Abstract This study explores lexical inferencing success among low- intermediate and low-advanced Turkish EFL learners who are attending an English preparatory school and investigates the relationship between inferential success and L2 reading proficiency among these learners. Additionally, it looks into the knowledge sources that learners make use of in inferring the meanings of unknown words from context. Analysis reveals that although low-advanced learners are relatively more successful than the low-intermediate learners, inferential success among Turkish EFL learners attending an English preparatory school is overall low. Furthermore, the participants apparently make use of mainly discourse knowledge and grammatical knowledge although they do not resort to knowledge sources frequently and in a widespread manner. This article discusses the findings in relation to previous research, and suggests pedagogical implications of the findings. Key words: Lexical inferencing, inferential success, guessing meaning from context, reading comprehension. 1
75
Embed
bildiri.anadolu.edu.trbildiri.anadolu.edu.tr/papers/bildirimakale/8357_b733b… · Web viewAbstract. This study explores lexical inferencing success among low-intermediate and low-advanced
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Abstract
This study explores lexical inferencing success among low-intermediate and low-
advanced Turkish EFL learners who are attending an English preparatory school and
investigates the relationship between inferential success and L2 reading proficiency
among these learners. Additionally, it looks into the knowledge sources that learners
make use of in inferring the meanings of unknown words from context. Analysis reveals
that although low-advanced learners are relatively more successful than the low-
intermediate learners, inferential success among Turkish EFL learners attending an
English preparatory school is overall low. Furthermore, the participants apparently
make use of mainly discourse knowledge and grammatical knowledge although they do
not resort to knowledge sources frequently and in a widespread manner. This article
discusses the findings in relation to previous research, and suggests pedagogical
implications of the findings.
Key words: Lexical inferencing, inferential success, guessing meaning from context,
reading comprehension.
1. Introduction
Reading is a multivariate skill that combines and integrates a variety of cognitive,
linguistic and non-linguistic skills rather than a single-factor process (Nassaji, 2003).
These skills involve the very basic low-level processing abilities employed in decoding
print and encoding visual configurations as well as high-level skills of syntax, semantics
and discourse, and higher-level knowledge of text representation and integration of
ideas with the reader’s global knowledge (Nassaji, 2003). Grabe (2009) also states that
it is crucial to recognize the role of lower-level and higher-level processes in
understanding the fluent reading process. Lower-level processes refer to word
1
recognition -which is related to phonological processing, semantic and syntactic
processing, lexical access, morphological processing, automaticity and context-
syntactic parsing, meaning proposition encoding and working memory. Higher level
processes, on the other hand, include a text model of reader comprehension, a situation
model of reader interpretation, a set of reading skills and resources under the command
of the executive control mechanisms in working memory such as strategies, goals,
inferences, background knowledge and comprehension monitoring. Similarly, Koda
(2005) indicates that there are a number of essential components of reading, namely
word recognition, vocabulary knowledge, intra-word awareness and word knowledge,
information integration in sentence processing, discourse processing and text structure.
Considering the multifaceted nature of reading skill, it is plausible to assume that
shortcomings regarding either lower or higher level processes and other essential
components of reading may negatively affect reading proficiency and thus even general
L2 proficiency of EFL learners. Encountering many unknown words in a text, for
example, may adversely affect the reading comprehension of L2 readers (Nassaji,
2003). Unknown words may also prevent L2 readers from perceiving the reading texts
as a whole, and cause them to read word by word (Nassaji, 2003). Alderson (1984) also
notes that a weak command of vocabulary knowledge is likely to threaten the reading
process and/or make it seem an insuperable impediment (as cited in Kaivanpanah and
Mogaddam, 2012). It is reported that in order to overcome such an impediment, EFL
readers resort to different kinds of strategies to make up for the incomprehensible input
in general and lack of vocabulary knowledge in particular (Read, 2000). Oxford (1990)
maintains that good language learners are able to make educated guesses when
confronted with unknown words and/or phrases. Less skillful language learners, on the
other hand, often panic, ignore, or appeal to a dictionary and try to look up every
2
unfamiliar word, which jeopardizes progress toward proficiency. Therefore, in addition
to unknown words, being unable to make profitable use of strategies in reading and
dealing with new words, as good language learners do, poses another obstacle in the
wake of effective and fluent reading. Accordingly, general reading strategies and
specifically the strategies to deal with the unknown words in a reading text become
more of an issue for EFL learners to make the most of reading.
Some procedures of reading strategies such as pre-reading exercises, skimming to
get the main idea and scanning for details, providing background information, helping
students infer word meanings, encouraging guessing, and focusing on global
comprehension in post-reading exercises matter to be skilled readers (Barnett, 1988).
Additionally, Carell (1987) underscores the necessity of pre-reading activities,
vocabulary instruction, instruction regarding how to comprehend conceptually complete
authentic texts and materials that are appropriate to the cultural schemata of the readers.
When it comes to the strategies to deal with the unknown words in a reading text,
Nassaji (2003) asserts that readers use a variety of strategies when they encounter new
words such as ignoring unknown words, looking up a dictionary, writing the unknown
word down to consult a teacher afterwards, and attempting to infer their meaning from
context. Research has shown that lexical inferencing, which simply refers to inferring
the meaning of unknown words thanks to all available linguistic and even non-linguistic
cues, is the most widely used strategy by L2 learners (Nassaji, 2003, Paribakht, 2004,
Qian, 2004). For this reason, a better understanding of lexical inferencing process and
the factors it involves that may potentially contribute to its accomplishment, a more
successful and fluent reading process and even incidental vocabulary learning is of
capital importance for ELT practitioners. Especially considering that EFL learners are
unlikely to be exposed to the target language outside the foreign language classroom
3
and that classrooms and printed materials are the primary sources for them to learn the
language, enabling them to make the most of these two sources is of considerable value.
Starting from this point of view, the present paper first modestly attempts to provide
some insights into lexical inferencing and related research, and then report a relevant
experimental study conducted with Turkish speaking learners of English, as a
representative of EFL context.
2. Review of Literature
2.1. Gaining Insights into Lexical Inferencing
2.1. 1.What is lexical inferencing and how does it relate to reading
comprehension and lexical development?
The term lexical inferencing was first made prominent by Haastrup (1991), who
defines it as follows:
“The process of lexical inferencing involves making informed guesses as to the
meaning of a word in light of all available linguistic cues in combination with the
learners’ general knowledge of the world, her awareness of context and her relevant
linguistic knowledge.” (Haastrup, 1991, p.13)
As for how it relates to reading comprehension and lexical development, how young
children acquire their mother tongue sets an example. It is known that young children
preliminarily learn novel words, word forms and their meanings once they are exposed
to them through oral input provided by others (parents and other speakers of the
language). Later on, they master literacy skills and hereby appeal to reading as an
important context to continue acquiring new vocabulary (Nagy et al., 1985;
Nation&Coady, 1988; Sternberg, 1987, as cited in Wesche & Paribakht, 2009). L2
learners, especially EFL learners, on the other hand, are dependent more on written
texts to encounter and learn novel words as oral input relatively more limited in their
context. Wesche & Paribakht (2009) purport that although the reader’s primary concern
4
is comprehension while reading, readers are likely to pay attention to a specific word
form, and accordingly may attempt to determine its meaning in the context it is
embedded. This may also initiate or push forward the challenging and progressive
process of vocabulary learning, and yield retention. For this very reason, lexical
inferencing is said to operate at the core of the relationship between reading
comprehension and vocabulary development (Wesche & Paribakht, 2009). Besides,
lexical inferencing seemingly contributes to the bidirectional relationship between
reading and vocabulary as reading provides opportunities to learn novel words, and
knowing many words gives way to a better understanding of reading texts.
At the beginning of 1980s, research demonstrated that boost in children’s L1
vocabulary once they start school might be related to incidental learning through
extensive reading. This being the case, L2 researchers also directed their attention to
reading as a means of vocabulary development. As a consequence, research on L2
reading comprehension, vocabulary learning through reading and lexical inferencing
accelerated (Wesche & Paribakht, 2009). Carton (1971) was the first to publish in
lexical inferencing (as cited in Wesche & Paribakht, 2009). He investigated the types of
cues to guess the meanings of unknown words, and categorized these cues into three as
intra-lingual (L2-based), inter-lingual (L1-based) and extra-lingual (world-knowledge-
based) cues. Following this, early research on lexical inferencing continued with
Bialystok (1983), Sternberg and Powell (1983), Bensoussan and Laufer (1984) and Liu
and Nation (1985), and the research gained prominence with Haastrup (1991), as
mentioned earlier in this paper (as cited in Wesche & Paribakht, 2009). Recent related
research will also be touched upon onwards.
5
2.1.2. What factors influence lexical inferencing and its outcomes?
Wesche and Paribakht, (2009) indicate that there are factors that influence whether
L2 readers will attempt to infer the meanings of unknown words citing a number of
research studies identifying major ones of these factors such as text factors (Sternberg,
1987), word features (Parry, 1993), readers’ perception of the word’s importance
(Paribakht & Wesche, 1999; Brown, 1993; Fraser, 1999; Kim, 2003), if the word seems
to be easy to guess (Bensoussan & Laufer, 1994), and the word’s class (Paribakht &
Wesche, 1999). They further indicate that there are some other factors that influence
lexical inferencing success such as clarity of contextual clues, cue location (adjacent
versus distant), L2 proficiency, conceptual knowledge (e.g. background knowledge,
unfamiliar cultural patterns or unfamiliar topics), learner factors like their attention to
details and their ability to use the context effectively and their motivation to identify
contextual meaning. As well as general L2 proficiency, L2 reading proficiency and L2
vocabulary knowledge have predictive value for successful inferencing, and thus have
been widely investigated.
To begin with the relationship between L2 reading proficiency and lexical
inferencing, Paribakht (2004) investigated the effect of EFL learners' L2 reading
proficiency on their L2 lexical inferencing while reading an English expository text. A
group of intermediate and advanced Arabic-speaking Libyan students participated in the
study. An authentic English expository text, consisting of approximately 1000 words -
26 of which were determined as the target words- on environmental refugees was
selected and used to elicit the participants' inferencing behavior through think-aloud
protocols. The think-aloud protocols were transcribed and analyzed to find out the
knowledge sources and contextual clues that the participants used during the lexical
inferencing test. As a result, the famous taxanomy of knowledge sources used in L2
6
lexical inferencing by Paribakht and Bengeleil, which has been widely used by many a a
researcher interested in lexical inferencing, was created. According to this taxanomy,
the sources fall into two main categories as linguistic sources and nonlinguistic sources.
Linguistic sources fall into two subcategories as intralingual sources and interlingual
sources. Intralingual sources refer to target word level sources such as word
morphology, homonymy and word association; sentence level sources such as sentence
meaning, syntagmatic relations, paradigmatic relations, grammar and punctuation, and
discourse level sources such as discourse meaning and formal schemata. Intralingual
sources, on the other hand, refer to lexical knowledge –like borrowed words- and word
collocation. Lastly, non-linguistic sources cover knowledge of topic and knowledge of
medical terms. It was found that the intermediate group used multiple sources more
often and showed more variation in the combinations of knowledge sources and
contextual cues they used in inferencing than the advanced group although it was
reported that the both groups made use of single as well as multiple knowledge sources
when guessing the meaning of a given target word. It was also found that the advanced
learners had a higher percentage of correct and partially correct and a lower percentage
of wrong inferences than the intermediate-level readers. Another finding of the study
was that there were gains in the participants’ knowledge of the words immediately after
inferencing and over time; however, these gains were not significant either within or
across groups. As for the combined sample, on the other hand, a significant rate of
learning and retention was found. Similarly, Shen and Wu (2009) examined if technical
university EFL learners’ reading proficiency affects their lexical inferencing
performance and their responses to the contextual instruction. The data were collected
in three stages. In the first stage (week 1), students took a standardized reading
proficiency test to determine their foreign language reading proficiency. In the second
7
stage (week 2), all participants took the first lexical inferencing test (pretest) and
responded retrospectively to the vocabulary strategy questionnaire. Upon completion of
the 15-week treatment program, a post-test was administered to all the participants. As
with the pre-test, four short stories in the same test format and same level of difficulty
were used in the post-test. After the test, the same strategies questionnaire was
administered to examine if each student’s perception of strategy use changed after
instruction. Results revealed a significant correlation between learners’ reading
proficiency and their lexical inferencing ability,. This indicates that the higher their
reading proficiency, the better their performance in lexical inferencing ability.
Additionally, a significant correlation was found between learners’ reading proficiency
and their strategy use for lexical inferencing, which means that the higher the learners’
reading proficiency is, the more frequent the learners use strategies for lexical
inferencing. To investigate if the learners’ foreign language reading proficiency has an
effect on their responses to the treatment of lexical inferencing ability and strategy use,
the score difference between the pre-test and post-test was calculated and analyzed. As a
result, positive correlation was found between the learners’ reading proficiency and
their pre-post-test lexical inferencing score difference. This finding indicates that the
learners with better reading proficiency demonstrated better treatment effect on their
lexical inferencing ability. However, when it comes to the treatment effect on strategy
use, a slightly negative correlation was found between the learners’ reading proficiency
and the post-and pre-test difference on their strategy use. This means that the learners
with higher reading proficiency did not demonstrate more treatment effect on their
strategy use for inferring word meaning. In a similar vein, Parel (2004) attempted to
demonstrate that low receptive vocabulary, which might negatively affect L2 reading
comprehension, could be compensated for thanks to the ability to select and implement
8
word-appropriate lexical inferencing strategies. For this purpose, pairs of students,
matched for reading proficiency level, but juxtaposed in relation to their vocabulary
knowledge, were administered a lexical inferencing test, in which they were required to
guess the meanings of underlined words embedded in sentences that they were shown
one by one, and predict the upcoming words. As a result, it was reported that
individuals at the same levels of reading proficiency but different levels of vocabulary
knowledge were reported to differ with regard to the ability to select and implement
word appropriate lexical inferencing strategies. The participants assumed to be better at
vocabulary were reported to be more successful in lexical inferencing and made a
higher number of plausible predictions. These findings indicate that sensitivity to word
structure, when combined with information from the context, can lower the threshold
for receptive vocabulary knowledge in relation to text demands.
Besides L2 reading proficiency, both depth and breadth of vocabulary knowledge
matter with regard to successful lexical inferencing. Nassaji (2006), for example,
scrutinized the relationship between ESL learners’ depth of vocabulary knowledge, their
lexical inferencing strategy use and their success in deriving word meaning from
context. 21 adult intermediate-level ESL learners coming from a variety of L1
backgrounds were divided into two groups as lexically skilled (LS) and lexically less
skilled (LLS) depending on their scores from a depth of vocabulary knowledge test, and
an 374-word text was administered as the lexical inferencing text, in which the
participants were asked to guess the meanings of 10 target words in think-aloud
protocols. LS group was found to have performed significantly better than LLS group
with regard to the lexical inferencing test, indicating that depth of vocabulary
knowledge plays a crucial role in lexical inferencing success among ESL learners. It
was also found that LS and LLS groups significantly differed in their overall rate of
9
success in inferring the meanings of unknown words. Results indicated that LS readers
employed local word-based strategies (identifying) less frequently than the LLS readers;
however, they benefitted more from this strategy. LS readers benefited more from the
monitoring strategy than the LLS readers did, too. When it comes to the evaluating
strategy, however, no significant difference was observed between the two groups. The
above findings indicate that depth of vocabulary knowledge plays an important role in
the types of lexical inferencing strategies that ESL learners employ as well as how
effective they use these strategies. Further findings revealed that learners’ depth of
vocabulary knowledge is a much stronger predictor of inferential success than lexical
inferencing strategy use. In other words, depth of vocabulary knowledge significantly
contributed to inferential success both on its own and over and above lexical inferencing
strategy use. Lexical inferencing strategy use, on the other hand, significantly predicted
lexical inferencing success when entered as the first variable to the model; however, its
contribution failed to reach significance when entered to the model after depth of
vocabulary knowledge.
Ehsanzadeh (2012) also aimed at exploring the roles of depth and breadth of lexical
repertoire in L2 lexical inferencing success and incidental vocabulary acquisition
through reading. Statistically significant intercorrelations were found between depth and
breadth of lexical repertoire and incidental vocabulary acquisition. However, the depth
of word knowledge was found to have a higher correlation with meaning retention than
does the breadth of word knowledge. Similarly, the intercorrelations among L2 lexical
inferencing success, depth and breadth of word knowledge were both positive and
significant. Yet, depth of vocabulary knowledge correlated more with the lexical
inferencing success than the breadth of vocabulary knowledge. Additionally, depth of
word knowledge accounted for 58% of the variance in the lexical inferencing success.
10
The breadth of word knowledge, on the other hand, contributed only an additional 8%
of the variance in lexical inferencing success. However, the breadth of vocabulary
knowledge alone explained 40% of the variance in lexical inferencing when first entered
to the regression model. The depth of lexical repertoire explained an additional 26% of
the variance in the lexical inferencing success. The additional contributions of both
depth and breadth were non-significant, though. It is clear from the findings that both
depth and breadth of vocabulary knowledge correlate with and contribute to lexical
inferencing and vocabulary retention; however, the role of depth of vocabulary
knowledge is greater than that of the breadth. On the contrary, Hatami and Tavakoli
(2012), in their study with 50 Iranian students majoring in English literature, found that
breadth of vocabulary knowledge significantly accounts for 37% of the variance in
lexical inferencing as well as significantly contributing to lexical inferencing above and
beyond depth of vocabulary knowledge. Depth of vocabulary knowledge, on the other
hand, significantly accounts for 29% of the variance in lexical inferencing. However, it
fails to significantly contribute to lexical inferencing above and beyond breadth of
vocabulary knowledge. These findings indicate that breadth of vocabulary knowledge is
a stronger predictor of lexical inferencing success than depth of vocabulary knowledge.
They further found that both depth and breadth of vocabulary knowledge are
inextricably linked to lexical inferencing success, whereas there is no significant
relationship between the two aspects of vocabulary knowledge (depth and breadth) and
the perceived ease of inferencing.
Two other factors that seemingly affect lexical inferencing are topic familiarity and
passage-sight vocabulary. Pulido (2007) investigated their impact on lexical inferencing
and retention conducting a study with 35 English speaking adult Spanish learners.
Results yielded a significant impact of topic familiarity on lexical inferencing. This
11
means that learners are able to come up with more correct guess while they are reading
about a more familiar scenario compared to less familiar scenarios. However, the effects
of topic familiarity on perceived ease or difficulty in guessing were inconsistent.
Similarly, it had inconsistent effects on lexical retention after the inference verification
task to promote revision, integration, and retention. When it comes to the passage sight
vocabulary, it was also reported to play a significant role in lexical inferencing for new
words during the reading tasks. This indicates that as passage sight vocabulary increases
so does ability to correctly infer target words’ meanings. Passage sight vocabulary also
consistently predicted lexical retention, although the effect sizes were not as strong as
on the lexical inferencing measure. Sara, Parviz and Keivan (2013) also explored the
effects of topic familiarity and passage sight vocabulary on L2 lexical inferencing. The
target group was Iranian EFL learners this time. It was found that topic familiarity had a
statistically significant impact on L2 lexical inferencing. There were considerably more
correct target word inferences when reading about a more familiar scenario compared to
a less familiar scenario. Similarly, increase in passage sight vocabulary knowledge
increased the ability to infer the meanings of the target words correctly. When it comes
to the relationship between topic familiarity/passage sight vocabulary knowledge and
perceived ease of lexical inferencing, it was reported that the difference in the
perception of the participants in inferring the meanings of the target words of the more
and less familiar scenarios was not statistically significant, indicating that the
participants’ perception of ease of lexical inferencing did not vary according to topic
familiarity. Finally, a slightly stronger relationship was found between passage sight
vocabulary knowledge and ease of lexical inferencing in the less familiar scenarios
compared to the more familiar ones. This finding was interpreted as follows: when the
participants did not have the relevant schematic knowledge to assist them in lexical
12
inferencing process (in other words, when they were less familiar with a reading text),
they resorted to passage sight vocabulary to help them.
Kaivanpanah and Alavi (2008) examined the contribution of grammatical
knowledge to inferring the meaning of unknown words through introspective think-
aloud protocols. Consequentially, Grammar knowledge was found to influence lexical
inferencing ability of L2 learners in simple and complex texts. The results also indicated
that learners representing different ability groups with respect to grammar knowledge
perform differently on the simple and complex texts. The comparison of the means
demonstrates that participants were more successful in inferencing in syntactically
simple texts.
Shen (2010) examined the effects of perceptual learning style preferences on L2
lexical inferencing and whether learners with certain perceptual learning styles
benefited more from an explicitly instructional program. A total of 145 Taiwanese
learners of English were classified into different learning styles such as visual, auditory,
kinesthetic, tactile, group and individual to be provided with the context clues and
inference strategies within a 15-week program. Results indicated that prior to the
instruction, learners with group learning style as their major style performed better than
the others in lexical inferencing ability. Group learners were followed by individual
learners, kinesthetic, tactile, auditory, and visual learners. When it comes to the post
test, it was found that auditory learners achieved the higher gain mean than the others,
while the group learners ranked the lowest, comparing pre-test and post-test results.
Additionally, the significant treatment effect was found on the visual learners as well as
the auditory learners. In sum, the order of the effect of instruction on contextual
inference was as follows: auditory learners > visual learners > individual learners >
kinesthetic learners > tactile learners > group learners. In other words, although
13
instruction of inference strategy was effective for the learners, learners with certain
perceptual learning style preference, namely auditory and visual learners, benefited
more from the explicit instruction.
Finally, lexicalization status of L2 words in a learner’s L1, which refers to L2
words’ having lexical equivalents in the L1, is said to be indirectly related to lexical
inferencing success (Wesche & Paribakht, 2009).
2.1.3. What processes are involved in lexical inferencing and how have they
been conceptualized and theoretically explained?
Wesche and Paribakht (2009) point out that researchers have found theoretical
frameworks to make lexical inference and its outcomes fit in. Within these frameworks,
lexical inferencing, like inferencing in general, is referred to as a multidimensional
cognitive process that involves both declarative and procedural components. Of these
frameworks, the first to spring to mind is knowledge sources, as a lexically linked
framework. Knowledge sources simply refers to finding information about words
(Nation, 2001). It includes analyzing the word (affixes and stems), using context,
consulting a reference source in L1 or L2 (written formal sources like dictionaries,
glossaries or concordances, and more spontaneous sources like asking teachers, native
speakers or other learners) and using parallels in L1 and L2 (Nation, 2001:218).
Descriptive taxonomies of knowledge sources use by different groups of learners have
been created by a variety of researchers. A very common one of these taxonomies,
formed by Paribakht and Bengeleil (2004), is provided below.
I. Linguistic sources
A. Intra-lingual sources
1. Target word levela. word morphology b. homonymyc. word association
14
2. Sentence levela. sentence meaningb. syntagmatic relations c. paradigmatic relations d. grammare. punctuation
3. Discourse levela. discourse meaning b. formal schemata
B. Inter-lingual sources
1. Lexical knowledge 2. Word collocation
II. Non-linguistic sources
A. Knowledge of topic
B. Knowledge of medical terms
(Paribakht and Bengeleil, 2004:231, Figure 1: The taxonomy of knowledge sources
used in L2 lexical inferencing,)
Nassaji (2003), on the other hand, defined five main categories of knowledge
sources, namely grammatical knowledge, morphological knowledge, world knowledge,
L1 knowledge and discourse knowledge. Qian (2005) makes a mention of three main
categories of knowledge sources as location of clues (e.g. within the test word, context
within T-unit and context beyond T-unit), knowledge of the world and intra-lingual
vocabulary knowledge (e.g. phonological/orthographic forms, morphology, syntax and
meaning). Although these taxonomies differ in name or number of categories they
include, they all rely on the starting point of Carton’s three major cue categories,
namely inter-lingual, intra-lingual and extra-lingual, as previously mentioned in the
present paper.
Within lexically linked frameworks of lexical inferencing, connectionism is also
worth mentioning. Connectionism, also known as associative networks or parallel
15
distributive processing theory, can be applied to motor aspects of articulation and to the
perceptual aspects of vocabulary acquisition (Ellis, 1994, 2002, as cited in Wesche &
Paribakht, 2009). In this sense, it helps explain the regular and steady familiarization
with novel words encountered in reading.
On the other hand, lexical inferencing also fits in comprehension-oriented
frameworks on the grounds that it is essential for word comprehension just as general
inferencing ability is so for text comprehension. As is seen, lexical inferencing is related
to both language comprehension and acquisition. In addition to this, it functions as a
bridge between comprehension and acquisition. In so doing, it is supported by Gass’
(1988, 1997) input processing, formulated to characterize stages in the acquisition of
new syntactic knowledge (as cited in Wesche & Paribakht, 2009). In other words, input
processing helps explain why some unknown words go unnoticed, while some others
are superficially perceived, and still some others are understood in context through
inferencing.
2.2. Other Related Research on Lexical Inferencing
One of the prominent studies on lexical inferencing was conducted by Nassaji
(2003), who examined the use of strategies and knowledge sources in L2 lexical
inferencing and their relationship with inferential success. Data were collected through
think-aloud protocols, in which 21 adult ESL learners (10 males and 11 females)
coming from a variety of L1 backgrounds were required to verbalize their thoughts
regarding how they guessed the meanings of the target words. As a result of counting
successful, partially successful and unsuccessful inferences, it was found that
intermediate-level ESL learners are not very good at guessing the meaning of unknown
words in a reading text. Moreover, there were five knowledge sources employed,
16
namely grammatical knowledge, morphological knowledge, L1 knowledge, world
knowledge and discourse knowledge, and six strategies, namely repeating (word
repeating and section repeating), verifying, analyzing, monitoring, self-inquiry and
analogy. World knowledge was reported to be the most frequently used strategy, while
word repeating was the most frequently used strategy. It is worth mentioning that not all
the students used all the knowledge sources and strategies and there was variation
among them with regard to the types of knowledge sources and strategies. Additionally,
morphological knowledge contributed to lexical inferencing success more than the other
knowledge sources. The percentages of successful inferences for each knowledge
source did not significantly differ from each other, whereas the percentages of
successful inferences for each lexical inferencing strategy significantly differed from
each other, indicating that inferential success depended significantly on what kind of
strategy learners used, but not what kind of knowledge source was used. Tavakoli and
Hayati (2011), on the other hand, examined how two different language proficiency
groups of Iranian EFL learners (high-intermediate versus low-intermediate) use lexical
inferencing strategies to guess the meanings of a number of unknown words embedded
in four reading passages. Results showed that for each reading passage, low-
intermediate learners made more inferences; however, intermediate learners made more
accurate guesses. It was also found that the two proficiency groups differed in their use
of knowledge sources. Further analysis of the data revealed that the students at low-
intermediate level focused more on the words and their syntactic category, while the
students at high-intermediate level preferred the discourse knowledge and tried to
understand the whole passage. It was also found that gender, one of the concerns of the
study, played no role regarding the type of inferencing strategy used and the success of
the participants’ guesses. In addition to language proficiency and gender, Gao (2012)
17
investigated the differentiations and homologies of Chinese undergraduates in terms of
lexical inference ability and lexical inference strategy use with regard to major. Results
indicated that students of different language proficiency, gender and major do perform
differentially in test in terms of lexical inferencing strategy use and lexical inferencing
ability, even though some are significant and some are not.
Some related studies were carried out in Turkish context, as well. For example,
Akkan (2008) investigated Turkish university preparatory school EFL learners’
awareness and use of English affixes as a knowledge source in guessing the meanings
of unknown words in written contexts and examined their use of prefixes and suffixes
separately. Results indicated that Turkish EFL learners use a variety of knowledge
sources, one of which is word morphology, when attempting to infer the meanings of
unknown words in written context. Despite this, it was reported that they did not do it
very often. Moreover, the participants appeared to use both suffixes and prefixes nearly
equally; however, their prefix use seemed to be associated with more successful guesses
than suffix use in spite of the participants’ unfamiliarity with prefixes in their L1
(Turkish). İstifçi (2009) analyzed how a group of Turkish EFL learners (intermediate
and low-intermediate level) guessed the meanings of a number of unknown words
embedded in four authentic texts. The participants were asked to guess the meanings of
the unknown words embedded in the reading passages, and write down how they did
this and what strategies they made use of. When it comes to the data analysis, all correct
inferences were counted and their percentages were taken, and the sources used by the
intermediate and low-intermediate level learners were compared. Results revealed that
for each reading passage, low-intermediate learners made more inferences (except the
forth reading passage in which low intermediate level learners 66 inferences, whereas
intermediate level learners made 79 inferences); however, intermediate learners made
18
more accurate guesses. The discourse knowledge category was the one which was
mostly used by the students at both levels. This indicates that both low-intermediate and
intermediate level learners used their knowledge of cohesive devices to link the words
in the texts, in which the intermediate level learners were more successful. The second
most frequently used category was the world knowledge. When it comes to the
homonymy category, the low-intermediate level students used more sound relations
between words in L2, and all their guesses were successful. This finding indicates that
low L2 proficiency learners often try to associate new words by using phonetic
similarities, which is also in line with the previous research. In terms of morphology,
low-intermediate level learners were reported to try to infer the meanings from their
inflections, derivations and the morphological relationship between words. It is worth
mentioning that they were successful in using this inferencing strategy. With regard to
cognates, low-intermediate level learners made use of them relatively more than the
intermediate levels. However, it is crucially important to note that “cognates” refers to
borrowed terms in this study. Considering all of the results mentioned above, the
researcher arrives at the conclusion that although low-intermediate level learners use
more lexical inferencing strategies in proportion to intermediate level learners, the latter
come up with more successful guesses. This implies that students at low-intermediate
level need to be trained to infer the meanings of unknown words. Akpinar (2013) also
examined how Turkish EFL learners handle unknown words while reading English texts
through think-aloud protocols. It further investigated the relationship between these
learners’ perceptions and actual practices regarding the employment of knowledge
sources while trying to guess the meaning of unknown words. The findings of this
study, in which a group of ELT learners were tested on Vocabulary Strategy Survey and
Lexical Inferencing Test, revealed a significant negative weak correlation between the
19
perceptions and actual practice of the participants regarding how they handle unknown
vocabulary while reading English texts. This indicates that the higher the perceptions of
the students are, the lower their actual practices in lexical inferencing.
Cross-linguistic studies investigating lexical inferencing do exist, too. For example,
Wang (2011) examined the lexical strategies used by Filipino graduate students and
Chinese graduate students when they encounter unknown words while reading. The
findings indicated that the Chinese graduate students, compared with the Filipino
graduate students, preferred using lexical strategies more to deal with unknown words.
Furthermore, notable distinctions were found in using inferring strategies between the
Filipino graduate students and the Chinese graduate students. First, the two groups
differed in the lexical strategies they preferred. For example, the Chinese graduate
students preferred making use of strategy of guessing using discourse context, that is,
using outside the sentence in which the word occurred, guessing using local (sentence-
level) context and guessing using visual form (similarity or morphological
understanding), whereas more Filipino graduate students than the Chinese graduate
students chose to utilize using syntactic knowledge and extra textual (thematic or world)
knowledge. Second, compared with the Filipino graduate students, the Chinese graduate
students made more wrong inferences of unknown words they encountered in reading
English although they could recognize the unknown words. Third, more unsuccessful
inference of the unknown words was made by Chinese students, and the discrepancy of
the Chinese students‟ level of vocabulary knowledge is much larger than that of the
Filipino students as far as the standard deviations are concerned. Fourth, the occurrence
of unknown words seemed to have an influence on the lexical inferencing used by the
Filipino graduate students but not on the inferencing strategies by the Chinese graduate
students. Lastly, the Chinese graduate students appeared to gain some words through
20
incidental learning when using lexical inferencing to deal with these unknown words,
while the Filipino graduate students did not report learning any other word except one.
3. The Present Study
The present study aims at exploring whether Turkish EFL learners from two
different reading proficiency groups, namely low-advanced and low-intermediate,
attending an English preparatory school, are able to infer the meanings of unknown
words while reading. . In addition to this, what knowledge sources they make use of in
so doing is also investigated. It is worth mentioning that the two reading proficiency
groups are compared in order to probe the role of L2 reading proficiency in lexical
inferencing success and use of knowledge sources. For these purposes, the present paper
seeks for answers to the following research questions:
1. How successful are the low-advanced and low-intermediate Turkish EFL learners
in guessing the meanings of unknown words?
a. What is the role of L2 reading proficiency in this success?
2. What are the knowledge sources that low-advanced and low-intermediate Turkish
EFL learners employ to infer the meanings of unknown words?
3.1. Participants
A total of fifty-two Turkish EFL students, who were studying at the preparatory
school of Anadolu University, participated in the present study. According to Common
European Framework of References for Languages (CEFR), half of the participants
were at B2 level, while the other half were at B1 level. However, Anadolu University
School of Foreign Languages prefers to split B1 and B2 levels into two as B1.1 - B1.2
and B2.1 - B2.2, aiming to elaborate the learning outcomes for each level and better
equip the learners with these outcomes. In view of this fact, the participants of the
current study fall into B2.2 and B1.2 levels, which are assumed to correspond to low-
21
intermediate and low-advanced levels respectively. These language proficiency levels of
the participants were determined by EMT (End of Module Test) that they sat a week
before the data collection commenced. The following explanation would better help to
confirm the participants’ L2 proficiency level. This study was conducted within the very
first week of the second module of 2013-2014 spring term. At the beginning of the fall
term, the participants, like all the other students at the same school, took a placement
test followed by a proficiency exam (including listening comprehension, reading
comprehension, language use, writing and speaking), which was determinant on the
proficiency level each student would be in in the first module. At the end of each two-
month time, they sat the EMT, which was determinant on the proficiency level they
would be in in the second module. As a result, considering that the participants of the
present study sat the same reliable and valid test, and the data collection of the present
study was carried out right after they sat this test, it is possible to safely arrive at the
conclusion that the participants were at the aforementioned language proficiency levels.
There were 30 males and 22 females, with their ages ranging from 18 to 27. They were
all monolinguals with the native language being Turkish, and no participant included in
the current study had an extended stay in an English speaking country.
3.2. Instruments
Two main testing instruments were utilized for the accomplishment of the present
study. First, the participants’ reading comprehension was assessed using a shortened
paper-based TOEFL reading post-test (Appendix A). The test was adapted from
Longman Preparation Course for the TOEFL Test (Philips, 2003). In the original test,
there were five short reading texts (composed of two or three paragraphs) followed by
nine to eleven multiple-choice questions. The time limit for the original test was 55
minutes. However, in the present study, the first four short reading passages were used c.
22
There were a total of forty multiple-choice questions. 45 minutes were allocated for this
shortened version of the test considering the number of questions and the time limit in
the original study. Prior to the administration of this test, the participants were told that
this test was designed to measure their ability to read and understand short passages
similar in topic and style to those that students in North American universities and
colleges are likely to encounter for their studies. They were additionally provided with a
sample paragraph followed by two sample questions. Second, a reading text was used as
the lexical inferencing instrument (Appendix B). It was a 374-word text used in Nassaji
(2003). There were 10 target words embedded in the original study. In the present study,
these target words have been modified according to the findings gathered from the pilot
study (See Appendix C for the pilot study). The target words in the original text were
squalor, affluence, waver, contract, sewage, curative, assess, hazard, unfathomable and
permeate. The pilot study revealed that two students (5%) knew the meaning of sewage;
another two students (5%) knew the meaning of curative; 12 students (28%) knew the
meaning of assess; 13 students (31%) knew the meaning of hazard, and 1 student (2%)
knew the meaning of permeate as they were used in the reading text to be used.
Therefore, these words (sewage, curative, assess, hazard, permeate) have been replaced
by the following ones respectively: sewerage, sanative, appraise, peril, imbue.
Frequency of occurrence was determinant on choosing these words. The following table
provides the tokens found for each word in the original text and those determined for
the purpose of the present study according to COCA (Corpus of Contemporary
American English).
23
Table 1: Frequency occurrence of the target words that were taken from the
original study along with that of the ones they have been replaced with.
Nassaji (2003) Tokens The present study Tokens
Sewage 3,178 Sewerage 133
Curative 317 Sanative 1
Assess 11,160 Appraise 317
Hazard 2,995 Peril 1,798
Permeate 426 Imbue 185
The target words are all content words consisting of four nouns, four verbs, and two
adjectives. The reading passage was chosen because it contained a minimum number of
words that the participants would not know. In other words, a big proportion of the
reading text with the vocabulary it contains –except the target words-, grammar
structures and length of sentences was comprehensible. Additionally, the topic was not
very familiar to the participants. Moreover, words invited the use of various knowledge
sources and a range of word classes were presented. The text statistics and readability
indices are shown in Table 2 . These results mean that the text is easily readable.
Table 2 : Text Statistics and Readability Indices
Text Statistics
Number of sentences 20
Number of words 375
Number of complex words 39
Percent of complex words 10.40%
Readability Indices
Flesh Kincaid Reading Ease 66.4
Flesh Kinkaid Grade Level 8.7
24
3.3. Data Collection
Data collection took two weeks. At the beginning of the first week, the very first
week of the new module, a pilot study was conducted to ensure that the target words
were unfamiliar to the participants. A total of 42 participants of varying L2 proficiency,
predominantly advanced learnersa, were given a list of the target words in a
decontextualized way, and asked to write down whether they knew the meanings of
those words, and to write L1 equivalents for each if they knew. At the end of the first
week, the actual participants of the present study sat a forty-item paper-based TOEFL
reading text. All of the participants were required to complete this test in a total of 45
minutes. b The following week, first of all, the participants were given a list of the target
words whose meanings they would be asked to infer later, and asked to write down their
meanings if they knew so that the data coming from the participants who provided the
correct definitions or L1 equivalents would be excluded from the analysis. However, as
it was expected, none of the participants provided correct definitions or L1 equivalents
for the decontextualized target words. Next, the lexical inferencing test was
administered with no time limit. The participants were instructed to read the text and
determine the meanings of the bold-faced target words along with describing what
helped them to do so on a separate piece of paper. The participants were allowed to use
L1 (Turkish).
4. Data Analysis and Results
Depending on its findings, the present study arrived at the conclusion that although
low-advanced learners were found to be relatively more successful than the low-
intermediate learners, inferential success among Turkish EFL learners attending a
preparatory school was overall low. In addition to this, it was found that the participants
25
made use of mainly discourse knowledge and grammatical knowledge although they did
not resort to knowledge sources frequently and in a widespread manner.
4.1.1. Success of Lexical Inferencing
To be able to answer the first research question -How successful are the low-
advanced and low-intermediate Turkish EFL learners in guessing the meanings of
unknown words, and what is the role of L2 reading proficiency in this success?-, first,
an independent samples t-test was conducted to see if the two groups –low-intermediate
and low-advanced learners- significantly differed in their reading comprehension so that
their inferential success could be compared. It was found that the low-advanced learners
(M=22, SD=4.24) scored significantly better than the low-intermediate learners (M=8,
SD=3.28) with regard to the scores they obtained from the TOEFL reading subtest
(t(50)=13.302, p<.001). This finding confirmed that the inferential success of the two
groups could be compared in terms of their varying reading proficiency.
Second, the data coming from the lexical inferencing task were analyzed by the
researcher and a colleague who is an experienced English instructor. It is worth
mentioning that the data coming from eight participants were excluded as they did not
seriously attempt to infer the meanings of the target words, thus making the total
number of low-advanced and low-intermediate participants 20 and 24 respectively. The
participants’ responses to each of the unknown words were rated using a three-point
scale (2=successful, 1=partially successful, 0=unsuccessful). Successful inferencing was
defined as responses that were semantically, syntactically and contextually appropriate.
It could be both a synonym and a full definition of the target word either in L2 (English)
or L1 (Turkish). Semantically appropriate but syntactically deviant responses, or vice
versa, were rated as partially successful. When the response was neither semantically
nor syntactically appropriate, it was rated as unsuccessful. To be able to provide a
26
framework of contextually appropriate responses, the reading text, with the target words
being removed, was presented to two native speakers of English, who were working as
language teaching assistants at the home university of the study during the
implementation. They were requested to complete the sentences with contextually
appropriate words. They were also reminded that they could write more than a single
word for each gap so that alternative words could be gathered. If the participants
responded to the inferencing task with any of the words provided by the native speakers,
they were rated successful. When there were disagreements between the two raters (the
researcher and a colleague) regarding if a response was successful, partially successful
or unsuccessful, these two native speakers of English were consulted.
Results let out a total of 362 inferences and 78 cases of no-attempt to infer the
meanings of the unknown words. Of these, 184 (50.8%) inferences belonged to the low-
advanced group, while 178 (49.2%) inferences came from the low-intermediate group.
Of the 184 inferences that the low-advanced group made, 59 (32%) were successful, 43
(23%) were partially successful, 66 (36%) were unsuccessful and 16 (9%) were ignored.
Of the 178 inferences that the low-intermediate group made, only 11 (6%) were
successful, 46 (26%) were partially successful, 59 (33%) were unsuccessful and 62
(35%) were ignored. These findings indicate that although low-advanced learners were
found to be relatively more successful than the low-intermediate learners, inferential
success is overall low. Table 3 below makes a summary of these findings.