L2 Vocabulary Learning From Context:

Strategies, Knowledge Sources, and

Their Relationship With Success in
L2 Lexical Inferencing
HOSSEIN NASSAJI
University of Victoria
Victoria, British Columbia, Canada

This study examines the use of strategies and knowledge sources in L2

lexical inferencing and their relationship with inferential success. Data
consist of introspective and retrospective think-aloud protocols of 21
intermediate ESL learners who attempted to infer new word meanings
from context. Analysis reveals that (a) overall, the rate of success was
low even when learners used the strategies and knowledge sources they
had at their disposal, (b) different strategies contributed differentially
to inferencing success, and (c) success was related more to the quality
rather than the quantity of the strategies used. Findings challenge a
unidimensional conception of the role of strategies in L2 lexical
inferencing and instead support an inferencing model that distin-
guishes between strategies and the ability to use them appropriately and
effectively in conjunction with various sources of knowledge in lexical
inferencing. This article discusses the pedagogical as well as theoretical
implications of the Ž ndings for an integrated model of lexical
inferencing.

M any ESOL teachers assume that an important process in learning

new vocabulary is the inference learners make about word mean-
ing when they encounter an unknown word in a text. Indeed, compel-
ling evidence suggests that L1 learners acquire much of their vocabulary
from inferring from context on the basis of multiple clues that might be
available to them. L2 learners’ lexical inferencing and its link with
vocabulary acquisition is not as well understood and therefore has
recently become the focus of much research (de Bot, Paribakht, &
Wesche, 1997; Dubin & Olshtain, 1993; Fraser, 1999; Haynes, 1993; Joe,
1995; Lewis, 1993; Morrison, 1996; Nation, 1990; Paribakht & Wesche,
1999; Parry, 1993; Prince, 1996). In L2 learning, much of this research

TESOL QUARTERLY Vol. 37, No. 4, Winter 2003 645

has failed to provide strong evidence; therefore, L2 professionals need
more information about this potentially important process if they are to
make good decisions about vocabulary teaching. This study examines the
success of intermediate ESL learners’ inferencing when they come across
unknown words in a written text. Through the use of think-aloud
protocols, I obtained evidence about the strategies and knowledge
sources learners rely on during the inferencing process. Knowledge
about these mechanisms helps in developing a theory of L2 vocabulary
learning from context as well as an effective and efŽ cient approach to
teaching and learning vocabulary in L2 classrooms.

L2 LEARNERS’ LEXICAL INFERENCING

Although evidence from L1 research suggests that L1 learners learn
much of their vocabulary from context, the results of L2 research in this
area are inconclusive. There is uncertainty among L2 researchers
regarding the role of lexical inferencing as an efŽ cient L2 vocabulary
learning strategy (Bensoussan & Laufer, 1984; Carter & McCarthy, 1988;
Haynes, 1993; Hulstijn, 1992; Scherfer, 1993). However, many research-
ers would agree that, if successful, inferencing can aid comprehension
and contribute to, if not lead to, immediate learning and retention of
lexical and semantic information about words (Ellis, 1997; Hulstijn,
1992; Paribakht & Wesche, 1999). Yet little is known about the exact
mechanisms underlying successful inferencing and, in particular, how
different strategies and knowledge sources used to infer word meanings
from context relate to ultimate success in lexical inferencing.

Inferencing Strategies in Reading

L2 reading comprehension processes are heavily in uenced by the
efŽ ciency of the lower level textual process (Nassaji, 2002, 2003).
Evidence from studies on L2 reading comprehension suggests that
encountering many unknown words in a text may negatively in uence
the reading comprehension of L2 readers. Unknown words may also
partly account for the observation that L2 readers read texts word by
word (e.g., Bernhardt, 1991; Bernhardt & Kamil, 1995; Carrell, 1988;
Clarke, 1980). Skilled L2 readers, on the other hand, due to their skills in
lower level word identiŽ cation processes and higher level syntactic and
semantic processes, can read and understand the text more efŽ ciently
than unskilled L2 readers (Nassaji, 2003). In addition, research investi-
gating more speciŽ c causes suggests that L2 readers are also weaker in
using effective strategies in reading and dealing with new words than are

646 TESOL QUARTERLY

L1 readers (e.g., Auerbach & Paxton, 1997; Block, 1992; Carrell, 1988;
Devine, 1993; Hudson, 1982; Kern, 1989). Thus, the type of strategies L2
learners use may be related to their ability to comprehend and infer
words successfully from context.
Research suggests that readers use a variety of strategies when they
encounter new words. These strategies include ignoring unknown words,
consulting a dictionary for their meaning, writing them down for further
consultation with a teacher, or attempting to infer their meaning from
context (Fraser, 1999; Harley & Hart, 2000; Sanaoui, 1995). Among
various word-learning strategies, lexical inferencing has been found to
be the strategy most widely used by L2 learners, a process that “involves
making informed guesses as to the meaning of an utterance in light of all
available linguistic cues in combination with the learner’s general
knowledge of the world, her awareness of context and her relevant
linguistic knowledge” (Haastrup, 1991, p. 40).
Fraser (1999) and Paribakht and Wesche (1999) found that lexical
inferencing was the most frequent and preferred strategy their adult L2
learners used to learn the meanings of new words when reading. Fraser
found that lexical inferencing alone accounted for 58% of the cases
where learners encountered a new word. Other strategies were used at a
lower percentage: consulting a dictionary (39%), ignoring (32%), and
not paying attention to the word (3%). Paribakht and Wesche found that
almost 80% of the strategies their university ESL students used in dealing
with new words were lexical inferencing, with all other strategies ac-
counting for about 20% of the learner’s strategy use. Inferencing has also
been reported to be the major processing strategy when learners attempt
to derive and learn idiomatic and Ž gurative meanings in reading.
Cooper (1999) found that 28% of the time, readers used inferring from
context as a strategy to identify the meaning of idioms. This percentage
was higher than that of all other strategies, including analyzing the idiom
(24%), using literal meaning (19%), requesting information (8%),
paraphrasing and repeating (7%), using background knowledge (7%),
and using L1 or other strategies (7%). In view of the dominance of this
strategy in L2 reading, its value for lexical acquisition for ESOL should
be further investigated.

Successful Vocabulary Inferencing

Studies investigating what is involved in inferencing have identiŽ ed
many factors that play an important role in successful inferencing. These
factors include the nature of the word and the text that contains the
word (Paribakht & Wesche, 1999; Parry, 1993), the kind of information
available in the text (Chern, 1993; Haastrup, 1991; Haynes, 1993), the

L2 VOCABULARY LEARNING FROM CONTEXT 647

importance of the word to the comprehension of the text (Brown, 1993),
the degree of cognitive and mental effort involved in the task (de Bot
et al., 1997; Fraser, 1999; Joe, 1995), and the degree of textual informa-
tion available in the surrounding context (Dubin & Olshtain, 1993).
Successful inferencing has been shown to depend heavily on learners’
prior knowledge as well as their ability to make effective use of extratextual
cues (de Bot et al., 1997; Haastrup, 1991). It also has been shown to
depend on learners’ having large vocabulary recognition knowledge, for
example, of around 5,000 word families (Coady, Magoto, Hubbard,
Graney, & Mokhtari, 1993; Laufer, 1997), and their ability to compre-
hend most of the words, at least 95%, in the text (Hirsh & Nation, 1992;
Laufer, 1988, 1989; Liu & Nation, 1985).
Successful inferencing strategies range from those related to the
internal structure of the words and their components, including the
various phonemic, phonetic, graphemic, and morphemic clues (Chern,
1993; Haynes 1993), to the information about the syntactic and semantic
relationship among words, and even to the various higher order,
extratextual, and discoursal clues (de Bot et al., 1997; Huckin & Bloch,
1993). In a recent study, de Bot et al. (1997) found that L2 readers used
knowledge sources ranging from knowledge of grammar, morphology,
phonology, and knowledge of the world, to knowledge of punctuation,
word association, and cognates.
Based on the results of an exploratory study on inferring word
meanings from context by three intermediate Chinese students, Huckin
and Bloch (1993) proposed a cognitive processing model of L2 lexical
inferencing. The model incorporates two separate components: a gen-
erator and evaluator component and a metalinguistic control compo-
nent. The generator and evaluator component includes numerous
interconnected knowledge-based modules, such as a vocabulary knowl-
edge module, a text schema module, a syntax and morphology module,
and a text representation module. The function of the generator and
evaluator component is to generate and evaluate hypotheses about the
meaning of the word encountered based on the various knowledge
sources in the module. The metalinguistic control component includes a
sequence of serial and parallel decision-making steps that the learner
goes through when trying to generate and test hypotheses. These
processes help the learner decide when and how to proceed and seek
help from context and various sources of knowledge available.
Huckin and Bloch (1993) showed that learners appealed to various
knowledge sources and employed various cognitive strategies in their
attempts to infer word meanings from context. They also provided
evidence for the degree to which these processes were used and how they
related to success. Paribakht and Wesche (1999) found that their
university ESL readers also appealed to a variety of linguistic and

648 TESOL QUARTERLY

nonlinguistic knowledge sources when attempting to derive the mean-
ings of new words from context. They also found that the use of these
knowledge sources was affected by a number of other factors, such as
type of text (summary vs. question), text characteristics (e.g., topic,
content, genre), and word characteristics (e.g., verb, adjective, noun).
This approach to explaining inferences is consistent with that of
Pressley, Borkowski, and Schneider (1987), who distinguish between
learners’ cognitive strategies and their knowledge base, but go a step
further to explain the relationship between strategy use and success.
Pressley et al. conceived of Ž ve factors that were considered important in
successful strategy use: (a) having a wide repertoire of general as well as
domain-speciŽ c strategies; (b) having the ability to use strategies appro-
priately and in appropriate contexts; (c) having an extensive task-
relevant knowledge base, ranging from general knowledge of the world
to knowledge about speciŽ c strategies and their causes of success and
failure; (d) being able to automatically execute and coordinate the use
of strategies with various knowledge sources; and (e) having an aware-
ness that, although success is related to efforts, efforts alone may not be
enough. In this context, Pressley et al. (1987) made a distinction
between “effort attribution and strategic effort attributions” (p. 104),
according to which, good strategy users realize that what matters is not
just efforts but efforts that are strategic and task matching.

Strategies, Knowledge, and Success

A fundamental question that has remained unanswered throughout
this research concerns the relationship between the range of strategies
and knowledge sources learners use and their success in lexical
inferencing. Most of the studies conducted so far on the role of learners’
strategies have been descriptive, so to what extent the strategies assist
them in deriving word meaning from context is not known. To address
this question and better understand these processes and their contribu-
tions to success in inferencing, the present study was designed to
determine (a) how successfully intermediate ESL learners infer word
meanings from context in a reading text, (b) what strategies and
knowledge sources they use to do so and to what extent, and (c) whether
there is any relationship between the range of strategies and knowledge
sources they use and their lexical inferencing success.

METHOD
I chose introspective methods for the research because the object of
investigation was the set of strategies and knowledge deployed during

L2 VOCABULARY LEARNING FROM CONTEXT 649

the reading process. I carefully selected the participants and reading
passage to ensure an appropriate level match.

Participants

Twenty-one adult ESL learners (10 males, 11 females) voluntarily

participated in the study. They represented Ž ve different language
backgrounds, including Arabic (2), Chinese (8), Persian (6), Portuguese
(2), and Spanish (3). They had all recently arrived in Canada and were
enrolled in a 12-week intermediate ESL program to improve their
English. All had met Level 4 of the Canadian Language Benchmark test
(e.g., Citizenship and Immigration Canada, 1996), which evaluates the
ESL proŽ ciency of adult newcomers to Canada for the purpose of
instructional placement. Using task- and performance-based criteria, the
test assesses learners in three language skills areas (speaking and
listening, reading, and writing) on a continuum of 1–12 benchmarks
(see Peirce & Stewart, 1997).

The Reading Passage

Because the research concerned successful inferencing, I had to Ž nd
a text that contained a minimum number of words that the participants
would know (e.g., Laufer, 1989; Liu & Nation, 1985; Nation, 2001). As
mentioned earlier, research suggests that readers should know a high
percentage (at least 95%) of the words in the text in order to be able to
infer successfully (Liu & Nation, 1985). The text also had to match the
comprehension ability of the intermediate readers. Among the passages
I examined, including those used in previous research, I chose the one
used in Haastrup’s (1991) study (see the Appendix). The passage
contained 374 words, 10 of which were target words that I used to focus
on inferencing strategies. The target words were all content words
consisting of four nouns, four verbs, and two adjectives.
I organized a panel of three ESL teachers who were working with the
students to determine the appropriateness of the passage in terms of its
reading level and content. To further check the appropriateness, I
piloted the passage with a group of ESL students assumed to be similar to
the participants in the main study with respect to language proŽ ciency
and level of reading comprehension measured in terms of the Canadian
Language Benchmark. The students were asked to underline any words
they did not know when reading the passage and further conŽ rm them
in retrospective interviews. The percentage of unknown words was then
calculated for each student by dividing the total number of words they

650 TESOL QUARTERLY

reported as unknown by the total number of words in the passage
multiplied by 100.
The pilot study revealed that the students had a good comprehension
of the text. Their mean scores on a comprehension test consisting of 10
questions was 7.6, and the number of unknown words for them,
including the 10 target words, ranged from 4.27% to 2.67% (which was
less than 5%). This percentage was later found to be very close to the
percentage obtained in the main phase of the study (4.01% to 1.06%).

Data Collection
Data were collected in individual sessions in which the researcher met
with each learner in a quiet room for about 45–60 minutes. To guarantee
the equality of procedures, I conducted all the data collection sessions.
Think-aloud techniques, which were used as the main data collection
tool, are procedures requiring participants to verbalize and report the
content of their thoughts while doing a task. Despite some criticism of
the think-aloud procedure, it is a common methodology used in strategy
research (Ericsson & Simon, 1993; Pressley & Af erbach, 1995). More-
over, such data, if not a true re ection of, have at least been assumed to
be associated with the processes participants use in processing language
(Olson, Duffy, & Mack, 1984). Introspective and immediate retrospective
reports were also used in this study. However, the data about the use of
strategies and knowledge sources derive mainly from the introspective
reports because they involve more direct and online reporting of what
learners are doing at the time of the task than do retrospective reports,
which ask learners to recall what they had done before (Olson, Duffy, &
Mack, 1984). The retrospective reports were used mainly to Ž nd out if
the learners had additional comments on their familiarity with the words
or their inferencing processes.
At the beginning of each session, I informed learners of the general
purpose of the study. They then participated in a training and practice
period in which I introduced them to the think-aloud procedure and
explained how they were to verbalize their thoughts. Participants also did
role plays using pictures as well as similar reading texts for training
purposes. After the training and practice period, when I felt that the
learners knew how to think aloud, I presented the reading passage and
instructed them to read it out loud. As they encountered each italicized
target word in the text, I asked them to try to infer its meaning from the
context, verbalizing and reporting whatever came to their mind. I also
asked them to underline and try to infer the meaning of any other words
whose meaning they did not know. I advised them that they could refer
back at any time to an unknown word to try to infer its meaning again.

L2 VOCABULARY LEARNING FROM CONTEXT 651

When they Ž nished reading, I asked them to review the passage and
make any additional comments they wished about any new words and
their thinking processes. For transcription purposes, I audiotaped the
think alouds and retrospective reports.

RESULTS
Results indicate that the intermediate-level ESL learners were not very
successful at inferring word meanings from context in a reading text.
More speciŽ cally, the results show what strategies and knowledge sources
these learners used during the process of inferencing in addition to the
relationship between strategies, knowledge, and inferencing success.

Success of Lexical Inferencing

To determine the degree to which learners were successful at
inferencing, I and an experienced native-English-speaking teacher inde-
pendently rated their responses to each of the unknown words using a
3-point scale (2 = successful, 1 = partially successful, 0 = unsuccessful ).
Successful inferencing was deŽ ned as responses that were semantically,
syntactically, and contextually appropriate. A successful response could
be a word representing an accurate semantic meaning of the target word
(e.g., a synonym) or an appropriate deŽ nition of the word. Because it
was possible to arrive at a completely accurate semantic meaning of a
word and yet associate the word with a wrong syntactic category (Gass,
1999), for rating purposes, we classiŽ ed responses that were semantically
appropriate but syntactically deviant, or vice versa, as partially successful.
In order not to underestimate learners’ success, if the meaning or the
deŽ nition they provided made sense in the context but when judged out
of context was not the meaning of the word, we still considered the
response partially successful. In cases where the response did not meet
any of the above conditions, we considered it unsuccessful.
During initial rating, we obtained an interrater agreement of 94%. We
then resolved disagreements through subsequent discussion to reach
100% interrater agreement on all items.
The question about success was addressed by looking at individual
lexical items. Because the 21 participants each responded to the 10
target lexical items, the initial data set contained 210 responses of
interest. However, the initial set of responses was reexamined in view of
some participants’ indications during the reading and retrospective
interviews that they knew some of the target words. Omitting the
responses for individual participants on the speciŽ c words they indicated

652 TESOL QUARTERLY

knowing reduced the number of target words to 199, for which responses
could be interpreted as inferencing of an unknown word.
Of the total 199 inferential responses, 51 (25.6%) were successful, 37
(18.6%) were partially successful, and 111 (55.8%) were unsuccessful.
The percentage of unsuccessful inferences indicates that more than half
of the time, students went completely wrong in their attempts to infer
the meanings of the new words from context. A chi-square test con-
ducted on the raw frequency of the data revealed that the frequency of
the unsuccessful versus the successful and partially successful inferences
was signiŽ cantly greater than the expected distribution would be ( 2 =
46.59, df = 2, p .0001).
An item-by-item analysis of the individual words indicated that the
degree of successful inferencing for each of the individual words was
quite low, ranging from 9.5% to 38.1%, with the mean percentage for all
the items being 25.6% (see Table 1). Results also showed that where at
least 95% of the words in the text were familiar, there was still a
signiŽ cant correlation between the learner’s inferential success (ob-
tained by adding up the success scores the learner gained for each
unknown word in the 3-point scale system) and the percentage of the
remaining unknown words in the surrounding context (obtained by
calculating the number of unknown words for each individual learner
and dividing them by the total number of words multiplied by 100) (r =
2.46, p = .03). The correlation coefŽ cient is negative because success was
correlated with the density of unknown words, thus indicating that the
higher the proportion of unknown words in the surrounding context,
the lower the likelihood of success. This Ž nding highlights the impor-
tance of lexical density, or the ratio of known to unknown words in the
context of an unfamiliar word, and suggests that with a large proportion
of words already familiar in the text (at least 95%), Ž ne-tuned knowledge
of the remaining words in the context is still a crucial factor in the
successful inferencing of unknown words.
Learners’ success seemed also to be related to the physical form of the
words and how they looked. Among the items guessed, the most difŽ cult
one was permeated, followed by squalor, afuence, and waver. The percent-
age of successful inferencing for these items ranged from 9.5% to 23.8%
(the mean percentage of correct responses being 16.65%). On average,
83.35% of the time these words were either unsuccessfully or partially
successfully inferred. The problem with these words may be related to
their misleading nature and confusion with similar-looking words. An
analysis of the learners’ protocols showed that many of the learners who
guessed the meanings of these words wrongly interpreted them by
confusing them with other similar-looking but semantically unrelated
words. For example, many of these learners mistakenly related permeated,
to meat, waver to wave, and afuence to inuence. Although some of these

L2 VOCABULARY LEARNING FROM CONTEXT 653

 TABLE 1
Successful, Partially Successful, and Unsuccessful Inferences for Unknown Words

Inferences
Partially
Successful successful Unsuccessful
Total number
Unknown words of responses n % n % n %

1. permeated 21 2 9.5 1 4.8 18 85.7

2. squalor 21 3 14.3 4 19.0 14 66.7
3. af uence 21 4 19.0 5 23.9 12 57.1
4. waver 21 5 23.8 9 42.9 7 33.3
5. contract 19 5 26.3 0 0.0 14 73.7
6. sewage 21 6 28.6 8 38.1 7 33.3
7. curative 19 6 31.6 4 21.1 9 47.4
8. assessing 15 5 33.3 0 0.0 10 66.7
9. hazard 20 5 35.0 7 25.0 8 40.0
10. unfathomable 21 8 38.1 1 4.8 12 57.1
Total n and mean % 51 25.6 37 18.6 111 55.8

words may be etymologically and historically related, they are currently

used in quite different semantic Ž elds. The following excerpt from the
protocol of one of the students who was attempting to infer the meaning
of the word permeated illustrates the typical problem with such words.

Permeated . . . meated . . . is a kind of meat . . . I think there are something

related with meat and the body . . . permeated with such things as chemical
and radio active.

This Ž nding suggests that the words’ appearance and their similarity with
other unrelated words may be a major source of problems in inferring
word meanings from context (see also Bensoussan & Laufer, 1984;
Laufer & Sim, 1985). It may also suggest that precise inferencing of
words may be related to how accurately learners recognize and decode
the orthographic form of the word (Ryan, 1997).

Strategies and Knowledge Sources

Following Pressley, Borkowski, and Schneider’s (1987) distinction
between learners’ cognitive strategies and knowledge base and Huckin
and Bloch’s (1993) distinction between knowledge sources and
metalinguistic decision-making strategies in their model of inferring
word meanings from context, this study makes a distinction between the

654 TESOL QUARTERLY

use of knowledge sources and strategies. I deŽ ned strategies as conscious
cognitive or metacognitive activities that the learner used to gain control
over or understand the problem without any explicit appeal to any
knowledge source as assistance. In contrast, I deŽ ned appeals to knowledge
sources as instances when the learner made an explicit reference to a
particular source of knowledge, such as grammatical, morphological,
discourse, world, or L1 knowledge.
To determine the different types of strategies and knowledge sources
learners used, I had all the introspective think-aloud protocols initially
transcribed verbatim before being carefully examined and coded twice,
once by me and then by a colleague. For coding categories, I consulted
the literature on vocabulary learning and lexical inferencing strategies
(e.g., de Bot et al., 1997; Haastrup, 1991; Huckin & Bloch, 1993; Parry,
1991, 1993; Schmitt, 1997). However, the coding scheme I used derives
mainly from the data and re ects the thinking of the learners participat-
ing in the study rather than from pre-existing categories imposed on the
data. Coding involved reading and rereading the protocols and identify-
ing in an inductive manner the kind of inferencing strategies and
knowledge sources used. I established the reliability of the coding by
calculating an intercoder agreement on a sample of 20% of the data,
selected from every Ž fth participant. The intercoder agreement for that
20% of the data was 89%. The second coder and I resolved discrepancies
through discussion to achieve 100% agreement. I then examined and
coded the remaining data.
I identiŽ ed a total of 11 categories of strategy types and knowledge
sources. Knowledge sources used included grammatical knowledge,
morphological knowledge, knowledge of L1, world knowledge, and
discourse knowledge. Strategy types included repeating, verifying, ana-
lyzing, monitoring, self-inquiry, and analogy. Repeating was further
divided into, and coded as, word repeating when the learner repeated
the word alone and as section repeating when the learner repeated a
bigger section in which the word had occurred, such as the clause or the
sentence. Table 2 presents the categories of knowledge sources and
strategy types identiŽ ed, along with deŽ nitions and examples from the
transcripts.
Of all the knowledge sources, students used world knowledge most
frequently (46.2%), followed by morphological knowledge (26.9%).
They did not use grammatical knowledge very widely (11.5%), and they
used discourse knowledge (8.7%) and L1 knowledge (6.7%) least
frequently (Table 3). The Ž ndings also showed variation among students
in terms of the types of knowledge sources they used. For example,
whereas 71% of the students used world knowledge, only 33% used
discourse knowledge. Due to the number of participants in the study, I
did not further analyze individual variations and their consequences.

L2 VOCABULARY LEARNING FROM CONTEXT 655

 TABLE 2
DeŽ nitions and Transcript Examples of Knowledge Sources and Strategies
Students Used to Make Lexical Inferences

Knowledge source DeŽ nition Example

Grammatical Using knowledge of grammatical “curative effect of medicine.” . . .

knowledge functions or syntactic categories, According to it is adjective . . .
such as verbs, adjectives, or adverbs mmm . . . it is something before
the effect.
Morphological Using knowledge of word “unfathomable.” . . . I don’t know
knowledge formation and word structure, unfathomable . . . ‘un’ . . . it is
including word derivations, negative of fathomable.
in ections, word stems, sufŽ xes,
and preŽ xes
World knowledge Using knowledge of the content or I think the “sewage” is like
the topic that goes beyond what is something that is produces, . . .
in the text because of some of some illness
that these people have, they are
talking about some problems that
the people have in Africa.
L1 knowledge Attempting to Ž gure out the “assessing . . . .” I forgot the idea
meaning of the new word by . . . Oh I got the meaning . . . I got
translating or Ž nding a similar it in Chinese, like if I want to apply
word in the L1 for position of professional
engineer I should pass the the
assessment of some organizations
like the professional engineering
organization.
Discourse Using knowledge about the “far from being mysterious and
knowledge relation between or within unfathomable . . .” unfathomable is
sentences and the devices that like mysterious something that is
make connections between the not known for everybody. Because
different parts of the text they are talking about the causes of
some disease and they they are
saying they are mysterious.

Continued on page 657

This question can be addressed in future research using more in-depth

case studies of individual learners (see Huckin & Bloch, 1993; Parry,
1991, 1993).
Of all the strategies, students used repeating (including word repeat-
ing and section repeating) most frequently, accounting for about two
thirds (63.7%) of the strategies used. Of the two types of repeating,
students used word repeating much more frequently than section
repeating (39.7% vs. 24%). Other strategies students used much less
frequently were analogy (8.5%), verifying (7.9%), monitoring (7.2%),
self-inquiry (7.2%), and analyzing (5.5%) (see Table 4). Moreover,
whereas all learners used repeating, only 66.6% used self-inquiry, 61.9%

656 TESOL QUARTERLY

 TABLE 2 (Continued)
DeŽ nitions and Transcript Examples of Knowledge Sources and Strategies
Students Used to Make Lexical Inferences

Strategies DeŽ nition Example

Repeating Repeating any portion of the text, “our beliefs waver . . . waver . . .
including the word, the phrase, or waver . . . .” May be . . . waver is
the sentence in which the word has something “beliefs waver . . .”
occurred
Verifying Examining the appropriateness of “but when we ourselves become ill,
the inferred meaning by checking our beliefs waver . . .” our beliefs
it against the wider context change . . . change . . . when we
become ill our beliefs change . . .
yeah.
Self-inquiry Asking oneself questions about the “hazards . . .” should it be pollution
text, words, or the meaning already according to the sentence?
inferred “pollutions?” No no . . . it should
not be that . . . it may be
something different.
Analyzing Attempting to Ž gure out the “and smell of sewage in their
meaning of the word by analyzing noses . . .” sew, age . . . should be a
it into various parts or components kind of smell. But sew is
something, may be it is a kind of
plant, wood.
Monitoring Showing a conscious awareness of “contract some of the serious and
the problem or the ease or infectious diseases . . .” contract . . .
difŽ culty of the task I think contract is is make from
boss and the staff . . . contract . . .
yes . . . this is easy . . . this easy . . .
maybe it’s difŽ cult, I am not sure.
Analogy Attempting to Ž gure out the “squalor . . .” may be it is like
meaning of the word based on its square . . . square . . . It should be
sound or form similarity with other something like that.
words

used verifying, 79.19% used analogy, 61.9% used analyzing, and 80.95%
used monitoring, suggesting that not all the students used all the
strategies and that there was variation among students in terms of types
of strategies used.

Relationship of Strategies to Success

To determine the relationship between successful inferencing and the
strategies and knowledge sources used, I Ž rst calculated the percentage
of successful, partially successful, and unsuccessful inferences for each

L2 VOCABULARY LEARNING FROM CONTEXT 657

 TABLE 3
Students’ Use of Knowledge Sources

Knowledge source n %

World knowledge 48 46.2

Morphological knowledge 28 26.9
Grammatical knowledge 12 11.5
Discourse knowledge 9 8.7
L1 knowledge 7 6.7
Total 104 100

strategy type and knowledge source. I then calculated a mean of success

for each strategy type and knowledge source. To that end, I divided the
sum of the scores obtained for success in inferencing the target words by
the total frequency of each strategy or each knowledge source used to
infer the meaning of those words.
Among the knowledge sources used (see Table 5), morphological
knowledge had the highest mean of success (.93), followed by world
knowledge (.83). These knowledge sources were associated with more
successful inferences than other knowledge sources (morphological
knowledge: 35.7% successful; world knowledge: 29.2% successful). L1
knowledge had the lowest mean of success (14.3%) and was least
associated with successful inferences. Statistical comparison of the means
of the knowledge sources and a two-way chi-square test on the frequency
of these knowledge sources and the degree of their success revealed no
statistically signiŽ cant differences in the contribution of different knowl-
edge sources—analysis of variance (ANOVA): F(4, 99) = .355, p = .84; 2 =
2.53, df = 8, p = .96. This indicates that whereas some of the knowledge

TABLE 4
Students’ Use of Strategies

Strategies n %

Word repeating 187 39.7

Section repeating 113 24.0
Analogy 40 8.5
Verifying 37 7.9
Monitoring 34 7.2
Self-inquiry 34 7.2
Analyzing 26 5.5
Total 471 100

658 TESOL QUARTERLY

 TABLE 5
Knowledge Sources and Inferential Success

Inferential success
Partially
Successful successful Unsuccessful Total
Knowledge M of
source success SD n % n % n % n %

Grammatical .67 .89 3 25.0 2 16.7 7 58.3 12 100

Morphological .93 .90 10 35.7 19 21.4 12 42.9 28 100
L1 .57 .79 1 14.3 6 28.6 4 57.1 7 100
World .83 .86 14 29.2 6 25.0 22 45.8 48 100
Discourse .78 .83 2 22.2 3 33.3 2 44.4 9 100
Total .82 .86 30 28.8 25 24.0 49 47.1 104 100

sources contributed more to successful inferencing than others, success

did not depend much on what kind of knowledge source was used.
Table 6 displays the means as well as the percentages of successful
inferences for the different types of strategies used. Among strategies,
verifying, self-inquiry, and section repeating were associated with higher
means of success than were other strategies. The means of success for
these strategies were 1.51, 1.15, and 1.05, respectively. Verifying and self-
inquiry were associated with the highest means of success and the
greatest proportion of successful inferences as compared to other
strategies (67.6% and 52.9%, respectively). Of the two subcategories of

TABLE 6
Types of Strategies and Inferential Success

Inferential success
Partially
Successful successful Unsuccessful Total
Knowledge M of
source success SD n % n % n % n %

Word repeating .66 .84 45 24.1 33 17.6 109 58.3 187 100
Section
repeating 1.05 .91 50 44.2 19 16.8 44 38.9 113 100
Verifying 1.51 .77 25 67.6 6 16.2 6 16.2 37 100
Analogy .40 .71 5 12.5 6 15.0 29 72.5 40 100
Self-inquiry 1.15 .96 18 52.9 3 8.8 13 38.2 34 100
Analyzing .73 .92 8 30.8 3 11.5 15 57.7 26 100
Monitoring .94 .92 13 38.2 6 17.6 15 44.1 34 100
Total .86 .91 164 34.8 76 16.1 231 49.0 471 100

L2 VOCABULARY LEARNING FROM CONTEXT 659

repeating, section repeating in comparison to word repeating was
associated with a higher mean of success (1.05 vs. .66) and a higher
percentage of successful inferencing (44.2% vs. 24.1%). Analogy was
associated with the lowest mean of success (.40) and the least proportion
of successful inferences (12.5%). An ANOVA conducted on the means of
success for strategies and a two-way chi-square test conducted on the
strategy types and their proportion of success revealed a statistically
signiŽ cant difference in the contribution of the different strategies—
ANOVA: F(6, 464) = 8.85, p .001; 2 = 51.10, df = 12, p .001. These
Ž ndings indicate that success in inferencing depended signiŽ cantly on
what kind of strategy learners used.
Further analysis of the data showed, however, that none of the
strategies and knowledge sources, even the most successful one, was
100% successful alone. In fact, none of the means of success reached 3
(i.e., fully successful). I also found variations in strategy use both within
individual participants and across items guessed. This suggests that
successful inferencing may be the result not of using one strategy or
knowledge source over and above other strategies but of the extent to
which various kinds of strategies and knowledge sources converge and
link. Due to the number of participants in the study, however, I did not
analyze such individual variations. This question can be addressed in
future research using more in-depth case studies of individual learners
(see Huckin & Bloch, 1993; Parry, 1991).
Another point concerns the relationship of success with the number
versus the kind of strategies used. Out of the total number of strategies
used (n = 471), about half (231) were associated with unsuccessful
inferencing. However, out of the total number of strategies associated
with unsuccessful inferencing (231), about half (109) were word repeat-
ing, with fewer instances of the other more elaborative strategies. In the
case of successful inferences, on the other hand, relatively more in-
stances of section repeating (50), verifying (25), and self-inquiry (18)
can be seen. These Ž ndings suggest that success in inferencing may not
be related as much to the quantity as to the quality of the strategies used
(see also Vann & Abraham, 1990).

DISCUSSION
The results of this research offer some insight into the process of
inferencing vocabulary meaning during L2 reading and guidance for
teaching vocabulary through reading.

660 TESOL QUARTERLY

Explanation of L2 Inferencing
The Ž ndings of this research can be explained through a combination
of the cognitive model of vocabulary learning from context developed by
Huckin and Bloch (1993) and the model of good strategy users (GSU)
by Pressley et al. (1987). Huckin and Bloch’s model includes knowledge
sources and strategies as the explanatory factors for vocabulary inferencing
in general, whereas the GSU model attempts to explain the differential
effects of strategy use.

Knowledge Sources

Among knowledge sources, students used general knowledge of the

world most frequently, indicating that they were very dependent on this
kind of knowledge when inferencing word meanings from context and
that this knowledge provided an important knowledge base for their
judgments. Students did not use grammatical knowledge very often,
which may indicate that information about the grammatical function of
the words was not something they needed to infer the meanings of the
new words from context. However, even when used, this knowledge
source was not associated with much success.
This Ž nding is consistent with what Parry (1993) found about the
usefulness of grammatical knowledge in a longitudinal study that investi-
gated how a Japanese student extracted the meaning of unknown words
from an academic text. Parry found a greater semantic than syntactic
relationship between the words and their inferred meanings. In her
study, the learner almost always had been able to infer the syntactic
information about the unfamiliar words even in cases where the inferred
meaning was a wrong meaning. This, then, may suggest that knowing the
grammatical function of a word or that a word belongs to certain
syntactic categories such as verbs, adjectives, or adverbs may not lead to
an accurate semantic representation of the word in context. Interpreted
that way, it may provide support for the idea that the process of
extracting an accurate conceptual meaning about an unknown word
from context is a lemma construction process (de Bot et al., 1997), in
which word meaning may not develop simply from resort to the syntactic
information about the word, something which can be done very easily by
most skilled learners. Rather, it may develop from a Ž nal semantic
speciŽ cation, which should be accessed and matched with the concep-
tual information existing in the learner’s conceptual system.

L2 VOCABULARY LEARNING FROM CONTEXT 661

Strategies

The strategies learners used included repeating, verifying, monitor-

ing, self-inquiry, analyzing, and analogy. Within Huckin and Bloch’s
(1993) model, the role of these strategies can be seen as examples of
cognitive decision-making processes learners use while interacting with
the text and formulating and testing their word meaning hypotheses.
Results showed that learners used repeating as the major strategy. The
fact that they used this strategy very often is not surprising because
repeating can aid both comprehension of and re ection on the content.
However, of the two types of repeating, section repeating led to signiŽ -
cantly higher means of success and was more associated with successful
inferencing than word repeating. The advantage of section repeating
may in part relate to the role of this strategy in assisting the learner to
relate the word to the phrase or sentence in which it has occurred and to
use the potential cues available in those contexts. Thus, it seems to
support the idea that success in inferencing is more related to the degree
to which the learner’s strategy in processing the word is global than local
(Dubin & Olshtain, 1993; Haynes, 1993; Huckin & Coady, 1999; Nation,
1990).
Among strategies, however, verifying and self-inquiry were related
more to successful inferencing than other strategies. This Ž nding seems
to provide evidence for the important role of these metacognitive
strategies in lexical inferencing. The advantage of these strategies may in
part relate to the role of these processes in assisting learners to examine
the accuracy of their guesses and revise or reevaluate them against the
information provided in the wider context. Self-inquiry, in particular,
may help learners to concentrate on their inferences by actively ques-
tioning them and then looking for alternative solutions in cases where
they Ž nd them to be wrong. Several research studies in L1 have shown
the potential beneŽ ts of using self-inquiry in reading and understanding
whole texts (Andre & Anderson, 1978; Frase & Schwartz, 1975; Singer &
Donlan, 1982). These studies suggest that self-inquiry may lead to more
active processing of materials being read and the activation of relevant
background knowledge (Wong, 1985). It is possible that the use of this
strategy in lexical inferencing may do the same. It may also make
students more conscious of the problem and then better enable them to
research solutions.
Analysis showed, however, that although some of the strategies were
more related to successful inferencing than others, the overall contribu-
tion of these strategies was partial and limited. This suggests that success
in inferencing may not depend just on the use of certain strategies but
also on how effectively the use of strategies is combined and coordinated

662 TESOL QUARTERLY

with the use of other sources of information in and outside the text. In
the present study, for example, students used word repeating and
analogy very often, but these efforts were not associated with much
success. Within Pressley et al.’s (1987) framework, these strategies could
be taken as examples of nonstrategic attempts.
Other examples of nonstrategic attempts were evident in the heavy
reliance on word repeating that may re ect a narrow word-based
approach to lexical inferencing used by these intermediate-level ESL
students, which may have contributed to the overall low rate of success as
well (see Qian, 1998, for similar results). Indeed, repeating the word
itself can be a helpful strategy, particularly, when it can help the student
access the meaning of a word through eliciting a phonological or
orthographic representation of the word in the lexicon (Ellis & Beaton,
1993). However, this strategy turned out to be ineffective in this study,
partly because most of the target words in the text were low-frequency
words and words that were completely unknown to the students. When
the word is completely unknown, students are unable to get much out of
repeating the word itself simply because there is little conceptual
reference for the word in their lexicon.
Analogy can also be a helpful strategy and can sometimes be used as a
means of retrieving the meaning of a word through associating it with
other neighboring words. However, it may fail if there are pseudo-similar
words in the text and if students fail to distinguish the word from those
that are deceptively similar (Bensoussan & Laufer, 1984; Haynes, 1984,
Huckin & Bloch, 1993). In such cases, analogy may lead to what Huckin
and Bloch (1993) called a “mistaken ID” (p. 166), a process whereby the
student takes a word for another similar-looking word.

Pedagogical Implications
This study demonstrated that the ESL students experienced difŽ culty
in successfully inferring the meanings of unknown words from context,
even though they reported knowing most of the words in the text and
used the strategies and knowledge sources they had at their disposal.
This Ž nding adds to and conŽ rms the literature in both L1 and L2
learning that inferring new word meanings from context is not an easy
task (e.g., Bensoussan & Laufer, 1984; Kelly, 1990; Prince, 1996; Schatz &
Baldwin, 1986; Shu, Anderson, & Zhang, 1995). The low mean percent-
age of correct inferences found in this study (25.6%) seems to replicate
the Ž ndings of Bensoussan and Laufer’s (1984) study, in which context
helped L2 learners with guessing only 24% of the words with no positive
effect on the remaining 76% of the words. These Ž ndings call into

L2 VOCABULARY LEARNING FROM CONTEXT 663

question the efŽ cacy of contextual inferencing in L2 vocabulary acquisi-
tion and consequently lend support to theories that posit a central role
for instructed L2 vocabulary learning.
This Ž nding suggests that in ESL learning classrooms, students should
not be pushed to rely too much on context to learn the meanings of new
words. Teachers should devote part of the class time to identifying,
deŽ ning, and explaining the new words to the students. I do not mean,
however, to downplay the role of context in L2 vocabulary learning. Two
distinctions are important here: Ž rst, between using context for learning
the meaning of words based on a single exposure and using context as a
means of providing multiple exposures to the same word, and second,
between using context as a means of generating new knowledge and as a
means of consolidating known knowledge. Although the results of this
study may call into question the efŽ cacy of context in the former sense
(i.e., generating new knowledge and learning words based on a single
exposure), there seems to be little question about the importance of
context in providing frequent exposure and a framework for consolidat-
ing and reinforcing vocabulary knowledge. This suggests that even if
explicit instruction of vocabulary is essential, it should always be consid-
ered a starting point (Schmitt & McCarthy, 1997). Once this initial stage
is undertaken, students must encounter the word in diverse contexts if
they are to build the various kinds of links and knowledge components
required for developing the full meaning of a word.
As for strategies, the study showed that successful strategies were those
that were evaluative and context-based rather than local and word-based
and that successful inferences were made by students who monitored,
considered, and judged the usefulness of the information present in the
wider context. These Ž ndings seem to suggest a need for training
learners and helping them use such strategies when attempting to derive
word meanings from context. Teachers should help students develop a
critical awareness of the problems of local and word-based strategies by
making them aware of the fact that these strategies alone may not be very
reliable sources of information for inferring word meaning from con-
text. Students should be encouraged to adopt a more context-based
approach by going beyond the word and paying attention to the phrase,
clause, sentence, and even the paragraph in which the word is located.
Teachers should also encourage students to always make sure that their
inferences are correct by checking and verifying them against the
existing clues in the wider context.
Among the different activities teachers might use to promote these
skills in language classrooms is the use of segmented texts. Teachers
could present students with short, segmented texts. The students could
infer the meanings of certain target words in each segment as they see
each new segment of the text. As they read each segment, they could

664 TESOL QUARTERLY

evaluate the information presented in the upcoming segment and their
relationship with their inferences and then verbalize their evidence for
their responses. As Porte (1988) suggested, students could initially be
presented with a sentence or segment of a text, including a target word.
When Ž nished with the Ž rst segment, they could receive a series of
written prompts or questions that they should answer in a step-by-step
manner. Questions could include those about the kind of information
and clues in the Ž rst segment, whether they used them, and also the kind
of information or clues they expect to Ž nd in the next segment. They
would also be asked to write down their initial guesses based on the
evidence they have collected so far. They would then be presented with
the next segment of the text, followed by a series of other questions
asking them whether they want to change or to stay with their initial
responses in light of the new information, and if so why. This cycle would
then continue for each target word in each segment of the text.
The assumption here is that by virtue of these explicit efforts, learners
may become more conscious of the role of contextual clues and
strategies and, hence, more active in gaining control over their search
for relevant information and knowledge sources in the wider section of
the text. However, the value and effectiveness of these activities should
be explored in empirical research. In addition, instruction focusing on
using strategies without taking into account the range of other mediating
variables may not be very effective. In the past, researchers have
conducted many studies of the effectiveness of training learners to use
comprehension and inferencing strategies. Although some have found a
positive effect for strategy training on reading comprehension and
inferencing ability (e.g., Carrell, Pharis, & Liberto, 1989; Fraser, 1999,
Kern, 1989), others have failed to produce such strong effects (e.g.,
Barnett, 1988). Thus, learners should be trained to use effective strate-
gies but should realize that inferencing is complex and that its success
involves not only the use of appropriate strategies but also the combina-
tion and coordination of those strategies with many other skills and
knowledge sources both inside and outside the text. Lexical inferencing
also depends heavily on students’ language and comprehension skills,
the types of tasks and texts, and the nature of the word as well as a host
of other individual and learner-related variables and differences.

CONCLUSION
Investigating the cognitive structures and processes of ESL students is
a revealing enterprise, offering important insights to ESOL teachers. At
the same time, results should be interpreted in view of the tentative
nature of the data examined in the research. First, the very act of asking

L2 VOCABULARY LEARNING FROM CONTEXT 665

students to report the use of strategies may have pushed them to use or
to report to have used strategies they may not have actually used.
Students were also asked to report their thinking in their L2, English.
This decision was made because the participants were of various L1
backgrounds and it was hard to collect, translate, and compare data in
different languages. However, the decision to require the use of one
language or another adds a particular set of challenges for the learners.
In using the L2, learners may not be able to fully articulate and report
their thought processes. Second, in attempting to gain depth into the
inferencing processes, I chose a relatively short passage. This decision
was made due to the qualitative nature of the research methodology and
the amount of time involved in collecting, transcribing, and analyzing
think-aloud data. Future research might include more extended, and if
possible, diverse types of reading passages. The use of different types of
reading passages, in particular, would allow the researcher not only to
explore the various strategies learners use but also to compare the use
and effectiveness of those strategies across different types of text.
With respect to the results about success, the relationship shown in
this study between some of the strategies and the students’ success
should not be taken as cause and effect relationships. As discussed
earlier, inferencing is a process consisting of multiple components and
involving a complex interaction and coordination of a number of skills,
strategies, and knowledge sources. Studies have also shown a relationship
between students’ use of inferencing strategies and their general learn-
ing styles (Parry, 1993, 1997; see also Ehrman & Oxford, 1990), which
suggests that studies are also needed to examine how success in lexical
inferencing interacts with and is mediated by other learner-related
variables, such as learners’ general cognitive and learning style prefer-
ences. The research reported here should prove useful in conceptualiz-
ing and conducting such future research on L2 inferencing.

THE AUTHOR
Hossein Nassaji is assistant professor of applied linguistics in the Department of
Linguistics at the University of Victoria. His research interests include L2 reading
comprehension and vocabulary acquisition, focus on form instruction and negoti-
ated feedback, L2 classroom discourse, and the application of sociocultural ap-
proaches to second language acquisition.

REFERENCES
Andre, M., & Anderson, T. (1978). The development and evolution of a self-
questioning study technique. Reading Research Quarterly, 14, 605–623.
Auerbach, E. R., & Paxton, D. (1997). “It’s not the English thing”: Bringing reading
research into the ESL classroom. TESOL Quarterly, 31, 237–261.

666 TESOL QUARTERLY

Barnett, M. A. (1988). Reading through context: How real and perceived strategy use
affects L2 comprehension. Modern Language Journal, 72, 50–62.
Bensoussan, M., & Laufer, B. (1984). Lexical guessing in context in EFL reading
comprehension. Journal of Research in Reading, 7, 15–31.
Bernhardt, E. B. (1991). Reading development in a second language. Norwood, NJ: Ablex.
Bernhardt, E. B, & Kamil, M. L. (1995). Interpreting relationships between L1 and
L2 reading: Consolidating the linguistic threshold and the linguistic interdepen-
dence hypotheses. Applied Linguistics, 16, 15–34.
Block, E. L. (1992). See how they read: Comprehension monitoring of L1 and L2
readers. TESOL Quarterly, 26, 319–343.
Brown, C. (1993). Factors affecting the acquisition of vocabulary: Frequency and
saliency of words. In T. Huckin, M. Haynes, & J. Coady (Eds.), Second language
reading and vocabulary learning (pp. 263–286). Norwood, NJ: Ablex.
Carrell, P. L. (1988). Some causes of text-boundedness and schema interference in
ESL reading. In P. L. Carrell, J. Devine, & D. E. Eskey (Eds.), Interactive approaches
to second language reading (pp. 101–113). New York: Cambridge University Press.
Carrell, P. L., Pharis, B. G., & Liberto, J. C. (1989). Metacognitive strategy training for
ESL reading. TESOL Quarterly, 23, 647–678.
Carter, R. A., & McCarthy, M. J. (1988). Vocabulary and language teaching. London:
Longman.
Chern, C.-L. (1993). Chinese students’ word-solving strategies in reading in English.
In T. Huckin, M. Haynes, & J. Coady (Eds.), Second language reading and vocabulary
learning (pp. 67–85). Norwood, NJ: Ablex.
Citizenship and Immigration Canada. (1996). Canadian Language Benchmarks: English
as a second language for adults/English as a second language for literacy learners
[Working document]. Ottawa, Ontario: Ministry of Supply and Services Canada.
Clarke, M. (1980). The short circuit hypothesis of ESL reading—or when language
competence interferes with reading performance. Modern Language Journal, 64,
203–209.
Coady, J., Magoto, J., Hubbard, P., Graney, J., & Mokhtari, K. (1993). High frequency
vocabulary and reading proŽ ciency in ESL readers. In T. Huckin, M. Haynes, &
J. Coady (Eds.), Second language reading and vocabulary learning (pp. 217–226).
Norwood, NJ: Ablex.
Cooper, T. C. (1999). Processing of idioms by L2 learners of English. TESOL
Quarterly, 33, 233–262.
de Bot, K., Paribakht, T. S., & Wesche, M. B. (1997). Toward a lexical processing
model for the study of second language vocabulary acquisition: Evidence from
ESL reading. Studies in Second Language Acquisition, 19, 309–329.
Devine, J. (1993). The role of metacognition in second language reading and
writing. In J. Carson & I. Leki (Eds.), Reading in the composition classroom: Second
language perspectives (pp. 105–121). Boston: Heinle & Heinle.
Dubin, F., & Olshtain, E. (1993). Predicting word meanings from contextual clues:
Evidence from L1 readers. In T. Huckin, M. Haynes, & J. Coady (Eds.), Second
language reading and vocabulary learning (pp. 181–202). Norwood, NJ: Ablex.
Ehrman, M. E., & Oxford, R. L. (1990). Adult language learning styles and strategies
in an intensive training setting. Modern Language Journal, 74, 311–327.
Ellis, N. C. (1997). Vocabulary acquisition: Word structure, collocation, word-class,
and meaning. In N. Schmitt & M. McCarthy (Eds.), Vocabulary: Description,
acquisition and pedagogy (pp. 122–139). Cambridge: Cambridge University Press.
Ellis, N., & Beaton, A. (1993). Factors affecting the learning of foreign language
vocabulary: Imagery keyword mediators and phonological short-term memory.
Quarterly Journal of Experimental Psychology, 46A, 3, 533–558.

L2 VOCABULARY LEARNING FROM CONTEXT 667

Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data (rev. ed.).
Cambridge, MA: MIT Press.
Frase, J., & Schwartz, B. (1975). Effect of question production and answering on
prose recall. Journal of Educational Psychology, 67, 628–635.
Fraser, C. A. (1999). Lexical processing strategy use and vocabulary learning through
reading. Studies in Second Language Acquisition, 21, 225–241.
Gass, S. (1999). Discussion: Incidental vocabulary learning. Studies in Second Language
Acquisition, 21, 319–333.
Haastrup, K. (1991). Lexical inferencing procedures or talking about words: Receptive
procedures in foreign language learning with special reference to English. Tubingen,
Germany: Gunter Narr.
Harley, B., & Hart, D. (2000). Vocabulary learning in the content-oriented second-
language classroom: Student perceptions and proŽ ciency. Language Awareness, 9,
78–96.
Haynes, M. (1993). Patterns and perils of guessing in second language reading. In
T. Huckin, M. Haynes, & J. Coady (Eds.), Second language reading and vocabulary
learning (pp. 46–64). Norwood, NJ: Ablex.
Hirsh, D., & Nation, P. (1992). What vocabulary size is needed to read unsimpliŽ ed
texts for pleasure? Reading in a Foreign Language, 8, 689–696.
Huckin, T., & Bloch, J. (1993). Strategies for inferring word meaning in context. In
T. Huckin, M. Haynes, & J. Coady (Eds.), Second language reading and vocabulary
learning (pp. 153–178). Norwood, NJ: Ablex.
Huckin, T., & Coady, J. (1999). Incidental vocabulary acquisition in a second
language. Studies in Second Language Acquisition, 21, 181–193.
Hudson, T. (1982). The effects of induced schemata on the “short circuit” in L2
reading: Non-decoding factors in L2 reading performance. Language Learning, 23,
1–31.
Hulstijn, J. (1992). Retention of inferred and given word meanings: Experiments in
incidental vocabulary learning. In P. Arnaud & H. Bejoint (Eds.), Vocabulary and
applied linguistics (pp. 113–125). London: Macmillan.
Joe, A. (1995). Text-based tasks and incidental vocabulary learning. Wellington, New
Zealand: English Language Institute.
Kelly, P. (1990). Guessing: No substitute for systematic learning of lexis. System, 18,
199–207.
Kern, R. G. (1989). Second language reading strategy instruction: Its effect on
comprehension and word inference ability. Modern Language Journal, 73, 135–148.
Laufer, B. (1988). What percentage of text-lexis is essential for comprehension? In
C. Lauren & M. Nordmann (Eds.), Special language: From humans thinking to
thinking machines (pp. 316–323). Clevedon, England: Multilingual Matters.
Laufer, B. (1989). A factor of difŽ culty in vocabulary learning, deceptive transpar-
ency. AILA Review, 10–20.
Laufer, B. (1997). What’s in a word that makes it hard or easy: Some interalexical
factors that affect the learning of words. In N. Schmitt & M. McCarthy (Eds.),
Vocabulary: Description, acquisition and pedagogy (pp. 140–180). Cambridge: Cam-
bridge University Press.
Laufer, B., & Sim, D. D. (1985). Taking the easy way out: Non-use and misuse of
contextual clues in EFL reading comprehension. English Teaching Forum, 23, 7–10,
20.
Lewis, M. (1993). The lexical approach. Hove, England: Language Teaching Publica-
tions.
Liu, N., & Nation, I. S. P. (1985). Factors affecting guessing vocabulary in context.
RELC Journal, 16, 33–42.

668 TESOL QUARTERLY

Morrison, L. (1996). Talking about words: A study of French as a second language
learners’ lexical inferencing procedures. Canadian Modern Language Review, 53,
41–75.
Nassaji, H. (2002). Schema theory and knowledge-based processes in second
language reading comprehension: A need for alternative perspectives. Language
Learning, 52, 439–481.
Nassaji, H. (2003). Higher-level and lower-level text processing skills in advanced ESL
reading comprehension. Modern Language Journal, 87, 261–276.
Nation, I. S. P. (1990). Teaching and learning vocabulary. New York: Newbury House.
Nation, I. S. P. (2001). Learning vocabulary in another language. Cambridge: Cambridge
University Press.
Olson, G. M., Duffy, S. A., & Mack, R. L. (1984). Thinking-out-loud as a method for
studying real-time comprehension processes. In D. E. Kieras & M. A. Just (Eds.),
New methods in reading comprehension research (pp. 253–286). Hillsdale, NJ: Erlbaum.
Paribakht, S., & Wesche, M. (1999). Reading and “incidental” L2 vocabulary
acquisition: An introspective study of lexical inferencing. Studies in Second Lan-
guage Acquisition, 21, 195–224.
Parry, K. (1991). Building a vocabulary through academic reading. TESOL Quarterly,
25, 629–653.
Parry, K. (1993). Two many words: Learning the vocabulary of an academic subject.
In T. Huckin, M. Haynes, & J. Coady (Eds.), Second language reading and vocabulary
learning (pp. 109–129). Norwood, NJ: Ablex.
Parry, K. (1997). Vocabulary and comprehension: Two portraits. In J. Coady &
T. Huckin (Eds.), Second language vocabulary acquisition: A rationale for pedagogy (pp.
55–68). New York: Cambridge University Press.
Peirce, B. N., & Stewart, G. (1997). The development of the Canadian language
benchmarks assessment. TESL Canada Journal, 14, 17–31.
Porte, G. (1988). Poor language learners and their strategies for dealing with new
vocabulary. ELT Journal, 42, 167–172.
Pressley, M., & Af erbach, P. (1995). Verbal protocols of reading: The nature of the
constructively responsive reading. Hillsdale, NJ: Erlbaum.
Pressley, M., Borkowski, J. G., & Schneider, W. (1987). Cognitive strategies: Good
strategy users coordinate metacognition and knowledge. Annals of Child Develop-
ment, 4, 89–129.
Prince, P. (1996). Second language vocabulary learning: The role of context versus
translations as a function of proŽ ciency. Modern Language Journal, 80, 478–493.
Qian, D. (1998). Depth of vocabulary knowledge: Assessing its role in adults’ reading
comprehension in English as a second language. Unpublished PhD thesis, OISE/
University of Toronto.
Ryan, A. (1997). Learning the orthographical form of L2 vocabulary—a receptive
and a productive process. In N. Schmitt & M. McCarthy (Eds.), Vocabulary:
Description, acquisition and pedagogy (pp. 181–198). Cambridge: Cambridge Univer-
sity Press.
Sanaoui, R. (1995). Adult learners’ approaches to learning vocabulary in second
languages. Modern Language Journal, 79,15–28.
Schatz, E. K., & Baldwin, R. S. (1986). Context clues are unreliable predictors of word
meanings. Reading Research Quarterly, 21, 439–453.
Scherfer, P. (1993). Indirect L2 vocabulary learning. Linguistics, 31, 1141–1153.
Schmitt, N. (1997). Vocabulary learning strategies. In N. Schmitt & M. McCarthy
(Eds.), Vocabulary: Description, acquisition and pedagogy (pp. 199–227). Cambridge:
Cambridge University Press.
Schmitt, N., & McCarthy, M. (1997). Introduction. In N. Schmitt & M. McCarthy

L2 VOCABULARY LEARNING FROM CONTEXT 669

 (Eds.), Vocabulary: Description, acquisition and pedagogy (pp. 1–5). Cambridge:
Cambridge University Press.
Shu, H., Anderson, R. C., & Zhang, H. (1995). Incidental learning of word meanings
while reading: A Chinese and American cross-cultural study. Reading Research
Quarterly, 30, 76–95.
Singer, H., & Donlan, D. (1982). Active comprehension: Problem solving schema
with question generation for comprehension of complex short stories. Reading
Research Quarterly, 27, 166–186.
Vann, R. J., & Abraham, R. G. (1990). Strategies of unsuccessful language learners.
TESOL Quarterly, 24, 177–198.
Wong, B. (1985). Self-questioning instructional research: A review. Review of Educa-
tional Research, 55, 227–268.

APPENDIX
Health in the Rich World and in the Poor
An American journalist, Dorothy Thompson, criticises the rich world’s health programmes in
the poor world. She describes her trip to Africa where she got food poisoning and her friend
malaria:

The town is very dirty. All the people are hot, have dust between their toes and the smell
of sewage in their noses. We both fell ill, and at ten o’clock in the morning I got frightened
and took my friend to the only private hospital in town, where you have to pay. After being
treated by a doctor, we caught the next aeroplane home.
Now, I believe that the money of the World Health Organisation (WHO) should be spent
on bringing health to all people of the world and not on expensive doctors and hospitals for
the few who can pay. But when we ourselves become ill, our beliefs waver. After we came
back to the States we thought a lot about our reaction to this sudden meeting with health
care in a poor country. When assessing modern medicine, we often forget that without more
money for food and clean water to drink, it is impossible to Ž ght the diseases that are caused
by infections.
Doctors seem to overlook this fact. They ought to spend much time thinking about why
they themselves do not contract some of the serious and infectious diseases that so many of
their patients die from. They do not realize that an illness must Ž nd a body that is weak
either because of stress or hunger. People are killed by the conditions they live under, the
lack of food and money and the squalor. Doctors should analyze why people become ill
rather than take such a keen interest in the curative effect of medicine.
In the rich world many diseases are caused by afuence. The causes of heart diseases, for
instance, are far from being mysterious and unfathomable—they are as well known as the
causes of tuberculosis. Other diseases are due to hazards in the natural conditions in which
we live. Imagine the typical American worker on his death-bed: every cell permeated with such
things as chemicals and radio-active materials. Such symptoms are true signs of an unhealthy
world.

From Haastrup, 1991, p. 234. Used with permission.

670 TESOL QUARTERLY