Student Speech: Email, Chat, and Spoken Language of Korean University Students

The results of research into the language learning potential of two forms of computer mediated communication (CMC)—asynchronous email and synchronous chat—have indicated that there are a number of advantages to the use of these forms of language production, both in terms of language produced and learner attitudes. This research focused on a detailed analysis of the language production of three classes of Korean freshman university students to assess how their language varied across the three forms of spoken language, email and synchronous chat. Class A wrote emails to each other, Class B communicated through synchronous chat, and students from Class C were recorded during small group conversations. The language was then analyzed for complexity and accuracy, and compared across a range of variables. The results showed that there was not a clear distinction between language produced in asynchronous email, language produced in synchronous chat, and language produced when engaging in facetoface communication. There were no clearly identifiable features to suggest they formed distinct language categories. Accordingly, it is suggested that other factors, such as task type and purpose of communication, may have as much influence on the language produced as the particular form in which it occurs.

157 STEM Journal, Vol 18, No 2, 2017 Student Speech: Email, Chat, and Spoken Language of Korean University Students Were, Kevin T (Kookmin University) Were, Kevin T (2017) Student speech: Email, chat, and spoken language of Korean university students STEM Journal, 18(2), 157-178 The results of research into the language learning potential of two forms of computermediated communication (CMC)—asynchronous email and synchronous chat—have indicated that there are a number of advantages to the use of these forms of language production, both in terms of language produced and learner attitudes This research focused on a detailed analysis of the language production of three classes of Korean freshman university students to assess how their language varied across the three forms of spoken language, email and synchronous chat Class A wrote emails to each other, Class B communicated through synchronous chat, and students from Class C were recorded during small group conversations The language was then analyzed for complexity and accuracy, and compared across a range of variables The results showed that there was not a clear distinction between language produced in asynchronous email, language produced in synchronous chat, and language produced when engaging in face-to-face communication There were no clearly identifiable features to suggest they formed distinct language categories Accordingly, it is suggested that other factors, such as task type and purpose of communication, may have as much influence on the language produced as the particular form in which it occurs I INTRODUCTION Since the early 1990’s, research into the use of the asynchronous exchange of e-mail in language learning has found that students produce more language when communicating through email compared to when they engage in face-to-face communication Beauvois (1998) concluded that CMC increased class participation, Beauvois (1995) and GonzalezBueno (1998) noted that students were more willing to participate in an electronic context compared to face-to-face interaction, and Gonglewski (2001) found email lists created 158 Were, Kevin T opportunities for authentic communication Gonzalez-Bueno (1998) even identified the same increase in productivity when comparing email with writing in-class paper and pencil assignments, suggesting that the difference may be more than that of the distinction between the immediacy of speaking and the ability to plan in writing Research into synchronous CMC (internet relay chat) has produced similar results with some research showing that learners’ face-to-face communication improves after participation in CMC activities (e.g., Beauvois, 1995; Blake, 2000; Kern, 1995; Warschauer, 1996), indicating that language skills may be transferable from CMC to face-to-face communication Kern (2006) notes, however, that we still need to learn how to make the best uses of computers to accomplish specific language learning goals, posing the question: What kind of language does the learner engage in during a CMC activity? Savignon (2007) notes that we are used to making a distinction between oral language and written language, but with the development of electronic communication it is becoming more difficult to find a difference between the two, although CMC may be closer to oral discourse than to written discourse Warschauer (2000) asserts that with the advent of computer based communication human interaction now occurs through a textbased form, and the division between speech and writing has been overcome, with the interactional and reflective aspects of language merged in a single medium, i.e CMC In addition, studies conducted on the social dynamics of CMC have found that it results in communication which is more equal in participation than face-to-face discussion, with those who normally don’t contribute to discussions benefiting most from the increased participation – those with less power and authority In Warschauer’s (2000) view CMC is creating new discourse communities, new literacies and new identities – a different language learning In my own classes I introduced an email component as part of the assessment for my freshman English language conversation classes Students were required to send the instructor a series of emails throughout the semester as part of their participation assessment in the class During this process it was noticeable that students communicated far more in their emails than they did when they were required to speak during class This could be explained simply as the difference between writing and speaking As a written medium, email allows students more time to plan and process their language output than is available when they have to communicate orally Communicative classroom environments emphasizing group-work also favor more social, outgoing students, so it could be expected that quieter, less outgoing students would feel more comfortable using a written medium where they aren’t face to face with other interlocutors It is also possible that task type and the speaking/writing difference contributed more to the difference than any other factor Other studies have researched the differences between one form of CMC and different Student Speech: Email, Chat, and Spoken Language of Korean University Students 159 forms of language production (most often speaking), but Abrams (2003) notes that research examining the possible differences between synchronous CMC and asynchronous CMC in terms of lexical richness, density, and syntactic complexity is not yet available The research described in this paper was devised in order to provide a contribution to fill this gap, while also comparing the CMC language varieties of Korean freshman university students with their spoken language output II LITERATURE REVIEW Writing and Speech EFL conversation classes typically focus on communication and teachers strive to maintain a good percentage of class time where students are listening to or speaking English While this oral practice is necessary, the development of communicative ability is achieved through more than simple listening/speaking time-on-task, however Savignon (2007), for instance, writes that, “the principles of communicative language teaching (CLT) apply equally to reading and writing activities that involve readers and writers engaged in the interpretation, expression, and negotiation of meaning” (p 213) In assessing the difference between speaking and writing, Biber (1988) studied a wide range of linguistic features taken from diverse text types, and concluded that the range of linguistic variation across spoken/written texts could not be reduced to a single factor, and there were “few, if any, absolute differences exist between speech and writing” (p 385) Halliday (1989) showed how the process of speaking was a different representation of language than the product of writing, while also noting that computer mediated communication (CMC) has further blurred the distinction between spoken language and written language Payne and Whitney (2002) write that the only cognitive difference between spoken language and CMC is step of Levelt’s articulator/speech comprehension system (Levelt, 1989) In this model of language production, utterances are conceptualized as pre-verbal thought (step 1) Then a preverbal message (communicative intention) enters the formulator for lexical access and grammatical and phonological encoding (step 2) After this, either speech-motor functions are utilized (step 3) to produce the utterance, or it is sent back through the speech comprehension system as sub-vocalized internal speech to check accuracy or comprehension Based on this model, it is argued that speaking and chatting are almost identical psycho- linguistically, with their only major difference being the involvement of the articulator (the speech-motor system, step 3) Shekary and Tahririan (2006) also note that in CMC interaction learners need not be concerned with such issues as pronunciation, hesitations, status, gender, race, and so on 160 Were, Kevin T In this view, language is a purely cognitive process and language learning is akin to structuring language competence with performance merely its externalization However, we still need to learn and develop recognizable pronunciation and intonation to be understood when speaking, develop listening ability, and confront those sociopsychological factors that play a large part in face-to-face interaction in order to function adequately when speaking in an L2 If, as Smith (2003) has argued, negotiation in CMC requires a new type of interaction, it may only be an intermediate model that has to be built upon by developing those skills necessary for oral interaction Ko (1996) found three distinct groups of occurrence when comparing linguistic features relating to interactivity and informativity in spoken, written and synchronous CMC language use, concluding that CMC output was more similar to spoken language overall as it was “interpersonally involved, syntactically fragmented, with a low degree of information focus and elaborateness” (p 35) These similarities of synchronous CMC interaction to spoken language as well as its benefits for second language learning have been elaborated by other researchers; for instance, Blake (2000), who asserts that CMC interaction produces similar language learning benefits to oral interaction, particularly in relation to collaborative tasks Chun (1994), Kern (1995), and Kitade (2000) all point out that CMC resembles oral communication although being written Uptake and Language Transfer Research results often assert that learner use of CMC satisfies the interactional conditions for optimal language learning, providing opportunities for learners to interact and negotiate meaning with an authentic audience, and allowing involvement in authentic tasks in a learner-centered atmosphere with ideal stress/anxiety level (Egbert & Jessup, 1996) At times it appears as if the research data only tenuously fits the theoretical construct however Hamzah’s (2004) examples of focus-on-form instances more often seem to be negotiations for clarifying meaning instead Smith (2003) presents data where an interlocutor says ‘OK’ at the end of an interaction routine which he interprets to indicate understanding of the meaning of what was said This could be interpreted in other ways though—for instance, a desire to move on and lack of interest in negotiating any further Ellis and Barkhuizen (2005) cite the research of Hawkins (1985) who showed how learners often fake comprehension Even confirmation of understanding doesn’t imply uptake will become procedural knowledge Shekary and Tahririan (2006) point out in this regard that in an interaction routine it is important that the final reply expresses the correct structure for uptake to occur Smith (2005) found though that, even if uptake occurs, there is a lack of positive effect of learner uptake in a SCMC environment: “uptake had no effect on whether target items were acquired or not” (p 51) Student Speech: Email, Chat, and Spoken Language of Korean University Students 161 The same tendency to fit data to the construct can be found in Kitade’s (2000) study where it is noted that feedback indicates students think their Japanese level has improved through chat (IC interaction) leading to the conclusion that the learners language improved through the collaborative meaning-making activity The following comment by the same student remains unacknowledged, however: “It would have improved even more if someone had observed our interaction and provided guidance for my Japanese” (p 160) In this case at least it appears that the interaction alone is not enough and indeed there are often social constraints on the amount of negotiation learners will undertake (Ellis & Barkhuizen, 2005) While the quantification of interactional processes and language features provides a model of the nature and form of language use in CMC, and it is asserted that CMC competence can be transferred to spoken discourse, there is little evidence for this In comparing the effects of synchronous CMC and asynchronous CMC on oral output, Abrams (2003) found an increase in quantity of language in the SCMC group but no significant difference in lexical density, diversity, or syntactic complexity amongst the three This is not enough to show transfer of competence and in fact her results support the view that transfer does not occur, which she explicitly states in relation to asynchronous CMC while confirming the obvious point that group discussion benefits oral interaction Liu (2011) similarly notes that CMC could help to prepare learners to participate in face-to-face communication, but it could delay language learning for intermediate and advanced learners if they become too accustomezd to untimed online chatting Language Mode and Task Type In contrast to the view that CMC represents a new approach to language learning, researchers such as Salaberry (2000) note only that, “CMC may constitute a new communication environment capable of creating new conditions for language interaction” (p 22) This note of restraint (‘may’) tempers the enthusiastic embrace of CMC as a pedagogical tool found in other studies to a consideration of exactly what pedagogical benefits may be available from incorporating CMC into language learning In noting that the use of technological resources should serve the pedagogical design of learning activities and that design must incorporate the constraints of the medium, he gives the example of the reduced means of expression in email which may force learners to focus more on morphological features of L2 useful for development of lexicon, morphology, and syntax There are different dynamics of interaction in online environments compared to face-to-face communication, and Kern (2000) asserts that, as a result, teachers need to develop new tasks, find new ways to guide and monitor interaction and to evaluate 162 Were, Kevin T performance, but it is still not clear how to transfer language abilities across the different modalities Halliday’s (1989) dictum that in “making language work for us in ways it never had to before, it will have to become a different language in order to cope” (p 82) has some support through the research into use of CMC, but the proclaimed benefits for language learning may be premature In some cases the results could be as much a function of the type of task as the mode in which it is performed, a conclusion reached by Bygate (1999), Crookes (1990) and others Yanguas (2010) for instance, found similarities in language in task-based interaction across different modes, determining that task nature may have been more significant than language mode in the type and focus of learner negotiations Finally, in a meta-analysis on the relative effectiveness of interaction in CMC and face-to-face contexts using journal articles and dissertations published between 1990 and 2012, Ziegler (2013) found no significant differences between the two modes, suggesting that mode of communication has no significant impact on L2 learning outcomes and language development III RESEARCH METHODOLOGY This study was designed to assess the comparative English performance of groups of Korean freshman students at Kookmin University in three different language modes: small group face-to-face discussion, paired interviews, paired email interaction, and paired internet chat (IRC) It is a descriptive research project, prepared with the objective of comparing how the different production modes affect language output and, in particular, whether the CMC results conform to data and conclusions from other studies Previous research (for example, Kern, 1995; Ko, 1996) has generally found a continuum of more complex and accurate language production in written modes to less complexity and accuracy in spoken language Part of the objective in this research was to see if this applies to the language output of this group of students, and if their language output, based on measures of complexity and accuracy, could be categorized as belonging to three distinct types Research Hypotheses The research hypotheses are: 1) There will be more accurate language use in CMC compared to face-to-face communication 2) There will be more complex language use in CMC compared to face-to-face Student Speech: Email, Chat, and Spoken Language of Korean University Students 163 communication 3) On a scale measuring complexity and accuracy, email will rank highest, followed by chat, and speaking will rank lowest Participants Students in three Freshman English conversation classes, a requisite one semester course for all students at Korean universities, were used as subjects for the research The students are from classes in Business IT, Korean Literature, and Communication Studies with a low-intermediate language level Class numbers ranged from 6-20 students Classes were taught the same subject matter in class, but were asked to additional assignment work in the different language modes The fact that students from the researcher’s classes were used as subjects for this study presented a number of difficulties The first was the ethical problem of using one’s own students for research The second was the requirement for all students to follow the same syllabus This made it problematic to employ an experimental research project that necessarily would require different treatments Accordingly, it was decided to maintain the same main syllabus content as input for all three groups and to give the same subject matter as additional assignment work for the groups, but to ask each group to present it in a different language mode The classes were consequently designated according to the following schema: Class A: email; Class B: internet chat; Class C: small group (3 - students) oral discussion Data Collection Classes all used the text English Firsthand (Helgesen, Brown, & Wiltshier, 2010) as the primary study material for the semester and class time centered on listening and pair or small group activities The data for analysis came from homework assignments based on topic themes in the text related to class content Students were required respond to the topics according to their group designation 1) Class A: Email Students were required to write a minimum of emails each between pairs of students on the topic Each email had to have a minimum of sentences In practice this meant that the amount of output reflected each student’s motivation or interest, from those who wrote the bare minimum of six single clause sentences with five to seven words in each (e.g “I feel sympathy for your problem I a part-time jobe in weekends But I bought too many 164 Were, Kevin T things So I can’t save money till now I hope you get a side job soon And let’s save money toghether See you.”) to others with as many as twelve sentences of up to fifteen words with coordination or subordination 2) Class B: Internet Chat Students were given the same instructions as the email group and asked to write a minimum of at least 10 lines of output per student In all cases this was exceeded with the number of turns ranging from 15 to 36 3) Class C: Small Group (3-4 students) Oral Discussion Students were placed in groups of or and required to meet with the instructor weekly for weeks In preparation for the meetings each student was asked to prepare a short one to two minute response to the topic After each short presentation was given the group was asked to comment or ask questions for a total session time of 10 to 15 minutes All sessions were recorded using a Sony portable cassette recorder (TCM-400DV) and recordings were subsequently transcribed for analysis Data Analysis Printed copies of email and chat transcripts were used for data from groups A and B Transcripts from recorded conversations were used for the data from group C The analysis of the data was based on Foster, Tonken and Wigglesworth (2000) level two analysis of oral data using the following measures TABLE Measure Number of AS-Units Total no of words Measures of Language Output How Calculated Defined by Foster et al as … a single speaker’s utterance consisting of an independent clause or sub-clauses Examples of independent clauses from Foster et al (with / marking the clause boundaries): That’s right/ Turn left/ You go to the main street of Twickenham Examples of independent sub-clauses: how long you stay there/ three months/ Oh, poor woman (pp 365-6) The total output for each student during each session calculated as a simple count of tokens Student Speech: Email, Chat, and Spoken Language of Korean University Students 165 TABLE Measure Mean turn length Amount of subordination (no of separate clauses/total no of AS-units) Number of different verb forms Type-token ratio Measures of Language Complexity How Calculated The total number of words (tokens) divided by the number of turns for the speaker/interlocutor This includes both subordination and coordination as indicators of increased complexity in language production Examples: 1) Actually I can’t play very well but when I was young / I was very active person / and I can ah I am, I was good at sports (analyzed as AS-unit with subordinate clauses marked with /) 2) So I’m ah I’m won a prize at the running contest / but nowadays I don’t have time for exercising, sports / so I can’t very well in sports,// but my parents manage a bowling center /so when I was young I always bowling in my free time, /so now I can well, uh I’m good at bowling (analyzed as AS-units divided at //, the first with subordinate clauses, and the second with subordinate clauses.) Verb forms identified were tense (present, past, future), modality (should, have to, must.), aspect (simple, continuous, perfect) The total number of different words used (types) was divided by the total number of tokens Tokens equaled the total number of words from each particular student It is important to use text segments of equal length as type/token measures decrease as text length increases (Ellis & Barkhuizen, 2005) and accordingly, this measure was followed with segments of 50 words each TABLE Measure Error-free clauses (errors per AS-unit)1 Target-like use of verb tenses Target-like use of vocabulary Measures of Language Accuracy How calculated Errors per AS-unit (Bygate, 1999) Total number of error free clauses divided by total number of independent clauses, subclausal units and subordinate clauses Number of correct finite verb phrases divided by the total number of verb phrases The percentage of clauses without lexical errors measured as the total number of lexical errors divided by the total number of ASunits Lexical errors are comprised of: (1) errors of omission, where a linguistic unit or units would have to be supplied in order to eradicate the error; (2) errors of over-suppliance, where a linguistic unit or units would have to be deleted to eradicate the error; (3) errors of permutation, where the order of linguistic units would have to be changed to eradicate the error; and (4) errors of substitution (a combination of and 2), where a linguistic unit or units would have to be deleted and another or others supplied to eradicate the error (Lennon, 1991) These include missing articles, copula, modal or other vocabulary word, prepositions; incorrect verb tense or aspect; incorrect word order; added words that not belong 166 Were, Kevin T IV RESULTS Results for Language Output TABLE Number of AS Units and Total Word Number Email Chat (averages for 10 (averages for students) students) No of AS Units 26.30 26.33 Total word number 204.60 151.17 Measures Speaking (averages for 10 students) 15.80 161.90 The higher number of AS units for both email and chat reflects the fact that in both CMC modes the AS units have less total words than in the spoken data What is noticeable is how short the conversational turns are in the chat transcripts Of the three transcripts (each between students) there is only one where sentences are extended into longer coordinated or subordinated structures Overwhelmingly, turns are structurally simple, containing one idea unit with no elaboration Examples from the chat data: Example A: How long have you been diet? B: Ummm … about one month A: We have to get some exercise Example A: What is the best animation you like? B: Ummm A: A lot of animation like … The exchanges are typically short in the manner of quick conversation In comparison, turns in the oral data alternate between longer explanatory utterances and short exchanges such as the following: Example A: Ah, yes we, ah our, our, our, … BIT students, business IT students went to Gangchon for membership training … this year B: This semester? A: Yes … Student Speech: Email, Chat, and Spoken Language of Korean University Students 167 B: Aha… A: …we went there … Generally though, the spoken exchanges tend towards being more extended coordinated or subordinated speech, such as in these examples: Example Most uh foreigner taste that and say it’s very spicy and hot but I enjoy this point Example I’m good at mathematics, science, … Korean, … but I didn’t study, I didn’t study because I’m, I was always play computer game so my brain goes to stupid … Results for Hypothesis There will be more accurate language use in CMC compared to face-to-face communication TABLE Results for Language Accuracy (Averages of Raw Scores) Measures Email Chat Error-free clauses (errors per AS-unit) 0.38 0.54 Target-like use of verb tenses 0.83 0.88 Target-like use of vocabulary (% of clauses without lexical 0.73 0.46 errors – total lexical errors/total no of AS-units) Speaking 0.37 0.72 0.75 As a proportion of total clauses, the higher percentage of error-free clauses in the chat data supports the view that sentence or text length may be a more significant indicator of accuracy than the language mode In accord with the expected order of accuracy, spoken language has the lowest target-like use of verb tenses compared to the data from both CMC modes Examples from the spoken data show the lack of consistency in the use of past tense verbs: Example Ah we booked train ticket we go we went there … ah we, rode train (Self correction) Example But my best teacher is high school grade 3… he often talk student and … (Present tense used instead of past) 168 Were, Kevin T Example One day I waked up early very early time because my, ah the first class started at nine I wake up and watch the clock … I spend about one hour to go to school I had 10 minutes left me And I prepared very quickly In the measure for target-like use of vocabulary, the lower the figure the less lexical errors there are The most accurate target use of vocabulary is in the chat data suggesting that there is a relationship between accuracy and length of text: the longer the text = the more complex the text = the more possibility of error and non-target vocabulary use In summary, the data in this study failed to find a regular patterning in the distribution of linguistic features across language samples Ellis and Barkhuizen (2005) have noted that the difference between planned and unplanned language is also significant in the type of language produced by learners In this analysis, both this factor and the amount of language produced appear to be as significant determining factors as the mode of language production Results for Hypothesis There will be more complex language use in CMC compared to face-to-face communication TABLE Results for Language Complexity (Averages of Raw Scores) Measures Email Chat Mean turn length 7.69 5.18 Amount of subordination (no of separate 1.19 1.06 clauses/total no of AS-units) Number of different verb forms 3.40 3.17 Type-token ratio 0.73 0.75 Speaking 11.10 1.34 2.55 0.63 Only two measures (verb forms and type-token ratio) support the hypothesis The mean turn length showed that turns in the spoken language were longer than either of the CMC forms Longer turn length in sentence production correlates with increased subordination and coordination In this measure, then, the effect of length of utterance appears to be the determining factor in the result For example, in this utterance, the student changes the verb in the repeated phrase until she finally gets to the past tense form that she wants: Example I’m just like smile and laugh and, and also I, … mmmm, I, I am I want, I wanted to be a Student Speech: Email, Chat, and Spoken Language of Korean University Students 169 comedian ah in elementary school The most frequent verb forms in the email data were: simple present, simple past, infinitive and forms of ‘be.’ Other forms were used but generally in only single or isolated instances These included modal forms (can, could, would, couldn’t, etc.), future, and past perfect forms In both the email and chat data a larger variety of auxiliary verb forms and more modal verbs were used There are no modal verbs in Korean language so their lack of use here may reflect the difficulty Korean learners of English have in understanding how and when to apply these Similarly, the future forms of English require the use of auxiliary forms, most commonly ‘will’ or ‘going to.’ It is noticeable that use of the future tense is almost non-existent The avoidance of their use here indicates a possible difficulty these students have with this form also In the email data seven types of verbs made up 87% of the total with 79.7% of these being either simple present, simple past or the copula In the chat data, 65% of verbs are present simple, past simple or the copula with a higher percentage of both modal and auxiliary forms as well as the future tense In the spoken data the same verb forms (simple present, simple past or the copula) account for 79.7% of total verbs (88% if ‘like’ and ‘want’ are included*) with auxiliary verbs, the infinitive and present progressive form accounting for another 11%, but there are no modal forms, few auxiliaries used and only one use of the future tense This may have been due to the nature of the tasks assigned in the spoken interviews which tended to be about past events but in both the email and chat tasks students were free to write about future events When they did, future forms were still not used This can be seen in the following two examples Example 10 ‘I try to apologize him’ [instead of I will try to …] Example 11 ‘If I and you have a problem, we talk about that like this’ [- instead of we (will/should/have to/must or other auxiliary use) talk about that …] Present tense verbs are more often used in unplanned speech styles and could indicate the difference between planned (writing) and unplanned speech (spoken language) with CMC as a kind of intermediate form yielding results between speech and writing This variation does not occur in the present study, however, suggesting either that all three modes represent similar unplanned language production styles or that in all three modes students lacked confidence in using the past tense and therefore overused the neutral form (present tense) to compensate for this This appears to be the case in instances such as the 170 Were, Kevin T following examples from the spoken transcript (the topic was ‘my best teacher’): Example 12 …my best teacher is high school grade three He always pick fighting to his student he often talk student and … uh … teach … In two further examples from the email transcripts, past and present tenses alternate in the same sentence: Example 13 The most interesting dream what I dreamed is that famous actor and me fall in love Example 14 In a dream, I was locked in a room That place was so dark, so I can’t see anything This kind of inconsistency can be seen predominantly in use of the past tense Future tense verb forms, on the other hand, were almost always regularized to the present tense form As the correct use of tense is fundamental in language use this difficulty points to an underlying lack of acquisition of very basic rules Beyond the difficulty with simple tenses lie the more complex challenge of modals, auxiliaries, perfect aspect, passives and other verb forms which are used with even more imprecision TABLE Examples From the Transcripts Data Sure, we make an appointment Are you finish report? Have ever been these dreams? In fact I usually not dreaming If I and you have a problem, we talk about that like this It’s so hard because time is shortage and money is also lack You need to patience because you love him Shall you a part-time job with me? Interpretation Missing modal/auxiliary ‘Be’ form of verb used instead of ‘do’ form ‘Be’ form of verb instead of ‘have’ form Participle instead of present tense verb Mixed ‘I’, and ‘you;’ missing modal Noun/adjective confusion and general verb form instead of participle Missing auxiliary Incorrect auxiliary; missing verb In the chat data the simple past occurs as only 5% of the total but forms of the copula are 27% of total verbs used In comparison, the related figures for email are 17% and 17%, and for spoken language the figures are 11% for past tense and 22% for the copula Adding the two figures gives totals of 32%, 34% and 33% respectively, indicating a similarity across the three modes in the use of these two verb forms with a possible correlation Student Speech: Email, Chat, and Spoken Language of Korean University Students 171 between the two as a lower figure for the past tense appears to be correlated to a higher figure for the copula in both spoken language and chat Use of the present tense is identical for the email and spoken language data, both with a figure of 36%, with chat slightly less at 33% These three verb forms account for 75% of all verbs used in the email data, 72% in the spoken data and 65% in the chat data In each of these cases, there appears to be a closer similarity between the spoken data and email with more variation in the chat data One further distinction can be seen in the use of modal verbs where both CMC forms are clearly differentiated from the spoken data In the chat data 3% of verbs are modal forms, in the email data 4% of verbs are modal forms, but in the spoken data there are no modals used at all Biber (1988) lists possibility modals (can, might, could, may, etc.) as indicators of lesser certainty about statements made In this data, there was a complete absence of these modals in spoken language, indicating a lack of acquisition for these particular students Other research has consistently found that L2 learners generally use modally modified expressions less often than native speakers (Holmes, 1988, for example) There were, however a relatively high proportion of both ‘can’ (13 times) and ‘could’ verbs (5 times) in the email data compared to chat (‘can’= and ‘could’= 0) This result, viewed together with the wider range of auxiliary verb use in email is the only indicator that, of the two forms of CMC, email is closer to writing (and therefore has more complex language use) than chat, a possible indicator of the (temporal) communicative constraints of the immediate discourse production required in synchronous CMC that is not such a significant factor in email The length of chat turns is less than half that of the spoken language with each turn having, on average, four more words than the email turns Perhaps the medium of internet chat necessitates short replies with its imperative to send a quick response to an interlocutor waiting unseen on another computer Just as in face-to-face interaction there would be no conversation if one partner had to wait an undue length of time – however long this may be - for each reply This effect may be exacerbated by chatting in an L2 without access to the quick processing and replies that would come automatically to someone interacting in their L1 Along with the brevity of the replies the content in the majority of the chat transcripts was also relatively simple A typical example of this is as follows: Example 15 A: I become very clean B: I envy you A: hohohohoho B: Are you happy? Hahahahaha 172 Were, Kevin T A: Yeah! B: Over the long time I go to a public bath A: there is so many people B: a public bath is so crowed A: Oh, yes Because today is holiday B: hahahaha I know In a second example, the turns are longer in length when the topic is more serious, discussing future plans, but the turns are shorter when this changes to a lighter mood, paralleling normal spoken discourse: Example 16 A: Next year, I will have to go army But if I get more than 780 point at TOEIC test, I can apply for KATUSA So I’d like to improve my English ability until next year B: wow… are you studying English now? A: No I am not Because there are lots of things to in this semester So I am going to start to study English on winter vacation B: no wonder you write English very well… A: haha Thank you I think English is so stressful thing to Korean What you like to on Next Year? B: I will quit this school… Just kidding… Maybe I fall in love with someone and date everyday S1: ‘O’!!!!!! Really?? School friend? S2: No … It’s just my hope … -_-; Type-token ratio is one of the two measures where both CMC forms indicate greater complexity than the spoken language form One possible reason for this result is because of the short replies that characterized the mean length of turn in both email and chat In fact there appears to be an inverse relationship here, as the language mode with the lowest mean length of turn (chat) has the highest type/token ratio; spoken language, with the longest mean length of turn has the lowest type/token ratio, and the email data falls in the middle Type/token measures decrease as text length increases (Ellis & Barkhuizen, 2005), but in this case the language segments were of equal length (50 words) The result here suggests that the actual clausal and phrasal units are more reliable indicators of the text Student Speech: Email, Chat, and Spoken Language of Korean University Students 173 length In this data, segments were counted in groups of 50 words for the analysis but a data set of 10 x word phrases/sentences would maintain features of short text length compared to a data set of x 10 word phrases/sentences Both are measures of complexity, but they indicate that the text length is more significant than the language mode in determining complexity Results for Hypothesis On a scale measuring complexity and accuracy, email will rank highest, followed by chat, and speaking will rank lowest Six measures were found where CMC showed either more complexity or more accuracy than spoken language, two measures where spoken language showed either more complexity or more accuracy than CMC, and one measure where spoken language lies between the two CMC modes There is no clearly identifiable pattern to indicate such an order This hypothesis is therefore not supported by the data V DISCUSSION AND CONCLUSION One measure of interactivity is a high occurrence of present tense verbs which indicate unplanned discourse (Ko, 1996; Ochs, 1979) In the data for this research, synchronous chat has very few past tense verbs (only 5% of total verbs) with 33.3% present tense verbs, indicating a high degree of unplanned speech style which, in this sense, is as speech-like as the spoken language data (which had 36% present tense verbs & 11% past tense verbs) Synchronous chat has also been found to be more accurate but shorter in length and more productive of informal language that serves an interactional requirement at a relatively superficial level with regard to topics and content This result suggests that different tasks may lead to quite different results and that these other factors also need to be taken into account It could be considered if there is any connection between Biber’s two parameters of informativity and interactivity and the measures of complexity and accuracy used in this study Ko (1996) found that electronic discourse followed the spoken discourse pattern in relation to features of interactive discourse (for example, frequency of pronouns and proverb do) creating an interpersonal focus and a less formal style In the data for this study the spoken language was longer in mean turn length with more subordination, but these measures of complexity can also be misleading, as the increase in the amount of language does not equate with a corresponding accuracy; in fact it has become less accurate in the data presented here 174 Were, Kevin T In addition, there are fewer verb types and a lower type/token ratio in the spoken language which not only indicates less complexity but also suggests a lack of risk-taking, a possible function of socio-psychological factors such as status, social anxiety, and communication apprehension (Young, 1990, 1991) that restrict or limit a learner’s willingness to attempt using more complex forms—as they increase the possibility of inaccuracy If learners in this study were more willing to take risks in the non-spoken language forms this may reflect the fact that CMC represents a language experience dissociated from these socio-psychological considerations It is interactive but without the interpersonal complications that face-to-face communication presents, without the requirement to engage Levelt’s step articulatory/speech comprehension system and without the need to process variations in language from another interlocutor before a response can be given, things such as pronunciation variations or logically obtuse responses, unfamiliar vocabulary, idiomatic speech styles, etc During the recording of interviews for the oral data for this study, a number of students expressed their difficulty with speaking English through their involuntary physical actions such as clapping, snapping fingers, flicking or hitting their legs, sighing, making clicking sounds with their mouths, and laughing excessively to hide embarrassment They obviously felt under enormous pressure to produce the relatively simple language that resulted This kind of pressure is not apparent in the decontextualized written modes of CMC The greater accuracy found in the CMC may be explained by the fact that more accurate language is needed in the decontextualized environment of CMC in order to convey clear meaning It doesn’t explain, however, why the chat data was more accurate than email, which is also decontextualized The explanation for this may lie in the interactivity of the synchronous chat, a functional quality that seems to demand short connected statements in order to maintain a conversation Email, on the other hand, with less immediacy in interaction, affords more time for topic consideration resulting in more informative and more elaborate text and, as in the spoken data, accuracy is compromised the more the language is extended Features which occurred with the lowest frequency in CMC chat include (1) type-token ratio, its reduced lexical diversity explained as reflecting the real-time communicative demand competing with the pressure to write quickly and the physical constraint of having to type, and (2) prepositional phrases, suggesting that under its combined time/production constraints chat discourse expresses simple idea units with basic clauses without optional adjuncts (such as prepositional phrases) – a feature more common in discourse with a greater informational load In this study the type-token ratio and the target use of vocabulary are both highest in synchronous chat which indicates a higher informational load and reduced real-time communicative demand In fact the data indicates that students Student Speech: Email, Chat, and Spoken Language of Korean University Students 175 don’t appear to be limited by the real-time communicative demand of chat and facing an increased cognitive load; rather it appears to enhance their communicative ability In comparison, the higher frequency use of the ‘want’ and ‘like’ verbs in the spoken data points to language that is concerned with expressing personal preferences more than with interaction or seeking information from others This research has shown that, for the sample of Korean learners of English, the use of CMC modes of language production does not result in language forms that are clearly identifiable into distinct categories There may be new possibilities for language learning through the incorporation of CMC language modes, but, on the basis of this research, it is still not clear if they will bring about significant changes in the way that second languages are acquired The data in this study rather seems to verify the importance of underlying communicative function and context of the language in determining its structural and grammatical forms (Biber, 1988) The language of two friends discussing their lives and experiences may not be so different if they are communicating face-to-face or through the medium of CMC What may count more is the immediacy of the communication and its purpose The final point in Abrams’ (2003) study that CMC be seen as a tool of interaction in its own right would seem to be a more fruitful approach than one which seeks to place it somewhere between the conventional duality of spoken and written language This research has been based on small sample sizes collected during one university semester More comprehensive data gathered over a longer period of time may yield more reliable or more valid results Furthermore, a larger scale study that compares CMC and other language modes across a range of similar tasks may give further insight into the range of possibilities and limitations of CMC as a means of developing second language acquisition REFERENCES Abrams, Z I (2003) The effect of synchronous and asynchronous CMC on oral performance in German The Modern Language Journal, 87(2), 157-167 Beauvois, M H (1995) E-Talk: Attitudes and motivation in computer-assisted classroom discussion Computers and the Humanities, 28(3), 177-190 Beauvois, M H (1998) Conversations in slow-motion: Computer-mediated communication in the foreign language classroom The Canadian Modern Language Review, 54(2), 198-217 Biber, D (1988) Variation across speech and writing Cambridge, UK: Cambridge University Press Blake, R (2000) Computer mediated communication: A window on L2 Spanish 176 Were, Kevin T interlanguage Language Learning & Technology, 4(1), 120-136 Bygate, M (1999) Quality of language and purpose of task: Patterns of learners’ language on two oral communication tasks Language Teaching Research, 3(3), 185-214 Chun, D M (1994) Using computer networking to facilitate the acquisition of interactive competence System, 22(1), 17-31 Crookes, G (1990) The utterance and other basic units for second language discourse analysis Applied Linguistics, 11(2), 183-199 Egbert, J L., & Jessup, L M (1996) Analytic and systematic analysis of computersupported language learning environments (Unpublished doctoral dissertation) University of Arizona, Tuscon, AZ Retrieved from http://www.cc.kyotosu.ac.jp/information/tesl-ej/ej06/aj.html Ellis, R., & Barkhuizen, G (2005) Analyzing learner language Oxford: Oxford University Press Foster, P., Tonken, A., & Wigglesworth, G (2000) Measuring spoken language: A unit for all reasons Applied Linguistics, 21(3), 354-375 Gonglewski, M (2001) Using email in foreign language teaching: Rationale and suggestions The Internet TESOL Journal, 7(3) Retrieved from http://iteslj.org/Techniques/Meloni-Email.html Gonzalez-Bueno, M (1998) The effects of electronic mail on Spanish L2 discourse Language Learning & Technology, 1(2), 55-70 Halliday, M A K (1989) Spoken and written English Oxford: Oxford University Press Hamzah, M (2004) Facilitating second language acquisition (SLA) in a computermediated communication (CMC) learning environment Internet Journal of eLanguage Learning and Teaching, 1(1), 15-30 Hawkins, B (1985) Is the appropriate response always so appropriate? In S Gass & C Madden (Eds.), Input in second language acquisition (pp 162-178) Rowley, MA: Newbury House Helgesen, M., Brown, S., & Wiltshier, J (2010) English firsthand Hong Kong: Pearson Longman Asia ELT Holmes, J (1988) Doubt and certainty in ESL textbooks Applied Linguistics, 9(1), 2144 Kern, R G (1995) Restructuring classroom interaction with networked computers: Effects on quantity and characteristics of language production The Modern Language Journal, 79(4), 457-476 Kern, R (2000) Literacy and language teaching Oxford: Oxford University Press Kern, R G (2006) Perspectives on technology in learning and teaching languages TESOL Quarterly, 40(1), 183-210 Kitade, K (2000) L2 learners’ discourse and SLA theories in CMC: Collaborative Student Speech: Email, Chat, and Spoken Language of Korean University Students 177 interaction in internet chat Computer Assisted Language Learning, 13(2), 143166 Ko, K K (1996) Structural characteristics of computer-mediated language: A comparative analysis of interchange discourse The Electronic Journal of Communication, 6(3) Retrieved from http://www.cios.org/EJCPUBLIC/006/3/ 006315.HTML Lennon, P (1991) Error: Some problems of definition, identification, and distinction Applied Linguistics, 12(2), 180-196 Levelt, W J M (1989) Speaking: From intention to articulation Cambridge, MA: The MIT Press Liu, Z (2011) CMC and face-to-face communication in L2 learning Columbia University Working Papers in TESOL & Applied Linguistics, 11(2), 23-24 Ochs, E (1979) Planned and unplanned discourse In T Givon (Ed.), Syntax and semantics 12: Discourse and syntax (pp 51-80) New York, NY: Academic Press Payne, J S., & Whitney P J (2002) Developing L2 oral proficiency through synchronous CMC: Output, working memory, and interlanguage development CALICO Journal, 20(1), 7-32 Salaberry, M R (2000) L2 morphosyntactic development in text-based computer mediated communication Computer Assisted Language Learning, 13(1), 5-27 Savignon, S J (2007) Beyond communicative language teaching: What’s ahead? Journal of Pragmatics, 39(1), 207-220 Shekary, M., & Tahririan, M (2006) Negotiation of meaning and noticing in text-based online chat The Modern Language Journal, 90(4), 557-573 Smith, B (2003) Computer-mediated negotiated interaction: An expanded model The Modern Language Journal, 87(1), 38-57 Smith, B (2005) The relationship between negotiated interaction, learner uptake, and lexical acquisition in task-based computer-mediated communication TESOL Quarterly, 39(1), 33-58 Warschauer, M (1996) Comparing face-to-face and electronic discussion in the second language classroom CALICO Journal, 13(2), 7-26 Warschauer, M (2000) The death of cyberspace and the rebirth of CALL English Teachers’ Journal, 53(1), 61-67 Yanguas, I (2010) Oral computer-mediated interaction between L2 learners: It’s about time! Language Learning & Technology, 14(3), 72–93 Retrieved from http://llt.msu.edu/issues/october2010/yanguas.pdf Young, D J (1990) An investigation of students’ perspectives on anxiety and speaking Foreign Language Annals, 23(6), 539-567 Young, D J (1991) Creating a low anxiety classroom environment: What does language 178 Were, Kevin T anxiety research suggest? The Modern Language Journal, 75(4), 426-437 Ziegler, M (2013) Synchronous computer-mediated communication and interaction: A research synthesis and meta-analysis (Unpublished doctoral dissertation) Georgetown University, Washington, DC Retrieved from https://repository library.georgetown.edu/bitstream/handle/10822/559497/Ziegler_georgetown_007 6D_12341.pdf?sequence=1 Applicable level: university education Keywords: computer mediated communication, language production, language complexity, language accuracy, language variables, task type Were, Kevin T Department of English Education Graduate School of Education, Kookmin University 77 Jeongneung-ro, Seongbuk-gu, Seoul, 02707, South Korea E-mail: ekevin1@gmail.com Received: April 9, 2017 Revised version: May 14, 2017 Accepted: May 21, 2017