response
without making any error. Likewise, if the gloss of a new word is provided in
reading, learners will avoid making wrong meaning guesses and thus learn vocabulary
more efficiently. In verbal learning, rehearsal, either the subvocal (covert) or vocal
(overt) repetitions of a target item, is considered highly facilitative to learning, which
suggests that the more times a word is looked up in its gloss, the better it will be
retained. Based on the two perspectives, Holley and King assumed that placing
glosses at a position that was neither too far from target words nor too close to them
(i.e. in the margin) was very likely to facilitate word retention because learners could
not only have timely access to the prompt but engage in some brief rehearsal as well.
Nation (2009) also argued for the effectiveness of marginal glosses over other
placements of glosses and Jacobs, Dufon and Fong’s (1994) survey showed that 94%
of their participants preferred glosses placed in the margin, with only 6% preferring
glosses at the bottom of the page or at the end of the passage. However, empirical
studies did not find any significant advantage of placing glosses in the margin on
vocabulary learning. For example, Holley and King themselves did not find that the
location of glosses made any significant difference in word learning probably due to
lack of rigid time control, the use of the simple reading material, the learners’ easy
access to meanings and no expectation from the students to learn glossed words. In
Cheng and Good (2009), marginal glosses even consistently yielded worse vocabulary
gain than in-text glosses or glosses with examples on another separate page across
immediate, one-week and 2-week delayed posttests.
Since glossing may increase verbal rehearsal, it has a potential to enhance word
learning in fewer exposures. When glosses are provided, multiple readings can be
triggered in a single word exposure (Rott, 2007). For example, a word glossed in the
margin is very likely to be read three times by students: see it in the passage for the
first time, go to the gloss to understand its meaning for the second time, and move
back to the passage to check how the meaning fits the context for the third time
(Watanabe, 1997). Such behavior of going back and forth between target words and
glosses can stimulate learners “to perform lexical processing, which may contribute to
the retention of the words” (Nagata, 1999, pp. 469-470). That may partially explain
the reason why glossing could increase the probability of one-exposure incidental
learning from 3% to 21% (Chun & Plass, 1996). Rott (2007) also reported that when
words were glossed once initially, reading one-exposure words could be as effective
as reading those occurring four times in acquiring and retaining word knowledge
across 4 to 6 weeks. Therefore, although learners need to encounter a word at least
six times to acquire it incidentally from unglossed texts (Ghadirian, 2002; Huang,
2007; Liu, 2002; Nation, 2001; Rott, 1999), the estimate of minimal exposure
frequency is likely to be revised under the glossing condition.
Many studies have observed learners’ positive attitudes toward glossing.
Jacobs et al. (1994) showed that 98.7% of their participants preferred the presence to
the absence of glosses, and on average, Jacobs’s (1994) participants rated 5.47 out of
6 points on the usefulness of glossing. Cheng and Good’s (2009) survey also
showed that 75% of their subjects experiencing the glossing treatments held a positive
view of glosses in facilitating text comprehension and vocabulary learning.
Likewise, in Bell and LeBlanc’s (2000) exit survey, all of the participants in the L1
gloss group and 74.2% of the L2 gloss group responded that glossing helped their text
comprehension, and 80.5% of all the participants in both groups desired more
vocabulary help for the unglossed words. Ko (2005) showed that 94% of her
participants preferred to read glossed materials as well. In explanation of why
glossing was favored, her participants expressed that glosses facilitated content
understanding, reduced long interruptions from dictionary consultation, supplied the
most appropriate contextual definitions, promoted incidental vocabulary learning, and
prevented wrong guesses.
Glosses can aid second language reading by facilitating bottom-up processing to
save more of readers’ mental resources for top-down processing (Gettys et al., 2001).
As Ko’s (2005) think-aloud data showed, glosses apparently made the students’
reading processes smoother and faster. While the no-gloss group was observed to
use more low-level strategies to deal with unknown words and to monitor their
uncertain comprehension more frequently, the gloss groups employed more high-level
reading strategies such as making inferences. When there are comprehension
breakdowns, referring to glosses can also be viewed as a repair strategy (Jacobs,
1994). In Rott (2007), there was a general tendency that the more times words were
glossed in their occurrences, the better the learners comprehended the main ideas,
which suggested that “repeated provision of meaning for the same word triggered not
only the comprehension of the target word but also the entire proposition” (p. 191).
Although consulting dictionaries is another way to facilitate word encoding processes,
vocabulary assistance, provided either before reading or during reading, can reduce
the extraneous cognitive load (i.e. cognitive load generated from unnecessary
characteristics of the learning context for content learning) resulting from dictionary
use (Alessi & Dwyer, 2008).
Aside from bottom-up processing, glossing may facilitate the learner-text
interaction. According to Stewart and Cross (1991), glosses can foster an interaction
between the text and the reader’s active stance in reading. Such active involvement
promotes the monitoring of meaning construction. When text and gloss are
processed concurrently, there will be greater text elaboration that leads to deeper
memory because the interaction with glosses may make information held long in
working memory and help readers incorporate prior learning into new knowledge
generated from the text.
From a top-down perspective, looking at glosses may switch learners’ attention
to individual words, disrupt reading flow, and hinder the mental creation of an overall
text representation (Jacobs, 1994). Lomicka’s (1998) qualitative data also indicated
that glosses were used mainly to achieve comprehension at a minimal level such as
translation and paraphrasing, and Johnson (1982) indicated that exposure to
vocabulary assistance contributed to reading comprehension less than prior cultural
knowledge. However, compared with other form-focus tasks, glossing is less
intrusive in reading comprehension. In Keating (2008), those who read a text with
target words bolded and glossed in L1 in the margin answered 83% of five true/false
comprehension questions correctly on average, but another group of learners only got
73% correct on the same comprehension test after reading the same text except that
the target words were blanked and required to be filled in the text based on the L2
definitions, examples, and L1 gloss provided on a separate page. Although the
comprehension scores of the two groups did not differ significantly, the researcher
suggested that the learners’ additional mental effort exerted on the fill-in task might
have interfered with comprehension.
Some empirical studies have shown that glossing significantly improves
comprehension more than reading without glossing. In Davis (1989), 71 university
undergraduates read an L2 story following one of the three conditions: (1) simply
read-write-read with no aid, (2) pre-read questions, comments and word definitions,
and then read the text, and (3) read the text with questions and word definitions
glossed in the margin. After the treatment, all of the participants were required to
recall the text in L1. The results showed that the students in the glosses provided
during and before reading conditions significantly recalled more pausal units (i.e.
linguistic units between natural pauses for breath catching, emphasis or meaning
enhancement, cited from Bell and LeBlanc (2000)) than those in the no aid condition.
Jacobs (1994) found that glossing facilitated the students’ reading comprehension.
In his study, 116 students learning Spanish as a second language split into two groups,
the gloss group and the no gloss group, and read a toy text in which there were target
words relevant to the main idea of the text. The results not only showed that
glossing helped the students significantly recall approximately 30% more idea units
but also demonstrated that overall, the gloss group spent less time reading the text
than the no gloss group. Watanabe (1997) reported that the learners reading the
glossed text answered open-ended comprehension questions significantly better than
those reading the original text. Bowles’s (2004) gloss groups comprehended a text
significantly better than the control group with no access to glosses. Her think-aloud
protocol data showed the participants used glosses mainly to understand the text, not
to learn vocabulary. Abraham’s (2008) meta-analysis indicated that L2 learners who
had access to computer-mediated text glosses consistently scored higher on reading
comprehension than those who did not. Yanguas’s (2009) three gloss groups were
significantly better than the no gloss group on the performance of a multiple-choice
comprehension test.
Other studies have not seen any significant effect of glossing on comprehension.
Stewart and Cross (1991) and Johnson (1982) found that whether learners read a text
with or without glosses did not make any significant difference in reading
comprehension. Joyce (1997) pointed out that the reason why glossing did not
facilitate reading comprehension in Johnson’s study was that the text was contrived
by the researcher and thus could have been tailor-made to match the participants’
vocabulary level, which might make glosses unnecessary in aiding comprehension.
Jacobs et al. (1994) and Joyce (1997) found that learners with glosses provided
generally recalled more idea units in L1 than those without but their difference did not
reach a significant level. Davis and Lyman-Hager (1997) also reported a weak
correlation between the use frequency of glosses and the recall protocol and a
negative correlation between the use frequency of glosses and the multiple-choice
comprehension test. They postulated that there was a ceiling on the effectiveness of
glossing when the passage was so difficult that the students’ limited comprehension
capacities were overloaded by the syntactic complexity of some sentence segments
and too many unfamiliar cultural reference and vocabulary items (660 glossed words
in a 1,754-word passage). In addition, “inaccurate intratextual perception of the
passage apparently had a more powerful effect upon these subjects’ recalls than did
the word definitions that they had consulted in the gloss” (pp. 65-66), which “provides
further evidentiary support for the psychological necessity of representing events in
memory in a manner that conforms with one’s view of reality, in combination with the
interpretation of a text one is reading” (p.67). Miyasako (2002) found that whether
the participants read a text with glosses or not, they answered comprehension
questions almost equally well. The researcher suspected that the comprehension
questions might not be well-designed so the learners could answer the questions
without understanding the glossed words and that glossing might interrupt reading
process. Cheng and Good’s (2009) gloss groups also did not show significant
superiority over the no-gloss group in reading comprehension probably because
glosses directed the learners to pay more attention to word meaning processing than to
text comprehension and detail memorization or because the number of comprehension
questions was too small to make any effective discrimination between subjects.
Although no obvious positive effects of glossing on reading comprehension are
observed in those studies, glossing is not found to severely interfere with text
comprehension, either.
There may be another reason to explain why the above studies did not show any
obvious effect of glossing on reading comprehension: glossing will not influence
reading comprehension significantly unless the unknown word text coverage is high
enough. In Johnson’s own opinion, the percentage of unknown words (around 19%)
in the text might not be high enough to affect her advanced students’ comprehension
and her students might be able to make use of text redundancy to figure out unfamiliar
words and to construct a text based on their background knowledge. Although it is
not known what percentage of unknown words is enough to make glossing effective,
Holley and King (1971) seemed to support that if the density of new words does not
reach a certain level, the performance of learners’ reading comprehension will be
similar regardless of how many new words are glossed. In their study, there was no
significant difference in the multiple choice comprehension test scores between the
learners reading two versions of a text, with each version designed with the coverage
of the glossed words only at 3.33% or 6.67%, respectively. Consistent with
Johnson’s explanation, Miyasako (2002) thought that the low percentage of unknown
words (only 5%) in the experimental text might also make glossing ineffective in
enhancing the learners’ comprehension. According to Miyasako, the result might
confirm Laufer’s (1997) prediction that knowing 95% of the words in a text without
the assistance of glosses enabled learners to read to a satisfactory degree but did not
sufficiently exhibit the effectiveness of glossing in reading comprehension.
However, as Davis and Lyman-Hager (1997) reported that more than 37% of new
words glossed in a text were too overwhelming for learners, an optimal range of
glossing percentage that can facilitate learners’ comprehension may fall between 19%
and 37% in a text.
In addition, language proficiency and text difficulty may be two interrelated
factors that need to be considered in judging the effect of glossing on reading
comprehension. To begin with, the conflicting results between the nonsignificant
effect of glossing observed among the advanced learners in Johnson (1982) and the
significant glossing effect reported among the intermediate learners in Alessi and
Dwyer (2008) on reading comprehension may also lie in the different proficiency
levels of the participants. Ko (2005) also noticed that even when glosses were
provided, the participants still sometimes misinterpreted sentences as a result of their
poor syntactic knowledge or general reading ability. Moreover, Jacobs et al. (1994)
found that there was a stronger association between text recall and glossing for the
higher proficiency learners, which suggested that if a text was too difficult, a massive
amount of glossing was needed but that if it was easy, glossing became superfluous.
Jacobs et al.’s suggestion makes it clear that the effect of glossing is determined
by whether there is an appropriate match between learners’ proficiency level and text
difficulty. When text difficulty is below learners’ proficiency level, glosses are of
little help. For example, Davis and Lyman-Hager’s (1997) multiple regression
analyses indicated a negative correlation between the use frequency of glossing and
the course grade, which meant that those whose general language ability was high
enough to read the passage might consider the reliance on glosses less necessary than
the low proficiency learners. In Taylor’s (2006) comments on Ko’s (2005) study, he
stated that if texts were fairly simple, learners might have no need to consult glosses.
However, when texts are far too difficult for learners, whether glosses are provided or
not does not make too much difference, either. Davis and Lyman-Hager (1997)
explained that having the learners read a text which was too difficult for them could
be one of the reasons why the effect of glosses on their reading performance was
rather limited in their study. As Rott et al. (2002) also added in their notes, when a
large proportion of the text was unintelligible to readers, glosses could hardly
facilitate text comprehension or lexical acquisition. Moreover, Abraham’s (2008)
meta-analysis revealed that glosses did not work effectively with beginners on
comprehension measures probably because learners at an elementary level might need
more than lexical glosses to monitor and facilitate their reading of authentic L2 texts.
Therefore, the above studies seem to suggest that instructors should take learners’
proficiency level and text difficulty into consideration when they implement glossing
in their teaching reading.
Proficiency levels may influence the effectiveness of glossing in vocabulary
learning as well. Abraham (2008) found that glossing had a robust positive effect on
intermediate learners but was the least effective for low proficiency learners.
According to Abraham,
Intermediate learners may possess deeper lexical knowledge allowing them to
connect vocabulary encountered in the glosses more easily to a pre-existing