Assessing the Effects of Word Exposure Frequency on Incidental Vocabulary Acquisition from Reading and Listening

Incidental vocabulary acquisition from reading

Previous studies have shown that reading could aid learners in incidentally acquiring partial word knowledge. For example, Song and Sardegna (2014) divided 24 secondary school EFL students into two groups. One group received enhanced reading instruction, while the other group did not. Pre-tests and post-tests were administered to measure the development of target words. Results revealed that the group of students who received reading instruction improved from a mean score of 53.67 to 67.59. The group of students who did not receive reading instruction only improved from 52.83 to 53.67. Students’ responses to retrospective interviews provided evidence for factors which affected their incidental acquisition of English vocabulary. One of the factors noted is frequent encounters of unknown words in meaningful contexts. This suggests that word exposure frequency is a main factor in incidental vocabulary learning, and it is an issue worth researching (Webb & Chang, 2012).

Webb (2007) examined the effects of repetition (1, 3, 7 and 10 encounters) on word knowledge. Frequently used words appearing in authentic reading texts were replaced with nonsense words. Results showed that greater gains in knowledge were found for at least one aspect of knowledge with each repetition. Chen and Truscott (2010) explored incidental vocabulary learning through a reading-only condition. The participants were 72 freshmen from two Asian universities. In contrast with Webb’s (2007) research, their study used 10 actual English words (five that were lexicalized in the participants’ native language and five that were not). These words were presented in 13 reading passages rather than in isolated sentences. Each passage included exactly one occurrence of each of the target items. Three groups were assigned a different number of passages: Group 1 (1 passage, target words occurred 1 time, Group 2 (3 passages, target words occurred 3 times), Group 3 (7 passages, target words occurred 7 times). Results revealed that scores consistently increased with the number of exposures. However, among the seven aspects of word knowledge that were measured, knowledge of orthography, part of speech, and meaning showed non-significant development with increasing encounters. In other words, only partial word knowledge was acquired. Moreover, repetition increased the acquisition of productive knowledge more than receptive knowledge. Their results reinforced Webb’s (2007) research findings.

Teng (2014a) explored incidental learning of 30 substituted non-words within five groups of frequency levels in reading a graded reader, Love or Money. Word form recognition, meaning recall and word usage were measured. The results revealed all word knowledge types improved consistently with the number of encounters. The acquisition profile of the three aspects of word knowledge is word form recognition, followed by meaning recall, and then usage. This study confirms Pellicer-Sánchez and Schmitt’s (2010) study that meaning recognition requires more exposures than form recognition (see also Pigada & Schmitt, 2006), but contradicts Waring and Takaki’s (2003) study that meaning recognition requires fewer exposures than form recognition.

A common point in the studies mentioned above is that that students benefited from gaining word knowledge through reading, and more exposures to words helped them move towards mastery of a new word’s form-meaning link. However, simply being exposed to a word frequently is not sufficient, and only partial word knowledge was acquired. This might be due to the different tests that were used to measure word knowledge. Hence, further study is needed for incidental vocabulary acquisition from reading.

Incidental vocabulary acquisition from listening

Van Zealand and Schmitt (2013) investigated incidental acquisition of three dimensions of word knowledge through listening. Results showed that participants scored highest on the form recognition test, followed by the grammar recognition test, and then the meaning recall test. On the immediate post-test, the three dimensions of word knowledge were significantly affected by frequency (p <.001). On the delayed post-tests, knowledge of grammar and meaning were not significantly affected by frequency. This suggests that word exposure frequency in listening is not the only factor affecting the acquisition of new words.

Brown et al. (2008) explored incidental vocabulary acquisition from listening, and compared gains from reading-while-listening and reading. Knowledge gains of 28 target words within four frequency groups were measured with two meaning recognition tests: a multiple-choice recognition test and a meaning-by-translation test. One week after the experiment, learners from the listening mode scored 8.2 (29.2%) for the multiple-choice recognition test and 0.55 (1.9%) for the multiple-choice recognition test. This result is significantly lower than reading-while-listening and reading-only modes. In addition, listening also presented significant attrition in retention of the new words. An effect of frequency was found in all three modes. More repetitions were needed in listening and reading than reading-while-listening for a substantial improvement in incidental vocabulary learning.

Likewise, Vidal (2011) compared incidental vocabulary gains from reading and listening. A vocabulary knowledge scale was used to measure vocabulary development. On the immediate post-test and one-month delayed post-test, readers scored significantly better than listeners did. An effect of frequency was stronger in reading than listening. This means that listening needs more word exposure frequency than reading for substantive vocabulary development.

These studies mentioned above indicated that listening has a less pronounced effect than reading alone in incidental vocabulary gains. A possible reason is that it may be more difficult to notice a new word in listening, because speech segmentation is difficult for EFL students (Vidal, 2003).

As mentioned above, many previous studies showed that learning gains found from listening were smaller than those from reading. However, there was one study conducted by Suggate, Leonhard, and Neudecker (2013) which showed that learners learned the smallest number of words from independent reading, but the largest number from storytelling (participants listened). In the process of analyzing the variance, the lack of sensitive word knowledge measurement in listening was noticed. This is particularly reflected on measuring a form-meaning link. As Nation (2001) stated, establishing a form-meaning link for a new word is complex. This provides rationale for the primary purpose of the current study of measuring the incidental uptake of four types of word knowledge (form recognition, form recall, meaning recognition, and meaning recall).

Participants

A total of 60 participants aged 18-20 were involved in the present study. The original number of participants was 126. They studied English as a foreign language and were enrolled in different majors at the university where the study was conducted. College English was the only subject for them in learning English. A survey before the study revealed that they had never studied in a country where English was the official language. To be in line with the nature of incidental vocabulary acquisition, participants were not informed of the real purpose of this study. They were informed that they would receive some reading exercises, and they were entitled to withdraw from the study at any time if they were not comfortable with it.

The criterion for selection was that they had a similar word level in Nation and Beglar’s (2007) Vocabulary Size Test (VST). The validity of this test was shown in Beglar’s (2010) study. The average score of the selected students in the current study was 31.16. According to Nation and Beglar (2007), they had an average vocabulary size of 3,116 when the score was multiplied by 100. They were randomly assigned to two groups: reading (Group 1, n = 30) or listening (Group 2, n = 30). The validity of comparing these two groups was confirmed since their average scores on the VST did not differ significantly (Group 1: 31.75 out of 140, Group 2: 31.04 out of 140, p =.88).

Target items

The target words were 15 words that are unlikely to be known to the participants based on the author’s teaching experience. The 15 words included verbs, adjectives, and nouns because they are the most common parts of speech found in natural texts (Webb, 2008).

The selected words did not have significantly different word lengths or syllables. Moreover, as shown by running the RANGE program (Heatley, Nation, Coxhead, 2002), the selected words were not within the list of the 3,000 most frequent words. Hence, in considering learners’ word level, they were unlikely to know these words. The 15 target words were arranged into three groups of different exposure frequencies (Table 1).

To make sure that the participants would not know the target words, a pre-test was administered four weeks before the study. The pre-test was a multiple-choice test. The participants had to choose a meaning similar to the target word. An I don’t know option was provided, and students were encouraged to select this option if they did not know the word meaning.

Example:

Intimidate

A. like B. enjoy C. threat D. walk E. I don’t know

This test included 30 items (15 target words and 15 high-frequency words). The 15 high-frequency words were selected from the New General Service List (NGSL, Browne, 2013). The reason for adding 15 high-frequency words and administering the test four weeks before the study was to alter the learners’ attention on the target words. In addition, adding 15 high-frequency words could avoid the difficulties that the participants might have in taking a test containing many unfamiliar words. The scores of the 15 high-frequency words were not calculated. All the participants chose the I don’t know option for 10 target words (epidemic, torpedo, tumultuous, contamination, thorough, alleviate, deterioration, prosperity, obsolete, prevalent), 25 chose the I don’t know option and 35 chose one of the three incorrect options for the other 5 target words (contumacious, condemn, contrive, diminish, inevitable). Thus following this test, it was confirmed that the participants had no prior knowledge of the target words. In addition, as this pre-test was administered four weeks before the study, the learners would not deliberately commit those words to their memory, particularly when they are in a EFL context where English is seldom spoken or used in their daily life. Care was also made to ensure that those words were not instructed by teachers.

Materials

Materials included 160 isolated sentences (80 target sentences and 80 distractor sentences). The target word occurred one time in each target sentence, and each context provided sufficient clues to infer the meaning of the target words. For example, the word epidemic was introduced in the sentence “The success of controlling bird flu epidemic provides us with clear evidence that even some terrible diseases can be successfully overcome” (see more in Appendix I). The 80 distractor sentences were all high-frequency words. Target words were not included. The reason for designing 80 sentences was to ease the difficulties participants might have in the experimental process. In addition, this can avoid learners’ over-emphasis on the target words, which may compromise the results. Although it is suggested that incidental vocabulary acquisition should be tested through reading/listening to passages rather than sentences, it is easier to control the context in sentences. As suggested by Webb (2008), an accurate control for unknown words in a sentence may provide accurate assessment of the effects of the encounter.

Difficult words were avoided in designing all the materials. In this case, apart from the 15 target words, all other words were within the 2,000-word list when analyzing through the RANGE program mentioned above. Considering the participants’ vocabulary size, it could be assumed with confidence that participants knew all the non-target words. This is the best assumption that could be made without asking every participant to underline every word they did not know, and this would be impossible in considering the nature of incidental vocabulary acquisition. The total running words for the 160 sentences were 2,850, so unknown words made up no more than 5% of the total number. This complies with the findings that learners need to know at least 95% of running words to infer the meaning of unknown words from context (Van Zealand & Schmitt, 2012, 2013).

Measurement instruments

Four tests were applied to measure the participants’ word learning improvement (see Appendix II). Reliability and application of the four tests was shown in Eckerth and Tavakoli’s (2012) study.

The first test measured passive recall of form. Participants needed to write down the target words according to the given context. For example,

To make somebody frightened is to ---------- (intimidate).

The second test measured receptive recognition of form. Participants needed to circle the correct spelling from three distractors. Differing from Webb’s (2007) study, an I don’t know option was provided to avoid wild guesses. For example,

A. intimitate B. intimidate C. intimisate D. intimidade E. I don’t know

On the third test, which measured passive recall of meaning, participants demonstrated their understanding by producing a similar meaning in English. For example,

If you are intimidating somebody, you are ----------

On the fourth test, participants were required to select the word meaning from four options. To avoid wild guesses, an I don’t know option was also provided. They were encouraged to choose the I don’t know option if they did not know the word meaning. For example,

Intimidate

A. threat B. make something better C. surprise D. agree with E. I don’t know

There are three noteworthy issues for the four tests. First, the four tests were conducted separately and carefully sequenced to avoid an earlier test giving hints about a later test. Second, participants received one point for a correct answer and zero points for an incorrect answer. This score system was consistent in the four tests. The author scored the four tests, and the total score for each test was 15 points. Finally, the four tests were administered in written forms. This was to capture fully the development of vocabulary knowledge. If participants receive the four tests in spoken forms, they may misunderstand some words. Thus, tests in written forms may accurately measure the different levels of word knowledge.

Procedures

The experiment was completed during class hours (Table 2).

Four weeks before the experiment (Session One), the participants were required to take the VST test to ensure that the two groups did not have a significant difference in vocabulary size. In addition, pre-tests were administered to ensure that none of the participants had any prior knowledge of the target words.

In Session Two, the author printed the same materials for the participants to read or listen to. They were randomly assigned into two groups. One group listened to a native speaking instructor’s reading of the material. The book was read at his normal pace for teaching (about 90 wpm). A time interval (10 seconds) was necessary between sentences. In order to focus participants’ attention on listening, they were required to write down on paper what they had heard. However, the accuracy of what they had written down was not evaluated, as it was not the scope of the present study. The total length of Session Two for Group One was about 40 minutes. Participants immediately completed four surprise tests after listening. Another group read the materials. In order to focus their attention on reading, they were told to translate these sentences into their native language (Chinese). However, the accuracy of their translations was also not evaluated because translation was not the aim of the present study. To ensure the similarity of conditions for the two groups, the time allotted for reading was also 40 minutes. Participants also immediately completed four surprise tests after reading. Both groups were allowed as much time as needed to finish the tests, but they were not allowed to have materials at their disposal. The whole experiment of reading or listening plus tests lasted for about 70 minutes.

Two weeks later, in Session Three, the participants were asked to take the four tests again. To avoid the difficulties in designing equal-level tests for measuring retention, the delayed tests were identical with the four tests administered immediately after the experiment, except for the order in which the items were presented. During the two weeks, they did not have an opportunity to read the materials again, and the target words were not used in class instruction, and thus it was unlikely for them to be exposed to the target items. In addition, participants were not told they had to complete tests for the whole process. Thus, this experimental study could be defined as incidental (Hulstijn, 2013).

Question 1: The occurrence of incidental learning from reading and listening input

Table 3 summarizes the data for the two input modes and the four test types for the immediate test.

As shown in Table 3, the gain in form recognition for the reading-only condition was 7.2 out of the 15 words (48%). The gain in form recognition for the listening-only condition was less remarkable standing at 5.1 (34%). The form recognition gains were better than that reported by Brown et al. (2008), wherein the gains made in reading-only and reading-listening condition were 48% and 29%, respectively. The form recall results in the current study show that 3.2 of the 15 words were learned in the reading-only condition (21.3%), and 1.1 in the listening-only condition (7.3%). The mean scores for the meaning recall test were: 5.1 (34%) for reading-only condition and 2.5 (16.6%) for listening-only condition. The mean scores for the meaning recognition test were: 7.0 (46.6%) for reading-only condition and 4.0 (26.6%) for listening-only condition.

Friedman Tests were applied to compare the scores of the four immediate tests in the two different modes. This showed that the difference between the four tests in the reading mode was significant for the immediate post-tests (χ² (2, n=30) = 61.542, p <.05). Likewise, the four dimensions in the listening mode were significant (χ² (2, n=30) = 50.115, p <.05). The effect sizes can be regarded as large based on Cohen’s (1988) criteria and medium-large based on Plonsky and Oswald’s (2014) recently suggested more conservative criteria. Therefore, it can be inferred that learners encounter different levels of difficulties in learning the four types of word knowledge.

Wilcoxon signed-rank tests were performed to compare the gains made in the two conditions for each test type. There was a significant difference between the reading-only and listening modes for the four tests (p < .05), with large effect sizes according to Cohen’s (1988) criteria and medium-large according to Plonsky and Oswald’s (2014) criteria. This means that the participants in the reading mode scored better for the four dimensions of word knowledge than the participants in the listening mode. This also reveals that it is far more difficult to acquire words incidentally from listening-only than from the reading-only condition.

Question 2: The retention of target words for the two modes

Scores collected from the two-week delayed tests were calculated to measure the learners’ retention ability. The decrease in delayed data for the two input modes at the two test times is shown in Table 4.

Table 4 shows that there was a relatively large decrease over a two-week period in the scores for the four test types by the two input modes. The participants in the listening mode suffered the greatest decrease in retaining a form-meaning link. Although the scores on recognition of form and meaning remained level, recalling form and meaning remained very low. Recalling word form showed the largest decrease. Thus, the knowledge needed to recall the productive word form seemed to be the most difficult one.

Question 3: The effects of word exposure frequency for the two input modes

Table 5 shows the mean number of correct answers that the participants from the reading mode achieved on the four immediate and delayed tests in each frequency group (occurring 1 time, 5 times, 10 times).

Table 6 shows the mean number of correct answers that the participants from the listening input mode achieved. Scores of the four immediate and delayed tests in each frequency group were summarized.

Tables 5 and 6 suggest that the participants scored better on the words that they encountered more frequently. This result was consistent across the four test types in the two modes.

Kruskal-Wallis Tests were then applied to analyze whether differences in different frequency groups for each dimension of word knowledge were significant. The results of the four test types in the two modes showed a significant effect of frequency (p <.05). Results of the Wilcoxon Tests also revealed that the scores for the words that occurred 10 times were more than those when words occurred 5 times (p < .05), and scores for the words that occurred 5 times was also significantly better than those that occurred only one time (p <.05) for the participants in the two modes. The results were consistent in four test types.

Overall findings

In summary, the present study provides evidence that there is very little or no chance of picking up a new word unless the word is heard at least 10 times. For a word that is met only one or five times in listening mode, it is difficult to learn it, including recognition of its word form. This is in line with previous studies (Brown et al., 2008). Van Zealand and Schmitt (2013) provided evidence that for listening to be a valuable source for incidental vocabulary learning, it appears that considerably more than 15 exposures are needed. This may be because EFL students tend to demonstrate a weaker ability in lexically segmenting the input. In other words, it is difficult for EFL students to detect word boundaries in connected speech. Although a coverage rate of 95% was met, this coverage rate seemed to be insufficient for listening, as inferring meaning of unknown words for the participants in the present study was a challenge. As stated in Nation (2001), it might need a coverage rate of 98% for incidental vocabulary learning from listening alone.

The reading experiment in the present study replicates a previous study (Waring & Takaki, 2003). The only difference was in the use of measurement tools. Multiple choice and meaning translation tests were used in Waring and Takaki’s (2003) study. The present study attempted to find out about the form-meaning link of a new word. However, similar findings are as follows: First, participants can only acquire a word that is encountered a sufficient number of times. However, the precise number of times the reader should meet a word is not clear. In Waring and Takaki’s (2003) study, more than half of the items that occurred 8 times were acquired. Chinese EFL learners in the present study could acquire at least 70% of the words that occurred 10 times. The results of the present study are in line with Teng (2015). Second, word knowledge that participants had acquired decayed if the target words were not met again soon after reading.

There are some similar results in incidental vocabulary learning from reading or listening. For example, to some extent, the reading and listening modes are a source of incidental vocabulary acquisition. The present study provided ample evidence that recognition of the word form could be achieved incidentally from reading or listening, and that is the first step towards developing a form-meaning link. Followed by word form recognition is recognition of meaning, recall of meaning, and then recall of word form, which is the general acquisition profile (Webb, 2007). This means that the type of instrument that is used to assess incidental vocabulary learning outcomes is related to the degree of success. The data also showed that the more frequently a word is encountered, the greater chance there is for students to build a form-meaning link. Moreover, it appeared that incidental learning from reading or listening was limited. Only partial word knowledge was acquired.

In terms of the effects by input mode, reading was better than listening in incidental vocabulary learning. A greater number of times for encountering target words is needed in listening than reading for acquisition to take place. In addition, more word knowledge decayed over the course of two weeks in listening than reading.

Implications for teaching and learning

Repeated encounters with words. The study has shown the importance of frequency in incidental vocabulary learning by reading or listening. As Peters (2014) states, “an increase in frequency resulted in higher recall scores of the target item to be learned” (p. 89). However, the data in the current study was not sufficient to pinpoint the precise number of times that a learner needs to encounter a word in order to build a formmeaning link. As Chen and Truscott (2010) point out, “the goal of research should not be to identify a definitive number of exposures needed, but rather to understand a complex process involving multiple, interacting variables” (p.694). The current study adds to the understanding that EFL learners need more exposure for word knowledge when listening than when reading. For listening and reading to be a valuable source of incidental vocabulary learning for students, teachers should be mindful of the frequency of difficult words when preparing materials.

Combine inferring meaning from context plus elaboration of word processing. Although EFL learners should be encouraged to infer meaning from context, and lexical inferencing is also an essential strategy for promoting vocabulary acquisition (Teng, 2014b; Teng & He, 2015). The present study provides evidence that simply inferring the meaning of unknown words from context of reading or listening is not sufficient. The elaboration of word processing should be supplemented for vocabulary acquisition. In other words, if the teaching aim is not incidental vocabulary learning, teachers should provide explicit vocabulary instruction. For example, teachers can provide word meaning of the target words and relevant exercises of processing the words. Providing elaborate word processing helps learners establish a more refined and durable form–meaning relationship (Eckerth & Tavakoli, 2012; Laufer & Rozovski-Roitblat, 2011).

Rehearsal of newly learned words. Another implication drawn from the current study was that learners needed to rehearse newly learned words. This means the teacher can help students review the target words after reading or listening. The results of the present study have shown that even word knowledge that had the greatest gains on the immediate test still suffered a significant decay in gains over a two-week period. Thus, in order to curb precipitous declines in word retention, teachers should provide opportunities for learners to recycle the target words frequently.

Incidental vocabulary learning from listening. The results of the present study showed EFL students have difficulties in incidental vocabulary learning by listening. It is evident that teachers of Chinese learners of English should not expect students to listen at the same 95% known vocabulary coverage rate at which they can read. When teaching listening, two unknown words in 100 words are preferable. This also has resonance for many global contexts where English is being taught and learned as a foreign language, especially for those learners whose first language phonological systems are significantly different from that of English.

Limitations and future directions

There were several limitations in the current study. The first was that it did not cover the effects of elaborate word processing, which is also an important factor in incidental vocabulary acquisition. Future studies on improving learners’ incidental vocabulary gains by comparing the effects of word exposure frequency and elaborate word processing are necessary.

The second limitation is that the present study only evaluated the form-meaning link of acquiring a new word. As stated in Teng (2016a), knowing a word means more than form and meaning. Future studies on measuring more dimensions of word knowledge are needed, such as collocation.

Third, as the sentence context provides more informative information for guessing the meaning of unfamiliar words, it would be more difficult for learners to infer the meaning when the context provides limited information or hints (Teng, 2016b). This is an issue that should be included when interpreting the findings in the present study. In addition, the study did not separately measure the learners’ proficiency in listening and reading comprehension, which is acknowledged as an influential factor in incidental vocabulary acquisition.

Fourth, although distractor sentences were developed to ease the difficulties participants might have in the experimental process, reading/listening to 160 sentences might bore them. In addition, there is a difference in listening to academic texts and spoken English. This issue was not covered in this study.

Finally, exposures to target words occurred within a close time period in the present study. If exposures were spread out over time, which is called “distributed practice” (Nakata, 2008), the results might have been different. Future studies on investigating this are needed.