A Study On Chinese EFL Learners' Processing Of Formulaic Sequences

Formulaic sequences play an important role in native-like selection and native-like fluency (Pawley&Syder,1983). They are found to constitute a large part of native-speakers'spoken and written discourse. However, second language learners are often found to overuse, underuse or misuse formulaic sequences, which lead to their unidiomatic linguistic output. A large number of researchers have attempted to explain how formulaic sequences can contribute to native-like selection and native-like fluency. The holistic processing hypothesis proposed by Wray and Perkins (2000) contends that formulaic sequences tend to be stored and retrieved as a whole, making them easier to use in communication. The Network model (Bybee,1985,1995) proposes the holistic representation for frequent morphologically complex words. Based on her model, Sosa and MacFarlane (2002) propose that larger units, such as non-idiomatic phrases and constructions, may also be stored in the lexicon and processed as single units. Many other researchers put forward similar assumptions. However, until recently, the holistic processing of formulaic sequences remained a theoretical assumption. Not many empirical studies were carried out to prove it.The past decade has witnessed a renewed interest in the role of formulaic sequences. Some researchers attempted to test the holistic processing hypothesis by experimental studies. Their findings generally support the view that formulaic sequences are processed faster than non-formulaic ones among native speakers, but whether it plays the same role among second language learners still needs further study.Some other studies examined the role of various factors in the processing of formulaic sequences among native speakers and non-native speakers. They found that frequency, mutual information (MI), and n-gram length influence native speakers and non-native speakers in different ways. Based on this finding, Ellis et al.(2008) propose a model of acquisition where frequency exerts more influence at the beginning stage of language learning and MI exerts more impact at a more advanced level. However, in their study, they did not consider the factor of familiarity, which might influence learners'processing of formulaic sequences, too. In addition, all previous studies take advanced second language learners as participants and do not include lower proficiency learners in their study. Neither do they adopt a cross-sectional study with learners at different proficiency levels and attempt to analyze the characteristics in processing formulaic sequences by learners at different proficiency levels. The frequency count of formulaic sequences in previous studies is mainly based on native speaker corpus and may not reflect the actual input characteristics for second language learners.On the basis of the above analysis, the present study takes a psycholinguistic approach to investigate the factors that influence Chinese EFL learners'processing of formulaic sequences. More particularly, it endeavors to explore:(a) how Chinese EFL learners process formulaic sequences;(b) compared with native-speaker controls, how frequency, mutual information (MI), n-gram length and familiarity of formulaic sequences influence Chinese EFL learners'processing of formulaic sequences, and (c) how the different factors can be explained by a model of language acquisition proposed by the author in this study. Six hypotheses were established as follows:H1:Chinese EFL learners process formulaic sequences faster than non-formulaic sequences.H2:Chinese EFL learners process higher-frequency formulaic sequences faster than lower-frequency formulaic sequences.H3:Chinese EFL learners do not process higher-MI formulaic sequences faster than lower-MI formulaic sequences.H4:Chinese EFL learners process larger n-gram formulaic sequences more slowly than smaller n-gram formulaic sequences.H5:Chinese EFL learners process higher-familiarity formulaic sequences faster than lower-familiarity formulaic sequences.H6:Chinese EFL learners at different proficiency levels process formulaic sequences in different ways.These hypotheses were investigated in three tasks, i.e., an eye-tracking reading task (ETRT), a grammaticality judgment task (GJT), and a reading task (RT). The latter two experiments were carried out using the E-prime software measuring participants'reaction times.20Chinese college non-English major students representing two English proficiency levels participated in the eye-tracking reading task.37Chinese high-school students representing two English proficiency levels and7native speakers took part in the grammaticality judgment task.21Chinese high-school students representing two English proficiency levels and2native speakers participated in the reading task.Empirically, the current research finds that frequency, MI, familiarity and proficiency influence Chinese EFL learners'processing of formulaic sequences in different ways. First, the study finds that Chinese EFL learners process formulaic sequences faster than non-formulaic sequences. Frequency and MI are found to influence Chinese EFL learners'processing of formulaic sequences. The higher the frequency of a formulaic sequence, the faster Chinese EFL learners process it. But for MI, the higher the MI score of a formulaic sequence, the slower the processing speed on the part of Chinese EFL learners. N-gram length does not influence Chinese EFL learners'processing of formulaic sequences, but it is a significant factor for native speakers'processing of formulaic sequences. Familiarity influences Chinese EFL learners'processing of formulaic sequences to a certain extent. The correlation between proficiency and processing speed is partially supported by the data.The results, to a large extent, confirmed the hypothesized developmental route. As the subjects in the present study represent the mid-proficiency and low-proficiency group, the data complement the model proposed by Ellis et al.(2008). Their model deals with advanced second language learners and native speakers and proposes that frequency influences learners at a lower proficiency and MI influences them at a higher proficiency. The data from the present study found that MI also influences lower proficiency learners, but in a different way. Instead of being a facilitating factor in language processing, it poses as a barrier in low proficiency learners'processing of formulaic sequences.The contribution of the study lies in the following aspects.Theoretically, the research finds that the nature of formulaic sequences among EFL learners differs from that of native-speakers. According to Wray (1999), formulaic sequences are stored and used as a whole among native-speakers. When they acquire their mother tongue, they memorize sentences and phrases they hear and later start to separate them and form rules. From the present study, it is found that frequency and familiarity are both influencing how a person processes a sequence. If learners are very familiar with a certain sequence, they will process it faster than other sequences. In other words, this processing advantage comes from exposure and practice, not from holistic storage in the first place. Learners process unfamiliar formulaic sequences in a similar way when they approach novel non-formulaic sequences, which indicates that their first reaction towards them is to use rules to interpret them instead of understanding them as a whole.Secondly, the present study is the first one in the field to include the role of familiarity in EFL learners'processing of formulaic sequences. Previous studies include frequency as a factor but neglect familiarity in the study of the processing of formulaic sequences and get inconsistent results. The addition of familiarity helps to shed new light on our understanding of the processing of formulaic sequences by EFL learners.Thirdly, the present study investigates the role of frequency, MI, familiarity, n-gram length and proficiency on the part of Chinese EFL learners. Most previous studies use advanced second language learners as participants and native speakers as controls. But only studying these two groups of subjects can not reveal the process of development on the part of the learners. Therefore, a cross-sectional study with learners at different proficiency levels is needed to explore how learners proceed when their proficiency level is enhanced. The advanced second language learners in existing studies are mostly college students studying in English-speaking countries. In contrast, Chinese EFL learners study English mainly through classroom instruction and practice. As one of the objectives of the present study is on the role of frequency in learners'processing of formulaic sequences, the input characteristics are important to the results. Therefore, taking Chinese EFL learners as subjects can help us to determine the difference between language learning in natural settings and in the classroom settings.Fourthly, the present study discusses a possible model of language acquisition and may deepen our understanding of the second language acquisition process. Ellis et al.(2008) found that in the processing of formulaic sequences, second language learners are influenced by frequency and n-gram length while native speakers are under the influence of MI and n-gram length. Based on their findings, they proposed a model of language acquisition. In their model, at the beginning of the language acquisition process, frequency and practice play a large role. Frequent exposure and practice can consolidate the mental representation of a linguistic form. However, learners still need to establish an association between the linguistic forms and their functions in communication. They gradually acquire the form-function mapping of the language. So they will be more influenced by MI. The higher the MI score of a formulaic sequence, the faster it will be processed. But in their model, they did not take into account the lower level learners'linguistic development process. Therefore, with the addition of intermediate learners as subjects, we complete this model of acquisition. Fifthly, the present study uses a textbook corpus to provide data on the input characteristics. Previous studies do not have such input characteristics and only use intuition or borrow from native speaker corpus to infer the input characteristics, which can not reflect the true properties of learners'input. In the present study, a corpus is made based on textbooks learners have learned in all their previous school years. It can be regarded as a more accurate source of learner input, as Chinese EFL learners mainly learn the language from classroom instruction and most of their language input is based on their textbooks.Sixthly, the present study utilizes on-line methods to study Chinese EFL learners' processing of formulaic sequences. In contrast to the traditional off-line methods, the on-line methods record learners'processing features while the subjects are processing the data. The results are more direct and are supposed to reflect accurate situations. The present study uses the eye-tracking paradigm to study the online processing of formulaic sequences among Chinese EFL learners.