L2 Word Frequency Effects

NB: We do not resell your papers. Upon ordering, we do an original paper exclusively for you.

NB: All your data is kept safe from the public.

Table of Contents

Introduction
Research Study
Theory
Analysis and Discussion
Conclusion
References

Introduction

The frequency effect is a widely recognized mechanism for lexical access, suggesting that the more frequently a lexical item is utilized, the quicker it is recognized. In their study Duyck(2008) compared word recognition frequency effect for bilingual individuals in their second language (L2). The researchers argue that the results demonstrate that the frequency effect can be explained by implicit learning. The theory of Rank Hypothesis proposed by Murray and Forster (2004) attributes FE to serial search in frequency ordered bins. This paper will critically examine the research study by Dyuck (2008) and discuss how well the Rank Hypothesis Theory explains the observed L2 word frequency effects.

Research Study

The research study by Duyck (2008) sought to evaluate lexical decision in the context of the word frequency effect on visual word recognition between L1 and L2 for Dutch-English bilingual individuals. The researchers justify their study by arguing that while the word frequency effect (FE) is recognized, there is a debate concerning the details of how the effect works; different models suggest alternate mechanisms to explain the effect. Furthermore, the researchers argue that no study has evaluated FE in a second language, which is inherently interesting since while the objective frequency of words is relatively similar in both languages, words in L2 are understandably less frequently encountered by a bilingual individual. The aim by Duyck (2008) was to study bilingual speakers to juxtapose FE in first language and second language word recognition.

The results demonstrated that mean reaction times were slower in L2 compared to L1 for both high and low frequency words. Participants responded more slowly to low frequency than to high frequency words, but the difference was larger in L2 than L1. Comparison indicates that the frequency effect was vital for both first language and second language, but much smaller for L1 (46 msec) than for L2 (103 msec) (Duyck, 2008). In a second experiment the researchers tested monolingual English participants, who demonstrated the same results for their primary language as the participants did for L1 (which was their native), indicating that the language-frequency interaction was not a variable inherent to English stimuli.

Theory

The rank hypothesis is based on the ‘bin’ model of lexical access and consists of two part of stages. The 1^st stage compares the input letter strings of the text being read with the orthography details stored in the mind of the reader. These orthography details are an array of lexical entries categorized into multiple subsets known as bins, which are essentially an orthographic access system. The words in each ‘bin’ share orthographic properties. The location of each word is computed into a hash-code functions of the input letter string. Upon finding a match, the information about the fuller word and its meaning is retrieved from the ‘master file’ (Forster, 1992). The rank hypothesis is concerned about the ordering of terms in each bin, suggesting that entries are ordered by frequency of occurrence in each bin for the most optimal efficiency during search. The bin model is vital as it allows for an expansive serial search of the whole lexicon in theory (Murray & Forster, 2004).

Access time in the bin model is based on relative frequency of occurrence. The ranking position of lexeme in its ‘bin’ is what determines excess time. In other words, the relative frequency ultimately decides the rank position of each word. Murray & Forster (2004) provide an example, that taking two randomized ‘bins’ – even if the first word in Bin A is 10,565 per million while in Bin B the first word is 7,198 per million, the access search time will be determined by rank in those respective bins alone, not the total frequency. Furthermore, the access time between the 1st and 10^th entries in each bin are virtually the same, even though the 1st and 10^thitems in each bin will have different frequencies. While frequency differences may be large in the high end, and smaller in the low end, the difference in reaction time should be the same. As search speed is linear within the bin, access time only depends on the rank position of the word in the frequency order, known as the rank hypothesis (Murray & Forster, 2004). Murray & Forster (2008) also clarify that the rank function is not specifically the function of frequency as it is normally understood. The rank hypothesis indicates that the access time is dependent not on the mathematical frequency, but a word’s ranking position in a frequency ordered list.

Analysis and Discussion

The authors Duyck (2008) directly discuss the application of Murray and Forster’s rank hypothesis model to their findings. They suggest that their results indicate constraints to the theory, potentially in the context of bilinguals. Murray and Forster assume that bins contain lexemes only for the single language and that bins for L1 and L2 are ranked similarly. Thus, even if a bilingual has encountered a word in L2 less frequently than its L1 equivalent, it is likely that the word would occupy the same place in the bin (i.e. top 10) for both languages (Duyck, 2008). That holds since the rank hypothesis theory emphasizes that the frequency is dependent on the respective orthographic bins not absolute frequency. Furthermore, reaction times within certain ranges, such as the top 10 words within bins, should be relatively the same (Murray & Forster, 2004). Therefore, theoretically, language should not be a barrier in the context of word identification and reaction times for a bilingual individual, with both languages should maintain similarly large FEs, and the researchers proved it was not an issue of the English language. As Duyck (2008) effectively explains,

“Because frequency ranks of translation equivalents’ lexical entries are similar across languages, FEs, originating from rank differences during lexical search, should also be similar; that is, the difference in lexical search time through the L1 bin between HF and LF words should be comparable to the difference in search time through the L2 bin between HF and LF words of similar corpus frequencies” (p. 852-853).

The only means by which the rank hypothesis model could apply in the situation is if a variation of it is used which suggests that bins in bilinguals include items from both languages, which goes against the theory that these persons have language-specific lexical access. The bins would be organized based on the same principle of serial rank of frequency. Therefore, L2 words which are less frequently encountered and used would be lower than L1 words. Since the word count in a frequency rank decreases as frequency band increases, the individual processing has to consider many more words and variations in L2 that are lower on the rank than they would for L1. Duyck (2008) indicates that the same frequency manipulation of bin contents would influence reaction times in lexical search much more for L2 than for L1 which agrees with their findings.

The arguments presented by Duyck (2008) are rational and seem to be supported by both empirical evidence of their and other studies as well as anecdotal evidence. The word recognition frequency effect is generally slower for L2 in bilingual and learning individuals. That is supported by lexical access and linguistic cognitive processes down the line as well. For example, Kaan (2014) evaluate how reading velocity impacts the processing of sentences in the second language. Even though their results showed no difference in reading speed and proficiency, the difference for the L2 group could be seen in the end-of-sentence verification data. L1 speakers have less difficulty computing and maintaining syntactic representations than the L2 speakers. The researchers argue that this occurs as the more difficulty linguistically the task is, the more resources it requires, and the difference become more apparent (Kaan, 2014).

Murray & Forster (2004) even discuss this element as an alternative to rank hypothesis, proposing that the FE reflects differences in the amount of learning. The FE is evident in skill acquisition (practice), educational level, and even age, as for example older adults with decades of reading experience demonstrate greater FE than for young adults. They argue that this supports the idea of relative frequency which influences access time, not absolute frequency, but some cases still require absolute frequency (Murray & Forster, 2004). Potentially the similar concept of the learning effect can be applied to language. Even if an individual is bilingual, Duyck (2008) proves that the native language reaction times and accuracy will remain higher because L2 will by all aspects remain a ‘learned language.’ Inherently, it is a skill that requires greater cognitive and lexical resources, even if the words have similar frequency within language-specific bins. That is supported by Murray & Forster (2008) which suggest that lexical decision time should not be used as complete measure of the time taken for the lexical access.

Conclusion

In conclusion, based on the findings of Duyck (2008) the rank hypothesis proposed by Murray and Forster (2004) does not fully explain word FE on visual word recognition between languages in bilinguals. Theoretically, the language-specific bins sharing similar frequencies of words would draw the similar frequency effect and reaction times. However, data demonstrated that L2 perception was consistently behind in reaction time and accuracy for both HF and LF words. Alterations to the traditional rank hypothesis model such as that languages can share the same bin or the presence of a learned effect on frequency offer better rationalizations to the experimental results.

References

Duyck, W., Vanderelst, D., Desmet, T., & Hartsuiker, R. J. (2008). The frequency effect in second-language visual word recognition. Psychonomic Bulletin & Review, 15(4), 850–855. Web.

Forster, K. I. (1992). Memory-addressing mechanisms and lexical access. In R. Frost & L. Katz (Eds.), Orthography, phonology, morphology, and meaning (pp. 413–434). Amsterdam: North-Holland.

Kaan, E., Ballantyne, J. C., & Wijnen, F. (2014). Effects of reading speed on second-language sentence processing. Applied Psycholinguistics, 36(4), 799–830. Web.

Murray, W. S., & Forster, K. I. (2004). Serial mechanisms in lexical access: The rank hypothesis. Psychological Review, 111(3), 721–756. Web.

Murray, W. S., & Forster, K. I. (2008). The rank hypothesis and lexical decision: A reply to Adelman and Brown (2008. Psychological Review, 115(1), 240–252. Web.

Click Here To Order Now!