Discussion My research examined the transcripts of four FTF courses, each consisting of nine classes which met for three hours, for a total of 108 hours of transcription. With the inclusion of the CMC courses [37], this resulted in the equivalent of about 130 hours of classroom dialogue.
Thanks to the scripts, I was able to examine nearly three times as much data as Bellack et al. and nearly 26 times as much as Sinclair and Coulthard's (1975) five-hour initial sample [38].
Once all of the sentences had been coded and the reliability of the coding determined, the data could be analysed. Each code had a certain number of records corresponding with each participant for each course; these were counted and converted to percentages as described in Appendix I. This process could be automated with an AppleScript, should one wish to perform a similar search among several databases; however, I was only examining one database, so it wasn't necessary.
Once the figures had been extracted from the database, I converted these amounts to percentages for comparisons across each course. So, for example, in Figure 11 one may view the percentage of teacher talk (ordinate) as a function of time (abscissa) for the average of the FTF courses. In the FTF Systems Analysis courses, much of the grade was based on a group project to be presented at the end of the course. Consequently, as the figure illustrates, the first 50% of the course indicates the teacher doing most of the talking. After this point, the students began using their new knowledge to ask questions and make their presentations. The grades in the FTF Database Management courses, however, were based on exams. As a result, although the quantity of teacher talk decreases 20% during the first 50% of the sentences of the course, in which students demonstrate their homework problems on the blackboard, it rebounds to build up for the final exam, and decreases slightly at the end for a discussion of the exam results.
Figure 11: Amount of teachers' talk in the FTF courses
The data for Purpose, Mechanism, and Subject were first entered into an Excel spreadsheet and then plotted on graphs, as in Figure 11. They were then examined for differences between teachers and students, modes of delivery, and the courses themselves. To minimise fluctuations and simplify each graph, the spreadsheets were reduced to ten-percent increments. These graphs permitted examination of the data by the dimensions of both quantity and time. This allows a detailed look at the nature of the classroom discourse.
As I described earlier, the coding is designed to look at discourse at the sentence level. To be thorough, though, I also analysed the average number of words in each sentence for all of the participants. Of course, one should keep in mind that any computer-generated word count will vary with the software used to obtain it. The text of the FTF teachers, for example, contained 401,571 words in FileMaker 3.0 (using the WordCount text function). This same text, when exported, contained 413,845 words (according to Word 5.1a) and 412,161 words (according to Word 6.0.1). The difference between FileMaker's sum and either of the Word programs occurred primarily because FileMaker ignores the dummy word character "~", rather than counting it as a word. The differences between the two versions of Word were probably due to different word-counting algorithms. For the sake of consistency, all word counts described in this work were derived from Word 6.0.1.
That said, an analysis of the amount of talk generated by participants is given in the following two tables. Table 1 indicates that of all of the FTF participants, the teachers say the longest sentences, with an average of 13.01 words per sentence. Both the male and female students utter shorter sentences, averaging 10.36 and 9.39 words per sentence, respectively.
| Role | n words | n sentences | words/sentence |
|---|---|---|---|
| Teachers | 412,161 | 31,689 | 13.01 |
| Male Students | 69,035 | 6,661 | 10.36 |
| Female Students | 46,493 | 4,950 | 9.39 |
Table 2 indicates that the CMC participants used longer sentences. The teachers, male students and female students increased the number of words in their sentences to 15.72, 13.79 and 13.89, an increase of 21%, 33% and 48% respectively. In addition to this increased quantity of words, the number of words in male and female students' sentences achieved more equal levels. Whereas in FTF interaction male and female students' sentences contained 80% and 72% respectively of the number of words in the teachers', in CMC these both increased to 88%.
| Role | n words | n sentences | words/sentence |
|---|---|---|---|
| Teachers | 66,530 | 4,232 | 15.72 |
| Male Students | 34,170 | 2,478 | 13.79 |
| Female Students | 28,793 | 2,073 | 13.89 |
In terms of sentences, the teachers still did most of the talking. In the FTF courses, the teachers uttered a total of 31,689 sentences (73%), compared with the male students' 6,661 (15%) and the female students' 4,950 (11%) [39]. In the CMC courses, the teachers provided an appreciably smaller percentage of the utterances, yet still wrote the most. Teachers wrote a total of 4,232 sentences (48%), compared with the male and female students' 2,478 (28%) and 2,073 (24%) sentences respectively.
An aspect of FTF interaction to be examined, specifically when discussing the nature of interpersonal interaction in the classroom, is the use of interruptions. Having coded sentences of the FTF transcripts in which interruptions were marked with an em dash, described earlier, it was a straightforward matter to search for the em dash in the text box of the database as well as by sex and/or role. This gave me the raw number of FTF participant interruptions, from which I was able to determine the percentage of interruptions (see Table 3).
| Role | n interruptions | in n sentences | amount |
|---|---|---|---|
| Teachers | 1,418 | 31,689 | .04 |
| Male Students | 989 | 6,661 | .15 |
| Female Students | 744 | 4,950 | .15 |
Although overall the teachers were interrupted more times than the students -- 1,418 -- the greater number of sentences spoken by them made this proportionally rare, occurring only four percent of the time. Both the male and female students were interrupted far more often, 15% of the time. Despite the difference in the number of interruptions they experienced, they were interrupted equally, due to their different numbers of sentences.
This discrepancy in the amount of interruption between the teachers and the students may be a result of the roles of turn-taking in classroom discourse (Atkinson, 1981), in which the options for the student are not the same as those of a teacher, for example, to interrupt. The fact that the teacher was interrupted less frequently might suggest the greater power the teacher has in the classroom, or there might be other elements at work.
Merely examining the extent to which FTF participants were interrupted only tells part of the story. The other side of the story is knowing by whom were these participants interrupted. This is important, for two reasons. First is the question of power dynamics; the students may have been interrupted by the teacher, or by other students. Second is the question of possible sex bias. It's one thing to know that both male and female students were interrupted 15% of the time, but it would be quite another to learn that a particular group was being interrupted by the teacher, or by another group.
To this end, I wrote the Who Interrupted Whom AppleScript. Given the record number of a sentence during which a class participant was interrupted, the script opens the database and determines the sex and role of the speaker of the next record, determining who interrupted whom. (A complete description of the script is given in Appendix K.)
An examination of the raw data (Table 4) indicates that the teachers interrupted male and female students about equally, at 84% and 80% respectively. This is not surprising, considering that the role of the teacher often embraces the privilege of interrupting students for the purpose of guiding dialogue.
| Role | n Teacher | n Male Student | n Female Student |
|---|---|---|---|
| Teacher | -- | 762 (.54) | 656 (.46) |
| Male Students | 827 (.84) | 66 (.07) | 96 (.10) [40] |
| Female Students | 595 (.80) | 90 (.12) | 59 (.08) |
Although it is to be expected that teachers will interrupt students more than students interrupt students, things tend to be more fair in inter-student interruptions. Male students were interrupted by other male students seven percent and by female students 10% of the time. Conversely, female students were interrupted by male students 12% and by other female students eight percent of the time. However, before any conclusions can be drawn about the nature of inter-student interruptions, the number of interruptions must be viewed in terms of their expected probability.
Although it is explained more fully in Appendix J, the basic idea of expected probability used in this example is that an examination of inter-student interaction must take into account the students' representation by sex. For example, if the Who Interrupted Whom AppleScript revealed that 75% of the interruptions of female students were done by a male student, it might be indicative of male students dominating the classroom. However, unless this is viewed with knowledge of the proportion of males in the courses, the information is inconclusive. If about 50% of the students in the course were male, then this would indicate that the male students were interrupting female students disproportionally often. On the other hand, if 99% of the students in the course were male, the contrary would be true (Paulos, 1995).
An analysis of the overall FTF course attendance records (see Appendix A) and the number of sentences uttered by students in each class appears in Table 5. In the courses observed, based on the ratios of males to females in each class of each course and the number of sentences which were uttered by students in each class, the expected probability of a male student interrupting another male student is .53. In a corresponding manner, the expected probability that a male student would be interrupted by a female student was .47. Since there were fewer female students overall, the expected probability that a female student would be interrupted by a male student was higher -- .61 -- and the expected probability that a female student would be interrupted by another female student was .39.
| Male Student | Female Student | |
|---|---|---|
| Male student interrupted by a | .53 | .47 |
| Female student interrupted by a | .61 | .39 |
The actual percentages of student-student interruptions, seen in Table 6, matched the expected probabilities in two of the four cases. The results indicate that female students were interrupted by other students in proportions which were closely tied to those students' representations in the FTF classes. Based on the distribution of student sex throughout the FTF courses, interruption of female students would be expected to have been performed by male students 61% of the time. In actuality, male students performed 60% of the interruptions of female students. In a complementary manner, female students interrupted other female students 40% of the time when it would have been expected to be 39% based on their populations within the FTF courses.
| Male Student | Female Student | |
|---|---|---|
| Male student interrupted by a | .41 | .59 |
| Female student interrupted by a | .60 | .40 |
The largest discrepancy came between the expected probabilities and actual results of who interrupted the male students. Based on their class representation, one would expect that 53% of the time that male students were interrupted it would have been by another male student. In actuality, though, this only happened 41% of the time.
Overall, female FTF students interrupted male FTF students more often than would be expected as a result of their sex's representation. The female students would be expected to interrupt male students 47% of the time when actually it turned out to be 59%. With regard to the other groups, the female interruptions matched expectations: One would expect that a female would interrupt another female student was 39% of the time when it turned out to be 40%. The expectation was that a female student would interrupt the teacher 44% of the time and it turned out to be 46%.
In a reciprocal manner, male FTF students interrupted male FTF participants less frequently than was expected based on their sex's representation. Male students would be expected to interrupt the teacher, female students or other male students 56%, 61% and 53% of the time respectively, when their actual results were 54%, 60% and 41% respectively.
What does this mean? It means that female students interrupted male students somewhat more often than would be expected based on their population in the FTF courses. And, correspondingly, male students interrupted other participants less than would be expected. By itself this does not reveal very much. These interruptions may be symptomatic of either the quick, interactive dialogue of a good discussion or rude interruptions of boorish monologues. To determine this, further examination of the results of the coding was performed to compare the results with the previously discussed models of interaction.
Since, as was discussed earlier, the recitation model is the norm for FTF classroom interaction, I shall examine the transcripts in terms of what one would expect to observe over the duration of the courses. (For an analysis of the average code distribution over the courses, see Appendix L.) Although some differences in the modes or interaction are to be expected due to the nature of the modes of delivery, these shall be addressed later.
There are several observable characteristics of talk which distinguish recitation from discussion. One characteristic of recitation is that the predominant speaker will be the teacher, who does two-thirds or more of the talking (Dillon, 1994; Flanders, 1970; Graddol, 1989). This was already addressed earlier in this work, in which it was revealed that FTF teachers (
= .73, sd = .06) spoke more than the CMC teachers ( = .49, sd = .10). However, this merely indicated the quantity of talk, rather than its quality. An analysis of classroom interaction should reveal differences in the nature of the interaction as well. If the recitation model were being used, one should be able to observe other indications of teacher dominance. One would expect to see teachers using the Purposes of Organising, Lecturing, and Idling more than students. In addition, one would expect teachers to use the Mechanisms of Filler and Rhetorical Device more than the students.
First, let's examine the use of Organising in the transcripts. In the recitation model, Organising fits squarely within the realm of the teacher. Quite simply, if one is reduced to a role of primarily answering questions, one has neither a need nor an opportunity to explain what one is going to say or do.
Pedagogically, Organising is used to indicate what the speaker intends to say. At its most basic level, Organising demonstrates control, for only a situation in which one has the freedom to speak or write in several possible ways requires the use of signposting or an advance organiser. Although one might argue that the statement "Tomorrow I shall do my homework" is Organising, in the context of the classroom it would most likely be a response to a teacher's Eliciting move.
Figure 12: Organising
Overall, due to the nature of CMC interaction, one would expect to see 6.5% more Organising from CMC participants [41]. The differences, however, were much greater (see Figure 12). Overall, CMC participants used Organising more ( = .22, sd = .06) than the FTF participants ( = .01, sd = .00). This difference between FTF and CMC uses of Organising may be, in part, due to the lack of non-verbal interaction. Whereas in FTF much can be communicated nonverbally, in CMC it has to be, quite literally, spelled out.
As expected, in both modes of interaction, the teachers used Organising more, with the FTF teachers ( = .02, sd = .01) slightly higher than the FTF students ( = .00, sd = .00) and the CMC teachers ( = .25, sd = .10) higher than the CMC students ( = .18, sd = .05). Although there is a slight increase in FTF student use of Organising in the last 20% of the courses, this is a result of the student presentations in the Systems Analysis courses.
Second, one should be able to observe other indications of teacher dominance, including a preponderance of Lecturing. As I defined it earlier, Lecturing is talk about the course content that is neither Organising, Eliciting, nor Responding. Like Organising, Lecturing requires one to have the control to be able to make statements. For example, a student Responding to a teacher's Eliciting move is neither Lecturing nor asking a question. In FTF classrooms, the teacher controls the dialogue, and may use his or her role to cut people off to prevent digressions -- such as students going off on long lectures. As a result, one would expect both the FTF and CMC teachers to have uttered most of the Lecturing sentences.
Figure 13: Lecturing
This, however, was not the case (see Figure 13). Although the FTF teachers used Lecturing more ( = .37, sd = .07) than the FTF students ( = .07, sd = .09), they were both surpassed by CMC participants. On average, the CMC teachers used Lecturing about as much as the FTF teachers, although with a great deal more variability ( = .40, sd = .17). The CMC students, however, used Lecturing the most ( = .59, sd = .08), ranging between 49 and 74 percent of the Purpose of their sentences.
These results suggest that the FTF courses approximated the recitation model, but the CMC courses, on the other hand, did not. This is most likely a result of the asynchronous nature of the CMC courses. These students were able to write about the course content for as long as they wanted before the teacher or other students saw it. They were not interrupted, nor forced to share a limited amount of class time. This freedom manifested itself as an increase in Lecturing. FTF students' largest use of Lecturing (which ranged from zero to 26 percent of their sentences) was in the last 30% of the course, a direct result of their Systems Analysis presentations.
A third indication of teacher dominance that would manifest itself according to the recitation model would be the use of Idling. In a FTF setting, Idling is mainly a reflection of a teacher's status, because, quite simply, teachers are more likely to get away with it. A teacher interested in having his or her questions answered is likely to interrupt a student who is blathering and pass the task to another student. The same teacher, however, is free to blather, as students are unlikely to challenge or interrupt. In the FTF courses, therefore, one would expect to see the teacher uttering more Idling sentences than students.
The asynchronous written discourse of CMC, on the other hand, should allow equal use of Idling by teachers and students. Both would have an equal amount of time to prepare, organise, and revise their thoughts. As such, one would expect to see a "cleaner" transcription from the CMC people, with fewer Idling sentences [42].
Figure 14: Idling
As Figure 14 illustrates, Idling was used relatively little in any of the courses, peaking at 4% for what is said by teachers in the first tenth of the FTF courses. As expected, the FTF participants used Idling more ( = .01, sd = .00) than the CMC participants ( = .00, sd = .00).
A fourth indication of teacher dominance would be the use of Filler. Like Idling, Filler is a manifestation of either hastily-prepared discourse or a means of stalling. Filler is often used to maintain control of the dialogue, so one would be more likely to see utterances coded as Filler from a teacher rather than a student.
As with Idling, Filler should be more apparent in synchronous than asynchronous discourse. First, teachers need to maintain control on a second-by-second basis, which they do not have to do in an asynchronous environment. Also, in CMC one has nearly unlimited time to think of responses, so stalling becomes irrelevant.
Figure 15: Filler
This prediction held true. As Figure 15 demonstrates, for the small amounts of Filler coded overall, the greatest amount was generated by the FTF participants. Specifically, the FTF teachers generated the most ( = .03, sd = .01), followed by the FTF students ( = .01, sd = .00). The CMC participants, as predicted, had the lowest percentages of Filler. However, the CMC teachers' levels ( = .00, sd = .00) were slightly lower than those of the CMC students.
A fifth, and final, manifestation of teacher dominance under the recitation model would be a greater use of the Rhetorical Device. Functionally, the Rhetorical Device serves as a means to demonstrate and maintain control, specifically for the purpose of guiding a discussion to a predetermined end. Moreover, there are two specific reasons why one would expect teachers to use Rhetorical Devices more than students. The first is that FTF students, being in a role of responding, i.e., having a lower status than the teacher, are more likely to respond with straightforward statements than by answering a question with a question. The second reason is that the use of Rhetorical Device requires a measure of time in which to speak in which one is confident he or she won't be interrupted; students don't tend to have this, while teachers do. Following this, one would expect to find that teachers use Rhetorical Devices the most and students the least.
Figure 16: Rhetorical Device
This prediction was proven both true and false in Figure 16. The prediction held true within the FTF courses in that Rhetorical Device was used by the FTF teachers ( = .03, sd = .00) the most and by the FTF students the least ( = .00, sd = .00). However, between these two extremes, the CMC teachers and CMC students used them the same amount, with ( = .01, sd = .01) and ( = .01, sd = .01), respectively.
It is possible that the increased use of Rhetorical Devices by the CMC students is caused by two factors. First, since the interactions were asynchronous, it would be impossible to interrupt a student. Second, in this asynchronous communication they would need to supply their own question and answer to have any form of immediate dialogue (Riedl, 1989). This would explain why CMC teachers and students use Rhetorical Devices the same amount.
It is also possible that some sort of threshold exists at which the more one talks -- or is permitted to talk -- the more one has an opportunity to become not merely a speaker or a classroom contributor, but an orator. And as an orator, one uses Rhetorical Devices. If this were the case, this would explain why the population uttering the largest percentage of sentences, the FTF teachers, would also use the Rhetorical Device mechanism the most. This, however, may simply be a function of oral discourse.
A second, fundamental characteristic distinguishing recitation from discussion is the nature of the exchange. In a recitational exchange, the teacher initiates with a question, the student responds, and only the teacher evaluates the response. This initiation-response-evaluation is the hallmark of conventional classroom interaction. In a discussion-oriented exchange, however, either the teacher or another student may also respond or evaluate. Continuing with the examination of the recitation model, one would therefore expect that the Purpose of most sentences from the transcripts should be either Eliciting or Responding.
Eliciting serves many purposes in the educational realm, including initiation of feedback "Did anyone have problems with the homework?" and commanding attention "Christopher, pay attention", but in the broadest sense, the underlying use of Eliciting is to get others involved. In the recitation model, since the teacher asks questions of the students, one would expect to see a great deal of Eliciting from teachers. In a similar manner, one would expect to see the least amount of Eliciting from the students.
Figure 17: Eliciting
This, however, was not the case, as Figure 17 illustrates. Overall, the FTF participants used Eliciting about five percent more than the CMC participants. Although the CMC teachers used Eliciting slightly more ( = .13, sd = .04) than the CMC students ( = .10, sd = .03), the FTF students used Eliciting the same amount more ( = .18, sd = .06) than the FTF teachers ( = .15, sd = .02).
Before dismissing the comparison to the recitation model, one should also examine the use of Response. In a typical exchange in a recitation, notes Dillon (1994), the teacher asks a question, a student answers it, and the teacher follows this with an evaluation and another question. In this traditional model of classroom discourse teachers ask a large number of questions, and students do a great deal of answering (Atkinson, 1981). As a result, one would expect to find Responding the most-used Purpose of students.
Figure 18: Responding
Although Figure 18 proves the prediction true for the FTF participants, the results were the opposite for the teachers and students in the CMC courses. In the FTF courses, the students used Responding about 17% more ( = .58, sd = .10) than the teachers ( = .41, sd = .08). In the CMC courses, the teachers used Responding about 11% more ( = .21, sd = .14) than the students ( = .10, sd = .06). It appears that Responding exists in a reciprocal relationship to Lecturing. Within the FTF courses, the teachers do more Lecturing and the FTF students do more Responding. In CMC, where the students do more Lecturing, the teachers do more Responding.
A related characteristic of discourse which distinguishes recitation from discussion is the overall pace of student-teacher interaction. Dillon (1994) notes that recitation is characterised by many brief, fast exchanges between teachers and students. In a discussion, however, there are fewer, longer, slower exchanges.
In examining the data to see if the observed classroom interaction follows this recitation model, there are two ways to approach the problem. The first is to examine the amount of teacher talk over time, as was done earlier in the Results section, only at a finer scale. For example, if a single course was composed of 10,000 sentences, one percent of the course would consist of 100 sentences. If students were permitted to speak for many sentences and several students were allowed to talk consecutively, this would result in a decrease in the amount of the teacher's utterances as shown on the scale. On the other hand, if students were restricted to short answers and were not able to respond to other students, then the amount of teachers' utterances would remain fairly constant.
Figure 19: Percentage of teacher talk
An examination of Figure 19 suggests that the FTF courses followed the recitation model whereas the CMC courses did not. As is evident in the graph, the amount the CMC teacher spoke varied greatly over time ( = .49, sd = .27). This is characteristic of a discussion rather than a recitation, in which the teacher may simply not talk at all for portions of the course. The FTF teachers, on the other hand, did most of the talking and spoke for a fairly consistent amount throughout the course ( = .73, sd = .10). Although they permitted the students to speak, they did not permit their students to speak for very long at any given time.
The second way to determine the overall pace of the classroom interaction would be to tally how many consecutive sentences were said by a teacher or a student in each course and construct a histogram to represent the data as I did with the Count Contiguous Sentences AppleScript (see Appendix M). For example, to illustrate how many contiguous sentences students or teachers uttered, one could construct a histogram in which the number of contiguous sentences were marked along the abscissa and their number of occurrences were plotted on the ordinate, as in Figure 20. A recitation would be revealed by an initial peak by both teachers and students, indicative of short turns. In addition, there would be an observable discrepancy between teachers and students. The teacher, tightly controlling the amount of talk in a recitation, would ensure that most of the students' responses were short. The teacher, however, being under no such controls, would have the option to take longer turns. A discussion, on the other hand, would be indicated by longer turns, primarily a decrease in the height of the initial peak. In addition, the histogram of both teachers and students would be somewhat similar.
Figure 20: Contiguous sentences
As Figure 20 indicates, the expected patterns appeared. There were large differences between the FTF and the CMC participants, particularly in the initial peaks of one- or two-sentence-long turn percentages. The FTF students peaked the highest in one or two-sentence turn percentages at 75 and 14% respectively. In a similar manner, yet not to as great an extent were the FTF teachers at 44 and 16% respectively. The CMC participants, however, had a much smaller percentage of short responses. The CMC teachers' percentages of one- and two-sentence turns were one and five percent respectively, and the CMC students' percentages were one and four percent.
As predicted, teachers in both the FTF and CMC courses took longer turns than their students. However, within their modes of communication, CMC participants took much longer turns than FTF participants. FTF teachers took longer ( = 5.19, sd = 11.43) turns than FTF students ( = 1.66, sd = 3.12), but these were still much shorter than the turns taken by CMC teachers ( = 36.48, sd = 43.05) and the CMC students ( = 22.20, sd = 21.06). Overall, CMC participants seemed to follow the expected pattern for discussion, while the FTF participants followed the recitation pattern.
Although the recitation model is the traditional norm for FTF classroom interaction, certain differences from the model are to be expected when a different mode of delivery, such as CMC, is involved. In other words, one would expect that due to the inherent properties of CMC one would see certain behaviours.
Based on the previous descriptions of the nature of CMC, one would expect to find great differences between it and FTF in several coding areas, specifically the Purpose of Humanising and the Mechanism of Opining. In addition, one would also expect to find a high agreement between the Purpose of Organising and the Mechanism of Repeating. Finally, based on a CMC model, one would expect to see a strong agreement between the Mechanisms of Explaining and Performing with the Subject of Procedure.
The first characteristic of CMC that one would expect to see a great deal of would be the use of Humanising. By my own definition Humanising consists of content-free dialogue used for the sole purpose of making people feel good. Some may consider Humanising as a sort of interactional garnish, decorative and ultimately edible, but hardly a substitute for the main course. Nevertheless, the literature recommends that teachers working at a distance make special efforts to involve students at remote sites or isolated students (Beauvois, 1995; Bruce & Shade, 1994; Monson, 1978). If this is the case, then one would expect to see more Humanising sentences from the CMC participants, particularly from the teachers.
Figure 21: Humanising
The data, as shown in Figure 21, do not completely agree, however. One would expect to find CMC participants using Humanising more than the FTF participants, and of the CMC participants, the teachers should be using Humanising more than the students. Although overall, the CMC participants did use Humanising more ( = .02, sd = .01) than the FTF participants ( = .01, sd = .00), the CMC students used Humanising for their sentences' Purpose more ( = .02, sd = .01) than the CMC teachers ( = .01, sd = .00).
Further examination of the figures indicate that both FTF and CMC teachers do some pure Humanising, about the same amount ( = .009 and .008, respectively). The FTF students also use Humanising approximately half the same amount, ( = .004). The CMC students, however, appear to making a more concerted effort to use Humanising in their interactions ( = .023). Although still representing a small percentage of the overall interactions, this is nevertheless used 2.5 to 5.75 times more than the teachers or the FTF students. This may be reflective of an informal and "chatty" style designed to compensate for the lack of non-verbal signals (Gray, 1989) and physical presence.
The second characteristic one would expect to see in courses taught via CMC would be an increase in opinion statements. In this work, this would be measured by an increase in sentences using the Mechanism of Opining. Based on the hierarchy of power in the classroom, one would expect to find that both the FTF and the CMC teachers, being in a position to give their opinions as they wish, would use Opining more than either of their students. The CMC students, being less constrained by social conventions, however, would be more likely to give voice to their beliefs and more likely to Opine. As a result, one would expect to find Opining used most by the teachers, then by the CMC students, and least by the FTF students.
Figure 22: Opining
The results, illustrated in Figure 22, reveal that this prediction holds mostly true. On average, the FTF and the CMC teachers do use Opining about the same amount ( = .04, sd = .01). In addition, the FTF students do use Opining the least ( = .03, sd = .01). While as predicted the CMC students do use Opining more than the FTF students, they also use Opining more than anyone else ( = .11, sd = .03). This finding concurs with the literature indicating that shy students, feeling confident to express themselves via CMC (Mabrito, 1991), would be more likely to "express their opinions" (Mason, 1989, p.129).
A third characteristic of CMC which one would expect would be a significant number of sentences coded as both Organising and Repeating. One would expect more of this type of sentence in a CMC context than in FTF, for two reasons. First, the purpose of the quoted sentence would be to establish context, to explain what one is going to talk about. (In my coding system, this is the Purpose of Organising.) Within synchronous communications, such as a traditional FTF classroom, context is provided by the contiguity of utterances. If a student is responding to a teacher or another student, unless he or she is talking about something further back in time, there is no need to establish the context for the response; indeed, it would be redundant and potentially annoying. In asynchronous communications, however, such as those in computer conferencing, one must establish the context to which one is responding. An excellent example of this pattern may be found in USENET newsgroups, where users typically respond to a post not merely by using the same subject line, but also by quoting a section of the previous person's text to set up context. Thus, one would expect to find more of this in the CMC courses. Second, since CMC makes it easy to quote someone else's text verbatim, this is coded as Repeating.
Figure 23: Organising and Repeating
As predicted, sentences containing both Organising and Repeating were used exclusively in CMC by both teachers and students. Whereas not a single instance of a sentence containing both of these codes was found within the FTF sentences, these constituted a measurable amount of the CMC sentences ( = .07, sd = .06). Further analysis of the CMC results indicate that the majority of these sentences were generated by the teachers quoting the students' contributions or questions. This shows the teachers quoting the students' text to establish the context for the answer.
A fourth characteristic of CMC that would be expected to distinguish it from FTF interaction would be the use of the Mechanism of Explaining or Performing coupled with the Subject of Procedure. This is because the basic forms of communication have to be explained.
Although participants in the CMC courses were all relatively knowledgeable about computers, one would still expect to see a measure of Explaining/Procedure used to explain how to do things related to the specific software, as well as use of Performing/Procedure to explain how to do things such as handing in homework or following up to others' posts [43]. This isn't directly related to the course content; this reflects the percentage of sentences used to explain why one does something or how to do something related to specific procedures, not the course content itself.
Figure 24: (Explaining or Performing) and Procedure
This prediction held true as illustrated in Figure 24. The CMC participants used Explaining and Procedure or Performing and Procedure more ( = .12, sd = .05) than the FTF participants ( = .02, sd = .01). I predict that as time passes, however, this difference between FTF and CMC participants will decrease, for two reasons. The first is the increasing sophistication of the hardware and software involved. Although there have been certain exceptions, over time hardware and software have generally become easier to configure and operate. Although I imagine it would be overly optimistic to predict that world-wide standards will emerge, I think it would be safe to assume that the trend toward increasing the ease of use in computer technologies will continue.
The second reason why I expect the difference between FTF and CMC participants to decrease is due to the increased sophistication of users. With continued exposure to the technologies, participants will need fewer sentences of explanation describing how to participate in a computer-based forum. Also, with continued exposure to the technologies, more students will have developed mental models that permit them to better operate and understand their hardware and software.
Vygotsky (1962; 1978) wrote that children use language as a tool to increase their understanding of the world around them and to transcend their physical limitations. Similarly, CMC permits one to transcend physical boundaries. Furthermore, it has the potential to change the role of the teacher from merely a disseminator of information and evaluation to that of a collaborator. Full and active participation in the CMC classroom gives all students an opportunity to validate their knowledge by negotiating the meaning of the course content with the teacher and other students. This negotiation permits the student to make learning a more active procedure, engaging the help of others, increasing his or her zone of proximal development, and ultimately helping him or her to learn.
Nowadays, teachers and students can use asynchronous CMC as a tool to communicate with each other and create a better collaborative learning environment. They will need some patience, since the technology is still in its infancy; fully 12% of the CMC utterances dealt with how-to material, versus only 2% in the more familiar FTF environment. Despite these practical limitations, CMC has a great deal to offer educationally.
From my research, I have found that discussion is one of the hallmarks of CMC. Of course, discussion can also take place in the FTF environment. However, teachers often fail to realise that they have slipped into recitation instead of discussion (Alvermann, O'Brien & Dillon, 1990; Connor & Chalmers-Neubauer, 1989).
To study these and other related phenomena, this case study examined the discourse used in FTF and CMC classroom interactions. The results indicated that, overall, CMC interaction resembled that of discussion whereas the FTF interaction resembled recitation. In the FTF courses, for example, the interaction among participants resembled that of a typical recitation (Dillon, 1994): The teachers uttered 73% of all sentences, and 58% of the students' sentences took the form of Responding. In comparison, in the CMC courses, the teachers generated only 49% of all sentences, and the students used Organising, Lecturing, and Rhetorical Devices in approximately the same percentages as their teachers. In addition, the CMC students took longer turns as measured by their number of contiguous sentences. All of these are elements typical of discussion.
The CMC courses also evidenced another characteristic of discussion, specifically the use of Opining sentences. The CMC students used Opining sentences 11% of the time in contrast to the FTF students, who used Opining the least (3%), and the FTF and CMC teachers (4%). This supports the research saying that due to the decrease in social identification and deindividuation from CMC, one would be more likely to express his or her opinion (Mabrito, 1991; Mason, 1989; McGuire et al., 1987; Sproull & Kiesler, 1991).
Of particular interest was the difference in the nature of interaction in the courses. In FTF interactions, the students uttered 17% more Responding sentences than their teachers, as would be expected in a recitation environment. In the CMC courses, however, the teachers Responded 11% more than their students. Although on average the FTF teachers said more contiguous sentences than the FTF students (5.19 and 1.66, respectively), these were both much shorter than the average number of contiguous sentences submitted by CMC students (22.20) or their teachers (36.48).
These results may have been due, at least in part, to the restrictions of CMC. In a similar manner in which a straitjacket prevents one from eating his or her salad with the cake fork, the current state of CMC prevents behaviours which are characteristic of recitation, and encourages behaviours characteristic of discussion. I believe that many of the benefits of discussion flow directly from the asynchronous mode of communication. The delay gives the students extra time to respond to the teacher, as well as to initiate an idea, develop an argument, or defend a point of view. This "sustained interaction" (Shale & Garrison, 1990) gives students the time to participate as an equal partner in the educational process. Moreover, the fact that they cannot be interrupted also seems to play an important role, as does the lack of social cues.
Traditionally, research has shown that in FTF environments, female students are dominated by teachers and male students. My research supports this to some extent, but with some important exceptions. In my study, the female students uttered shorter sentences than the male students (an average of 9.39 and 10.36 words, respectively), while speaking for approximately the expected percentage of time (47% of the FTF student population uttered 43% of the sentences). However, the female students interrupted their teachers, male students, and other female students more than expected. Female students were expected to interrupt the teacher, male students, or other female students 44%, 47%, and 39% of the time respectively, when their actual results were 46%, 59%, and 40% respectively.
In the CMC environment, I expected to find fewer differences between the sexes, given the relative lack of social cues. This held true, in that both female and male students' sentences were the same length (13.79 and 13.89 words, respectively), but the female students contributed much more in total volume of utterances than expected (46%), given their 32% representation in class.
Taken together, one might conclude from these results that the female students were simply more verbose than their male counterparts. In the FTF environment, social presence and social cues may have kept female students from participating fully in class discussions, but they seem to have circumvented that to some extent through interruptions. In CMC, where they were free from such restricting influences and were able to contribute as they liked, they spoke proportionally more than expected. Further study comparing FTF and CMC interaction could be used to examine specifically whether female students experience a greater need for involvement and interaction than male students (Gabriel & Davey, 1995), and whether teachers should take this into account when interacting with their students.
To perform this research, I had to develop the tools I used to analyse the data. Rather than restrict myself to writing inflexible, stand-alone programs that would have to be re-written to add features for further analyses, I used AppleScript. In addition to providing a means to compare the environments of FTF and CMC classrooms, this research demonstrates that with the use of multi-program scripts, commonly available software can be used for sophisticated analyses of large amounts of data, and can continue to do so in the future.
This work was a relatively small case study. However, the automated aspects permit much greater amounts of data to be analysed, which might ultimately allow one to make useful generalisations about the nature of classroom interaction. Furthermore, creating the tools to automate analysis permits more data to be created and examined. If researchers made these tools and resulting data sets available on ftp sites or the World Wide Web, more research could be done, including replication and refinement of existing coding systems and tools.
When I first considered working with AppleScripts for this thesis, I was hesitant about their applicability in the future. What would be the use of writing programs that wouldn't work on computers available three years from now? I justified my actions for two reasons. First, even if the AppleScript language changed or Word and Excel ceased to exist, the programs would still be useful. Besides commenting extensively within each AppleScript, I have explained how each AppleScript works. Even if AppleScript becomes defunct, it would not be difficult to convert the scripts to a different language.
The second rationale is my belief that AppleScript or something similar to it will be the next mode of computer applications. Admittedly, anyone who's watched the computer industry in the last decade should hesitate before making any sort of predictions. However, there are some indicators that this may come to pass.
First, Apple Computer are actively supporting future versions of AppleScript. Although few applications support AppleScript presently, Apple appear to "have been impressed by grassroots support for AppleScript by its customers and developers" [44]. This is particularly welcome news for anyone who wishes to do collaborative work with these scripts in international or non-English settings. AppleScripts may be converted to other human languages simply by changing an option within the Script Editor. For example, one could -- if one so desired -- write an AppleScript that would replace the thirteenth word of the third paragraph of a document with "weasel" (see Figure 1):
Figure 25: Basic AppleScript
One could then change the language of the AppleScript to French for use with a foreign colleague (see Figure 26):
Figure 26: Basic AppleScript converted to French
Although AppleScript has not been upgraded since 1993, an update is promised which will add native Power Mac support and links to OpenDoc [45]. In particular, the links to OpenDoc give me reason to believe that these scripts have a bright future.
OpenDoc was included as an optional extension for the System 7.5.3 update for Macintosh in July 1996 (Apple Computer, 1996b; Crotty & Gruman, 1996). OpenDoc and OpenDoc components will form an integral part of future versions of the Macintosh operating system (Apple Computer, 1996a; Crotty & Gruman, 1996; DiNucci et al., 1994; Goodman, 1994; Oakley, 1996; Rizzo, 1996a). OpenDoc is an open, vendor-neutral, multiplatform standard for building modular software components that work together (Apple Computer, 1996a). OpenDoc makes applications available as small pieces of software, called parts, that one can choose to install and combine for their specific features under a common interface (Rizzo, 1996a; Streeter, 1995). These parts work together with the operating system, similar to the way in which plug-ins work in Adobe's Photoshop and Illustrator, Netscape's Navigator, and Quark's XTensions for XPress (Crotty & Gruman, 1996).
While AppleScript permits commands and data to be exchanged between applications, OpenDoc permits commands and data to be exchanged between parts. The difference is that with AppleScript one is required to use an entire application even if one wants to use a single feature of it. With OpenDoc, instead of working with bulky "bloatware" (Apple Computer, 1996a), applications which provide hundreds of functions one may never use, one can select only the desired features from a set of application components. For example, the scripts used in this work require the use of Word, even though it is only being used to open text files and recognise paragraphs, sentences, and words. If there were a OpenDoc part that could serve these elementary functions, the scripts wouldn't require Word, with its large RAM and hard-drive requirements.
OpenDoc parts work with each other the same way AppleScript can work with Macintosh computers across a network. Unlike AppleScript, however, OpenDoc will not be limited to the Macintosh computing platform. Currently, users of Windows and Windows NT operating systems do not have anything comparable to AppleScript available to them (Oakley, 1996; Weger, 1995). Soon, however, OpenDoc will be able to run on Macintosh, Windows, Windows NT, OS/2, and AIX operating systems (Apple Computer, 1996b; Orfali, Karkey & Edwards, 1994), as well as mainframe and midrange operating systems such as AS/400 OS, HP-UX, and MVS (Rizzo, 1995). Furthermore, OpenDoc parts can be scripted using several scripting languages, including AppleScript, LotusScript, and IBM's REXX, on different computing platforms (Rizzo, 1995). Thus, for example, an OpenDoc script could be written on any of these machines that would involve using parts on a Macintosh, a Windows machine, and a UNIX mainframe (Apple Computer, 1996c).
OpenDoc and the scripts I have devised ensure a strong future for this sort of analysis in three ways. First, by permitting inter-application scripting on other computing platforms, the use of the scripts is not limited to users of Macintosh computers. Admittedly, unless AppleScript is ported across platforms or the script is run from a Macintosh computer, my scripts would have to be rewritten in each platform's native scripting environment (Streeter, 1995). However, with a small amount of work, the same scripts could work on a wide variety of operating systems.
Secondly, since the parts required for the scripts are rather simple, they could be given away or sold very cheaply. Rather than having to purchase big, expensive, feature-laden software, one could purchase only the small, simple, necessary parts. In the case of my scripts, for example, one would require merely an extremely elementary text processor capable of understanding paragraph, sentence, word, and character and of copying them to another part. Such a simple text conveyor could be designed as a freeware or shareware addition to the scripts and made available at an ftp site or on the World Wide Web for anyone's use.
Third, since the use of parts rather than large applications decreases the hardware requirements to use the scripts, more people could use them. Besides permitting the use of much smaller applications, this would permit more flexible use of a computer's available RAM. When an OpenDoc document is open, OpenDoc will load the correct component into RAM as needed and unload it when done, based on what type of object is selected (Crotty & Gruman, 1996).
OpenDoc will permit a continuation of this sort of analysis to continue in the future, regardless of which software or computing platform is selected. The adoption of OpenDoc standards and parts would permit these AppleScripts to be used on almost any computer for free or nearly so. This would allow anyone, particularly those in cash-starved educational settings, to use these tools easily and inexpensively.
Although this research and these techniques are useful for examining discourse involving all classroom participants, further research is warranted. For example, the coding system and scripts could be modified to aid the analysis of threaded discourse (Grimaldi, 1995; Levin, Kim & Riel, 1990). Presently, the technology is limited to distinguishing teachers from students and males from females. Additional fields in the databases and minor modifications to the programs could permit one to track patterns of response by specific students to specific sentences. In addition, this would permit a way to determine if a few students (male or female) are monopolising the classroom discussions. Finally, by tracking each participant, one would be able to analyse each student's participation, correlate this with other measures of evaluation such as grades and interviews, and, ultimately, help to determine the effectiveness of specific modes of teaching.
| hillman@cantab.net | 1997 |