A Multidimensional Study of Synchronous and Supersynchronous Computer-Mediated Communication
Chapter 6. Discussion
Throughout this study, texts from the conversational writing genres have been contrasted with texts from spoken conversations, and written texts, to elucidate their qualitative, functional aspects as well as their lexico-grammatical patterns. Chapter 4, for instance, explored the interpersonal functions of modals in conversational writing and speech, and the similar lexical density in both types of communication, indicating their close relationship, and detailed the incidence in oral conversations of the salient features found in conversational writing. Ultimately, in chapter 5, the conversational writing genres and the genre of face-to-face conversations from SBC were positioned on Biber’s (1988) dimensions of linguistic variation, alongside the written and spoken genres studied by Biber, and the dimension scores of Collot’s (1991) genre of BBS conferencing were presented. The purpose of the present chapter is to bring together and discuss the results of the full investigation. The chapter starts out on a quantitative note and proceeds towards increasingly qualitative, multifaceted assessments.
Firstly (in section 6.2), the two hypotheses underlying the study (stated in section 1.2) are revisited quantitatively to begin to determine the relative degrees of orality in conversational writing and asynchronous CMC. This is first done by relating the positions of the conversational writing genres, and Collot’s (1991) genre of ACMC, to the oral conversational genres on Biber’s (1988) dimensions. Secondly (in section 6.3), the overall picture afforded by all dimensions in chapter 5 is closely examined to achieve multidimensional characterizations of the conversational writing genres and the genre of ACMC. The multidimensional characterizations provide the requisite input for determining the most prevalent “text types” (Biber 1989, 1995) in the CMC genres, which informs the indispensable, qualitative assessment of the results in relation to the hypotheses (Biber’s notion of text types is introduced in the same section). The chapter proceeds (in section 6.4) to revisit the four research questions posed at the beginning of the study (section 1.2), the first three of which have been addressed throughout, to identify and discuss the answers to these. Among other things, the section provides a summary of the findings from the comparisons of conversational writing to writing and speech. The fourth research question, as to whether conversational writing constitutes a modality ← 255 | 256 → of its own, is then addressed and answered. Finally, the working definition of conversational writing (offered in section 1.1) is revisited, in order to find out whether the definition needs to be elaborated on the basis of the findings in the full study. The last section (6.5) sums up the chapter.
This section revisits the hypotheses stated in section 1.2 from a quantitative viewpoint and discusses the relationships found in the present study between the CMC genres and the spoken conversational genres. The quantitative findings will then be complemented with gradually more qualitative assessments in section 6.3, before any final conclusions can be reached regarding the hypotheses.
In chapter 1, the synchronicity of communication was presumed to contribute decisively to the linguistic character of a genre. Genres with similar synchronicity of communication were predicted to display textual similarities, despite being communicated in different media (e.g. in a medium of CMC or through the medium of speech). Conversational writing was thus expected to display similarities with oral conversation, as both involve dialogs carried out in real time. It was also suggested that the CMC genres, representing asynchronous, synchronous and supersynchronous CMC, would display different degrees of orality. The degree of orality in conversational writing was defined as the degree of linguistic correspondence to oral conversations (face-to-face and telephone conversations). The two hypotheses stated in section 1.2 are the following:
- Synchronous conversational writing displays a higher degree of orality than asynchronous CMC
- Supersynchronous conversational writing displays a higher degree of orality than synchronous conversational writing
To test the two hypotheses quantitatively, the discussion here utilizes the positions of the conversational genres on Biber’s dimensions (see figures 5.1a through 5.6a in chapter 5) and the dimension scores of the genre of asynchronous CMC (presented in table 5.5). Five conversational genres are plotted on the dimensions in chapter 5, namely Biber’s (1988) two genres “face-to-face conversations LLC” and “telephone conversations,” and the three genres introduced in this study: “face-to-face conversations SBC” and the conversational writing genres “Internet relay chat” and “split-window ICQ chat.” The dimension scores of the genre of asynchronous CMC (given in table 5.5) are those of “BBS conferencing” (studied ← 256 | 257 → by Collot 1991 and originally labeled “ELC other”). The genre was not plotted on the dimensions but will nevertheless be considered here.110
As a genre’s degree of orality is crucially informed in this study by the genre’s proximity to oral conversations on Biber’s dimensions, it should be productive, over and above a visual inspection of the chapter 5 graphs, to measure the distances between the relevant genres. On Dimension 1, for instance, as seen in figure 5.1a, the two conversational writing genres both range in the vicinity of oral conversations, with texts displaying intense personal involvement between the interlocutors. The section accordingly explores the linguistic features that by their frequent occurrence contribute to the high scores of conversations on the dimension (first and second person pronouns, present tense verbs, direct WH-questions, etc.). From the visualization of the five conversational genres on Dimension 1 (figure 5.1a) it may be inferred that split-window ICQ chat approximates the three oral conversational genres slightly more than does Internet relay chat. By measuring the distance between the genres, in standard deviation units on the dimension, it is possible to begin to assess this quantitatively. Table 6.1 indicates the distance between each conversational writing genre and the oral conversational genres on all of the six dimensions. The table also presents the figures for BBS conferencing relative to the oral conversational genres. ← 257 | 258 →111
As seen in table 6.1, the predicted relationship between the conversational writing genres and oral conversations on Dimension 1 appears to hold true; split-window ICQ (with a distance “total” of 25.4 units; see the leftmost column) is indeed closer to the oral conversational genres than is IRC (with a distance “total” of 39.4 units). BBS conferencing, whose Dimension 1 score is nearly identical to IRC’s (see tables 5.5 and 5.1), naturally displays a total on Dimension 1 in table 6.1 similar to that of IRC, but on Dimensions 3 and 5 it deviates markedly from the conversational genres. Judging from the totals, split-window ICQ (SSCMC) is closest to oral conversations throughout the six dimensions, and except on Dimensions 2 and 6, BBS conferencing (ACMC) is most distant from oral conversations. IRC communication (SCMC) typically ranks in the interval between these two. The totals in the rightmost column are indicative of the pattern throughout; split-window ICQ is closest to oral conversations, IRC is intermediate, and BBS conferencing is the least oral genre.112 Put differently (see the bulleted hypotheses above), synchronous conversational writing (IRC) displays a higher degree of orality than asynchronous CMC (BBS conferencing) but is surpassed by supersynchronous conversational writing (split-window ICQ chat), ← 258 | 259 → which displays the highest degree of orality. Judging from this calculation, both hypotheses appear to be supported.
A conscientious, statistical analysis of the relationship between the relevant genres on the dimensions, however, needs to take into account not just the crude distances between genres, but statistically valid measurements. By calculating the t-values obtaining between the genres, the variation in the data is taken into account (cf. tables 5.1b through 5.6b) as well as the number of texts in each genre (cf. table 3.1 and Appendix I). In other words, the t-value indicates the distance between the genres’ mean dimension scores after accounting for these factors. T-values were obtained by using the two equations below, in which x is a conversational writing genre, y is an oral conversational genre, and n is the number of texts in the genre. The equations take into account for each genre its mean dimension score (mean) as well as its standard error of the mean (SEM), based on the standard deviation of texts. Table 6.2 presents the relevant statistics for the genres under consideration (in normal font), as well as the results of the calculation (in bold). Unfortunately, the genre of ACMC must be left out of the account here as the requisite data is unavailable for BBS conferencing (Collot 1991) and, as a result, the first hypothesis cannot be statistically tested.
The figures given in bold in table 6.2, nevertheless, are viable for addressing the second hypothesis, by which conversational writing in SSCMC should display a higher degree of orality than in SCMC. In principle, the lower the t-value (in absolute value, i.e. ignoring incidental minus signs in the comparison), the less likely is a significant difference between the genres compared. In table 6.2, the t-values for the relationship between split-window ICQ chat (SSCMC) and the spoken conversational genres are indeed generally lower than those obtaining between Internet relay chat (SCMC) and the spoken conversational genres. On Dimension 1, for instance, the t-values for split-window ICQ chat are 0.6, 2.9 and 2.3 compared to the oral conversational genres, respectively, whereas those for Internet relay chat range around 4. Similarly, on Dimensions 2 and 6, the t-values for the relationship between split-window ICQ and the oral conversational genres are also all lower than those between Internet relay chat and the latter. On Dimensions 4 and 5, the same general impression comes through, even though IRC is closer than ICQ to face-to-face SBC on Dimension 4, and to telephone conversations on Dimension 5. For the conclusive interpretation of the t-values, however, table 6.2 needs to be paired with the statistical significance of the results, i.e. with p-values (p). Table 6.3 presents the p-values for the relationship between conversational writing and conversational speech (i.e. for the same genres). ← 260 | 261 →113
The p-values in table 6.3 specify for the corresponding t-values given in bold in table 6.2 the probability of obtaining the distances (measured in t-values) by chance if there were no difference between the dimension scores of the genres. Table 6.3 shows that a few more of the values for IRC than for ICQ are statistically significant, i.e. p<.05 (in bold). The p-values for IRC on Dimensions 6 all indicate a significant difference from the oral genres, as do two of IRC’s p-values on Dimension 2 and one on Dimension 1; i.e. they indicate that the dimension scores of IRC are significantly different from the oral conversational genres in question. As it is the proximity of genres that is at issue here, however, high p-values are also informative for the interpretation of the results; the preponderance of high p-values (in normal script) for the relationship between split-window ICQ and oral conversations shows that split-window ICQ chat is not significantly distinct from oral conversations; on Dimensions 1 through 5 no difference obtains; the only indications of a discrepancy are found on Dimension 6. The wealth of high p-values for IRC, moreover, indicates roughly the same relationship; IRC is generally not distinct from the oral conversational genres, either. On Dimensions 3 and 5, all the conversational genres in effect coincide (“p=1.000” meaning that no measurable difference was found) and on Dimension 4, no statistical difference obtains between the written and the oral conversations. The statistical tests thus establish that on most dimensions, supersynchronous CMC, as represented by split-window ICQ chat, is lexico-grammatically more similar to spoken conversations than is synchronous CMC, as represented by Internet relay chat. On ← 261 | 262 → the first five dimensions, split-window ICQ is an inherently “oral” genre, whereas IRC is marginally less “oral.”
On Dimension 6, labeled “On-line Informational Elaboration,” both conversational writing genres deviate from face-to-face conversations (see the p-values in table 6.3). This finding mirrors the discussion of Dimension 6, in section 5.2.6, in which the sparsity in conversational writing of dependent clauses (THAT complement clauses on verbs and adjectives and THAT relative clauses on object position) and demonstratives (that, this, these, those preceding nominals) were found to yield noticeably low dimension scores for the conversational writing genres. Chatters’ turns contain extremely few elaborations added on “live” by way of complement clauses, as in for instance spontaneous speeches, but also contain fewer complement clauses and demonstratives preceding nominals than most of the written genres. On the other hand, while IRC largely lacks markers of online informational elaboration, split-window ICQ and telephone conversations remain similar even on this dimension (as no measurable difference was found between the latter two). In the assessment of the orality of conversational writing here, however, we must be cautious not to attach too much importance to Dimension 6. As noted in section 5.2.6, the dimension has few linguistic features with important loadings (Biber 1988) and most studies, including Biber’s more recent ones, have consequently left it out of account. The discourse observations with respect to Dimension 6 in section 5.2.6 are thus only tentative, and in the further discussion, section 6.3 below, the sixth dimension is eventually phased out.
Before proceeding, a few remarks to sum up the present section are in order. In the quantitative assessment of the hypotheses here, two main findings have emerged. Firstly, the dimension scores of the ACMC genre generally position the genre at a greater distance than the conversational writing genres from the oral conversational genres on Biber’s (1988) dimensions. This finding lends support to the first hypothesis, by which asynchronous CMC should be less “oral” than synchronous CMC (or conversational writing at large). Secondly, in the statistical tests of the positions of the conversational writing genres on the dimensions, the supersynchronous genre (split-window ICQ) was found to be substantially correspondent to oral conversations, whereas the synchronous genre (IRC) appeared to be marginally less “oral.” In other words, the statistical tests offer a measure of evidence supporting the second hypothesis, that the supersynchronous genre displays a higher degree of orality than the synchronous genre. In the next section, with gradually more qualitative assessments, we will find out whether these initial propositions hold. The section will show that, by looking beyond genre boundaries ← 262 | 263 → to find similarities and differences across texts, a complementary approach to discourse variation is feasible, one which provides illuminating results.
What is the linguistic nature of conversational writing? The question was raised as the first of four research questions in section 1.2, laying out the aim and scope of the study. Chapters 4 and 5 have provided ample textual examples to illustrate typical messages exchanged over the computer networks. Discussions of these recurrently recognized that the texts from both conversational writing genres resemble oral conversational texts to a great degree, syntactically and lexico-grammatically, and deviate notably from traditional written genres in many respects. The most salient features in conversational writing (frequent first and second person pronouns, direct WH-questions, analytic negation, demonstrative and indefinite pronouns, present tense verbs, predicative adjectives and contractions, and infrequent prepositional phrases), explored in chapter 4, were found to be decisive contributors to the oral character of the textual chats. Most of the features were then revisited in chapter 5 in the discussion of Dimension 1, distinguishing involved texts from informational texts. On other dimensions, conversational writing was set apart from most written genres by, for instance, a marked paucity of certain linguistic features, as on Dimension 5 on which the non-abstract chats are diametrically opposed to the genres of stereotypical abstract writing (official documents and academic prose). At this point, the multitude of findings can be interrelated and the discussion of the hypotheses (stated in section 1.2 and repeated in section 6.2) brought forward. First, the discussion here reflects on the overall picture afforded by all dimensions (see also chapter 5 for dimension graphs and concomitant in-depth descriptions). Next, the dimension score patterns across the first five dimensions will be traced to identify the “text types” (Biber 1989, 1995) of the conversational writing texts and the ACMC genre, which, in turn, will enable the conclusive assessment of the hypotheses.
The following slightly simplistic characterization of conversational writing can be made on the basis of the genre means on Biber’s dimensions explored in chapter 5 (parentheses indicating Dimension numbers). Conversational writing is involved (1), non-narrative (2), situation-dependent (3), non-argumentative (4), non-abstract (5) discourse, containing very little real-time informational elaboration (6). Oral conversations are also typically involved (1), situation-dependent (3), non-argumentative (4) and non-abstract (5), but are slightly more narrative (2) and contain more real-time informational elaboration (6) than ← 263 | 264 → chats. The multidimensional characterization of the ACMC genre will be traced shortly; the discussion here first zooms in on the conversational genres.
In section 6.2, the results from statistical tests of the relations between conversational writing and conversational speech were presented. It was seen there that, except on Dimension 6, split-window ICQ is not distinct from oral conversations, and that IRC is only marginally less “oral” than the former. The discrepancies between IRC and the spoken conversational genres all appeared on Dimensions 1, 2 and 6 (see p-values in table 6.3). The findings on Dimension 6 were discussed in section 6.2 with regard to both conversational writing genres; the remaining discrepancies (i.e. those between IRC and oral conversations) are touched upon here, although as will be seen, the similarities across the conversational genres outweigh the differences.
On Dimension 1, the five conversational genres contrasted in this study (IRC, face-to-face and telephone conversations from LLC, face-to-face conversations from SBC and split-window ICQ chat) all reside, in the order given, at the involved end of the scale. The Dimension 1 graph (figure 5.1a) visualizes their distinctive positions, as the dimension separates the texts with involved production from those with informational production. Prima facie, split-window ICQ chat appears to be more “involved” than the oral conversational genres, i.e. to display a degree of orality beyond all of theirs (if the positive end of the dimension is taken to be the oral end). On closer inspection, however, split-window ICQ is not significantly different from either face-to-face or telephone conversations (as p>.05 in table 6.3). Instead, split-window ICQ is greatly akin to these, which also rules out the possibility theorized in chapter 1 of supersynchronous chats exceeding oral conversations in orality, a relationship that would have called for a redefinition of orality here. In other words, the definition of orality employed in this study (the similarity to oral conversations) still holds.
On Dimension 1, split-window ICQ, like the oral conversational genres, displays abundant markers of involvement (private verbs, THAT deletion, contractions, etc.; see section 5.2.1), reflecting real-time production circumstances. In fact, both conversational writing genres are lexico-grammatically akin to the oral conversations on this dimension, even though the IRC discourse is more moderately involved than the split-window ICQ conversations and also statistically different from these (p<.05 in table 5.3), as well as from face-to-face conversations SBC (p<.05 in table 6.3). On Dimension 2, “Narrative versus Non-Narrative Concerns,” IRC displays non-narrative discourse similar to that of telephone conversations (with few past tense verbs, few third person pronouns, etc.; see section 5.2.2) and has fewer narrative features than the face-to-face conversational ← 264 | 265 → genres (as <p.05 in table 6.3). The latter range in intermediate position on the dimension, unmarked for narrativity, and are thus not particularly concerned with narration, either. (As Dimension 2 is not associated with a literate-oral polarity, the position of IRC on the dimension is unproblematic for the definition of orality; only on Dimensions 1, 3 and 5 can a genre “exceed” oral conversations, i.e. surpass them at the oral end, and this not the case, statistically, for any conversational writing genre.)
Apart from the few discrepancies mentioned (pertaining to IRC), the overall examination of Dimensions 1 through 5 in chapter 5 yields solid results for both conversational writing genres as regards their similarity to oral conversations, not least on Dimensions 1, 3 and 5, the three dimensions seen to “identify sharp distinctions between ‘oral’ and ‘literate’ registers” (Biber 2008: 843, also noted in e.g. Biber & Finegan 2001). On Dimension 3, “Explicit/Elaborated versus Situation-Dependent Reference,” neither of the chat genres is different from oral conversations (cf. table 6.3); conversational writing, like the oral conversational genres, displays discourse with frequent markers of situation-dependent reference (e.g. time adverbials) and a sparsity of elaborating devices (such as WH relative clauses); see section 5.2.3. On Dimension 4, the conversational genres are all generally unmarked for overt expression of persuasion/argumentation and, even though certain split-window ICQ texts contain more opinionated discourse than most IRC texts, no statistical difference obtains between the conversational genres on the dimension (cf. table 6.3). On Dimension 5 (as seen in figure 5.5a), the conversational writing genres both practically coincide with the oral conversational genres on the non-abstract/non-impersonal end of the dimension. Neither split-window ICQ nor IRC is statistically different from the oral conversations; rather, all conversational genres display a decisive paucity of markers of abstract information (e.g. conjuncts and agentless passives).
As touched upon in the previous section, Biber and other linguists carrying out post-Biber (1988) MD analyses have, over the years, paid diminishing attention to Dimension 6, identified in Biber (1988). As early as 1989, Biber ignores the dimension, asserting that “five major dimensions have been identified in English” (1989: 7). Biber’s (2008) account of multidimensional approaches mentions the sixth dimension, but also elaborates only on the first five, seeing that the sixth dimension “has few salient linguistic features” (2008: 836). Moreover, Biber (2008) notes that Dimensions 2 and 4 “have no systematic relationship to speech and writing” (2008: 843). Even though all dimensions must be considered for the full picture (as described in section 1.2), Biber (2008) argues that Dimensions 1, 3 and 5 have been seen to most clearly set the oral genres, especially ← 265 | 266 → conversations (“stereotypical speech” 2008: 843), apart from the written genres. In the present study, however, conversational writing is observed to intermingle with oral conversations on most dimensions, and most notably on Dimensions 1, 3 and 5. Conversational writing and conversational speech have been found to be closely related functionally, irrespective of their genre classifications, as ways for writers/speakers and readers/listeners to interact personally in the immediate present (synchronously and/or supersynchronously), generally with the purpose of furthering interpersonal relationships. The multitude of textual properties found to be common to all the conversational genres makes their affiliations with the written or spoken medium rather irrelevant; instead, it is the immediacy of the situation, the synchronicity, the presence of a responsive audience, and the attendant social practices that determine the nature of the discourse (as found in section 4.2). Texts are not confined to genre boundaries; rather, texts may display similar linguistic characteristics across genres.
In his 1989 and 1995 studies, Biber offers an apt, complementary view of textual variation, bringing the 1988 study forward from defining the genres in situational/functional terms to defining the “text types” with maximally similar linguistic properties (Biber 1989, 1995). While genres are determined on the basis of external criteria such as the purpose of the author/speaker and the production circumstances, text types are groupings of texts that are similar in their linguistic form (with respect to dimension characteristics), irrespective of their genre classifications. Text types thus cut across genre boundaries, offering variationists a complementary way “to dissect the textual space of a language” (Biber 1995: 320). A single text type might include texts from several different genres. The text type “scientific exposition” for instance, marked as very informational (integrated) on Dimension 1, non-narrative on Dimension 2, elaborated on Dimension 3, etc., includes texts from academic prose, official documents, and a few more genres, all texts sharing the same linguistic characteristics. Conversely, texts from a single genre can be distributed across several text types; academic prose, for instance, is represented across four text types (3, 4, 6 and 8). Table 6.4 summarizes the eight text types identified among the Biber (1988) texts, types detailed in Biber (1989, 1995), indicating for each text type genres in which it occurs and its multidimensional characterization.
The text types in the English language (table 6.4) were identified empirically in Biber (1989) by way of a multivariate statistical procedure called cluster analysis. As input to the analysis, Biber used the dimension scores of the Biber (1988) texts. The analysis grouped texts with maximally similar dimension scores on Dimensions 1 through 5 into clusters, assigning every text to some cluster. Biber found ← 266 | 267 → each cluster to represent one text type, and assigned each text type an interpretive label; thus, cluster 1 represents text type 1, which he labeled “intimate interpersonal interaction”; cluster 2 represents text type 2, which he labeled “informational interaction,” etc. For each text type, Biber traced the typical dimension characteristics of texts, which resulted in the multidimensional characterizations given in table 6.4. The multidimensional characterizations represent the most central texts of the cluster (those closest to the centroid of the cluster).114 In what follows, we will identify the text type(s) of conversational writing by relating its texts
to the text types (i.e. clusters) identified in Biber (1989, 1995). The consideration of text types here is intended to elucidate not just the character of conversational writing, but also, eventually, that of asynchronous BBS conferencing (Collot 1991), all in order to inform the assessment of the hypotheses.
Biber (1989: 15) explains the cluster patterns by first illustrating the clusters formed by combining the Dimension 1 and 3 scores of individual texts into a graph, in which Dimension 1 constitutes the x-axis and Dimension 3 the y-axis. The dimension scores for the individual text on both dimensions are plotted in their point of intersection in the graph, each numbered with the text’s text type. To begin to identify the text type(s) of the conversational writing texts in the present study, their dimension scores were plotted onto this graph, which assigned most of the split-window ICQ chats positions in the text type 1 cluster and most of the IRC texts plots in the text type 2 cluster of texts. The dimension scores of the face-to-face conversations from SBC were also plotted onto the graph, expectedly yielding a position for most of the texts in the text type 1 ← 269 | 270 → cluster.115 Finally, the mean dimension 1 and 3 scores of Collot’s (1991) genre of ACMC, named “BBS conferencing” in the present study (table 5.5), were plotted onto Biber’s (1989: 15) graph of cluster distributions, giving the genre a position amidst the text type 2 texts, albeit more distant from the cluster centroid than most of the IRC texts.116
Throughout this study, conversational writing has, implicitly or explicitly, been suspected to be maximally similar to the texts of face-to-face and telephone conversations, which all appear in text types 1 and 2 (except for one outlier telephone conversation in text type 8). By first relating the texts on Dimensions 1 and 3, and next, analyzing the dimension score distribution of the conversational writing texts with respect to Dimension 2 (ICQ texts relatively unmarked, IRC texts non-narrative), Dimension 4 (most texts relatively unmarked for persuasion/argumentation) and Dimension 5 (most texts non-abstract), it is possible to conclude that most of the split-window ICQ texts indeed adhere to text type 1 and that most of the Internet relay chats fall under text type 2 (as no other text type suits their multidimensional characterization better; see table 6.4). Furthermore, most of the face-to-face conversations from SBC follow the same text type 1 pattern as split-window ICQ across the five dimensions.117
That split-window ICQ chatting, like most of the face-to-face conversations from SBC, represents “intimate interpersonal interaction” (text type 1) has been exemplified recurrently throughout the present study (prior to this identification ← 270 | 271 → of the text type). The split-window ICQ texts display, for instance, abundant private verbs, contractions, first and second person pronouns and present tense verbs, and a sparsity of nouns, prepositional phrases, etc., associated with extremely high scores on Dimension 1; frequent time and place adverbials, and few WH relative clauses, contributing to low scores on Dimension 3; and particularly infrequent conjuncts, agentless passives, etc., yielding low scores on Dimension 5. Labeling Internet relay chatting as “informational interaction” (text type 2), however, is slightly counter-intuitive.
The IRC texts have been more difficult than the split-window ICQ chats to characterize throughout. A few of the textual examples (e.g. examples 2, in section 3.2, and 4, in section 4.2) show interlocutors engaged in fairly involved, interpersonal interaction, but such passages are usually brief in the flow of IRC communication (i.e. extend over a restricted number of turns). Owing to the multitude of participants, conversational turns are more often interrupted. Typically, newly arrived participants’ greetings and questions break up adjacency pairs (as in example 21 in section 4.4), fragmenting the discourse. Herring (2013b: 252) notes that such disrupted adjacency makes multiparty chat systems “noisy communication environments.” As mentioned in connection with example (3) in section 5.2.1, IRC conversations (i.e. those involving two or more interlocutors in coherent exchanges) are frequently interspersed with turns consisting of one mere keystroke (?, ;, 2), compliments, phatic expressions and other attention-attracting tropes (hey, gret thanks cheeky, grrrr u), which eventually cause coherent conversations to wane. In this “noisy” environment, chatters are less inclined than those in split-window ICQ to produce extremely involved discourse, judging by the slightly fewer “involved” features on Dimension 1 (e.g. present tense verbs, analytic negation, first person pronouns) that surface in the IRC texts as compared to the split-window ICQ chats. Nevertheless, the IRC texts assume a “very involved” position on Dimension 1, which renders the text-type label “informational interaction” for IRC communication a misnomer.
In a study of web chat, that is, chat rooms similar to the channels of IRC, Sveningsson (2001: 58) observes that the communication resembles “multiparty telephone conversations (telephone chat lines), of the kind that used to be called ‘Heta Linjen’ (the Hot Line) in Sweden.” Sveningsson explains:
These multiparty telephone conversations should not be confused with what is referred to as hotlines today, where the main purpose seems to be phone sex. The former type [i.e. the Hot Line] consisted in telephone numbers that had no subscriber, to which people could call free of charge. The knowledge of those numbers was spread through personal communication between young people, and can indeed be seen as one of many strategies ← 271 | 272 → to avoid the governance of adults and organizations, since these media provided a free and un-moderated space, in which adults had little insight. (Sveningsson 2001: 58)
Hot Line communication, popular among young Swedes in the 1980s, involved multiparty telephone conversation between several individuals, all with the intention of finding new friends. The discourse greatly resembled the web chat, and IRC, interaction of today in that it typically contained “a jumble of voices shouting ‘Hello?’, ‘Hello?’, ‘Hello?’, ‘Who are you?’, ‘What’s your name?’, ‘Hello?’, and so on” (2001: 58). Sveningsson also notes that “callers often exchanged personal telephone numbers at an early stage, to be able to call each other up and have a more coherent conversation” (2001: 58), which in IRC is accomplished by initiating the direct client-to-client protocol (see section 4.5). Speaking from personal experience, the present author agrees with Sveningsson’s description of the telephone chat lines and their resemblance to online chat rooms/channels. The discourse in public Internet relay chat channels is indeed as disjointed and ephemeral as was that on the telephone chat lines, even though both also contain passages of coherent conversation.
The participants competing for attention in IRC all value extreme brevity and quick responses. The profusion of greetings and attention-attracting tropes, however, brings about “a low signal to noise ratio” (Herring 1996b: 105) in the discourse, meaning that the verbal “flurry” of the multiparty chat does not transmit a great deal of useful information (Mann & Stewart 2000: 184). In the words of Crystal (2001), it rather resembles “a cocktail party in which everyone is talking at once – except that it is worse, because every guest can ‘hear’ every conversation equally, and every guest needs to keep talking in order to prove to others that they are still involved in the interchange” (2001: 159). Put simply, IRC discourse is disjointed multiparty conversational writing. The characterization of IRC as “informational interaction” (text type 2) here is thus misrepresentative, but apparently inevitable given the multidimensional distribution of the texts’ linguistic features. Certainly, the multidimensional character of text type 2 embraces the IRC texts more closely than does any other text type, but upon including Internet relay chats among the texts in text type 2, altering the text type label would perhaps be justified, which necessarily requires further research (outside the scope of the present study).118
Asynchronous CMC discourse for social interaction, such as that in Collot’s (1991) genre of BBS conferencing (“ELC other”; see inter alia sections 4.1, 4.2 ← 272 | 273 → and 5.1), on the other hand, appears to more justly conform to the original text type 2 label, although more for its resemblance to the non-conversational texts in text type 2 than any resemblance to Internet relay chat or oral conversations. The dimension scores of the genre indicate that its discourse, like that of IRC, is moderately involved on Dimension 1 and, like ICQ, non-narrative on Dimension 2, but unlike conversational writing, the asynchronous discourse is unmarked for situation-dependent reference on Dimension 3, persuasive/argumentative on Dimension 4 and remarkably abstract on Dimension 5. Despite these differences on Dimensions 3 through 5, the overall distance between the dimension scores of the BBS conferencing genre and the cluster centroids of Biber’s texts objectively indicate that the genre adheres to the text type 2 cluster (specified in Biber 1995 and in Appendix IX here), albeit more peripherally so than most of the IRC texts (the ACMC genre being closer to text types 3 through 8 than are most of the conversational writing texts). At the same time, the dimension scores of the asynchronous genre are more distant from the text type 1 cluster than are those of most IRC texts.
While the correspondents in BBS conferencing may formulate long, thought-out verbalizations, the IRC and ICQ interlocutors, like oral conversationalists, are subject to time constraints and produce brief, impromptu turns analogous to those in unplanned, spoken interaction. Biber (1995: 328) points out that the principal difference between text type 2 and text type 1 texts “relates to the primary purpose of the interaction: to convey information in text type 2 and to maintain the interpersonal relationship in text type 1.” Collot (1991) notes that “one of the primary purposes for participating in a BBS is to seek and impart information” (1991: 86) and that the ACMC texts are easily compared to interviews, as well as to letters: “In much the same way as personal and professional correspondents, participants in [BBS conferencing] share neither the same physical nor the same temporal context” (1991: 89). The latter fact, to be sure, sets the ACMC texts apart from the synchronous and supersynchronous CMC texts; Collot (1991) observes that participants’ separation in time and space “may be at the root of the resemblance between the [BBS conferencing corpus] on the one hand, and personal and professional letters on the other” (1991: 90). As the ACMC texts are produced for asynchronous delivery, the authors rely less on the immediate situation and instead, like letter-writers, produce for instance more WH relatives in subject position and more nominalizations (on Dimension 3) than in conversational writing. They usually have the required time to study incoming messages and to carefully prepare their argumentation and thus include e.g. more conditional adverbial subordinators: if, unless (on Dimension ← 273 | 274 → 4). The most striking difference between the ACMC genre and the conversational writing genres here, however, is the far more abstract discourse produced in the asycnhronous genre, giving it a score on Dimension 5 on a par with official documents. Abstract/impersonal content is composed by way of e.g. conjuncts (furthermore, moreover, nevertheless, etc.), adverbial subordinators (such as since, while and whereas) and BY passives (Biber 1988), features largely lacking in conversational writing as well as in conversational speech. On the whole, it is on two of the three dimensions separating literate and oral genres, Dimensions 3 and 5, that the asynchronous genre deviates most from the conversational genres (cf. table 6.1). Korsgaard Sorensen (1993) finds CMC texts to reflect the time interval between exchanges; when exchanges “occur at longer intervals in time,” they display features of “prototypical written interaction” (1993: 406). That the production of asynchronous computer-mediated texts admits of thought-out, carefully composed verbalizations is evident not just in the more elaborate references and the frequent features of abstract discourse, but also in the word length and TTR of the ACMC texts, both more similar to those of writing than of speech (as seen in section 4.3) and in the high lexical density of ACMC texts (although the latter is not one of Biber’s features). The genre of asynchronous CMC, in sum, is deemed to consist of texts that, unlike the conversational writing texts, are peripheral to oral conversations.
In conclusion, rating from their convergence to the multidimensional characterizations for the text types in Biber (1989, 1995), the conversational writing texts and the asynchronous genre studied in this analysis divide up. The split-window ICQ chat texts (SSCMC) most closely resemble oral conversations of text type 1, “intimate interpersonal interaction,” and the Internet relay chat texts (SCMC), despite their defying easy classification, adhere to text type 2. However, judging from the incidence of oral conversations in both text types, and the qualitative, functional assessments here, IRC communication is no less “oral” than split-window ICQ chat, only less extremely involved (intimate). BBS conferencing (ACMC), on the other hand, more peripherally than the IRC texts in multidimensional character, but more justly in terms of its function, conforms to the text type 2 label “informational interaction.” The asynchronous genre shares more functional and lexico-grammatical characteristics with the non-conversational genres in text type 2 than does Internet relay chat and may, consequently, be regarded as more distant from face-to-face and telephone conversations than is conversational writing, i.e. as not an “oral” genre in the terms of this study. The discussion in this section has thus complemented the quantitative assessment in the previous section and necessitated a refinement of the proposition ← 274 | 275 → enounced there regarding the hypotheses. The findings, in combination, support the first hypothesis, i.e. that conversational writing displays a higher degree of orality than the genre of asynchronous CMC, but do not provide evidence for confirming the second hypothesis, that split-window ICQ chatting should be more “oral” than IRC communication. Rather, the conversational writing texts, in both SSCMC and SCMC, are all closely related to oral conversations; the split-window ICQ chats are merely more intimately interpersonal in character than the Internet relay chats.
In section 1.1 of the present study, a working definition of conversational writing was formulated, by which conversational writing is written communication 1) for social interaction 2) which requires the simultaneous presence (physical or virtual) of producer and recipient, 3) in which interlocutors expect immediate feedback (i.e. within seconds) and 4) during which the discourse may be reconfigured by the participants while under construction (e.g. as interlocutors are able to influence each other’s line of thought). None of the findings that have emerged in the study have occasioned an alteration of these premises (other than that the conversational writing discourse investigated here is exclusively computer-mediated and not conveyed via note-passing, also regarded as conversational writing in section 1.1). In the present section, a selective overview of the results will be given with the primary aim of discussing and summarizing some of the answers provided to the research questions posed in section 1.2, and the secondary aim of finding out whether the definition of conversational writing needs to be elaborated on account of the discussion.
Four research questions were posed at the outset of this study (the first three of which have been addressed and the fourth will be addressed shortly):
- What is the linguistic nature of conversational writing and the genres studied here, IRC and split-window ICQ chat?
- How does conversational writing carried out in SCMC and SSCMC, respectively, relate to writing and speech?
- How do the genres of SCMC, SSCMC and ACMC relate to oral conversations on Biber’s (1988) dimensions?
- Does conversational writing carried out in SCMC and SSCMC constitute a modality of its own?
It stands to reason that any attempt to summarize the answers to the research questions in one brief section is bound to be slightly simplistic and might fail to ← 275 | 276 → reflect the complexity of the results (thus, for the comprehensive answers readers are referred to chapters 4 and 5, as well as to the entirety of the present chapter). Nevertheless, the task is taken on here; firstly with regard to the first research question, as to what conclusions about the nature of the conversational writing genres may be drawn on the basis of the full investigation. (The nature of conversational writing is first discussed in functional linguistic terms, before the more specific, text-linguistic findings are summarized in bullet points.)
Conversational writing is not a homogeneous entity; rather, just like oral conversations, it occurs in endless constellations of contexts, for a variety of purposes, on infinite numbers of topics, between any two or more people. A major difference between conversational writing and face-to-face conversations, however, is the limited shared context in the former, a context even more limited than in telephone conversations. While face-to-face interlocutors share the physical surrounding (and with all senses available to them take in the visible objects, background sounds, scents, temperature, etc., as well as the non-verbal cues signaled by the conversational partner) and telephone conversationalists share the audible surroundings (and are able to perceive clues as to each other’s sentiments and e.g. turn-yielding signals, such as changes in vocal pitch), conversational writers (in the CMC media studied here) are confined to the interface shared on their screens, mostly to the text conveyed. This limited semiotic field (Halliday 1985a, 2004, Martin 2001a; cf. section 2.4) naturally impinges on interlocutors’ language, but not as much as one might first imagine. In this study, conversational writing texts have been found to be remarkably similar to transcribed oral conversations (with prosodic annotations removed in the latter), while at the same time inherently different from most other written genres. The split-window ICQ chats, however, have been found to be of a more close, interpersonal character than the Internet relay chat texts (cf. section 6.3).
The social relationships formed in the conversational writing interface (i.e. the semiotic tenor of the interaction; cf. section 2.4) largely depend on what the particular chat client allows (in terms of number of participants). In IRC, multiple individuals (dozens at once in the IRC corpus studied) in remotely separated localities convene in the virtual rooms. Some chatters frequent the same channels, keeping the same nickname, which means that close relationships may form between regulars (Mar 2000), although these are more often maintained in the direct client-to-client protocol than in the public channel. Most IRC chatters in public channels, however, are not previous acquaintances and rarely meet in real life. As a result, they have little at stake if they are not appreciated or accepted in the chats, especially in the public channels, as they can simply leave the channel ← 276 | 277 → without embarrassment (Mar 2000). At the same time, chatters in ICQ more often use the medium to complement or extend real-life interactions and discuss matters from their occasionally shared real-life context, expecting to be held responsible for views expressed. The results presented in section 5.2.4, for instance, show that IRC chatters in public channels rarely exchange views in animated ways, whereas the semiotic mode in ICQ occasionally brings about extended supportive or challenging argumentation.
The ICQ chatters in the corpus studied, unlike most IRC chatters, have met, and regularly meet (met), face-to-face. At the time of the recording, the ICQ chat client was designed to handle only a few (two to three) participants chatting at a time, and was mostly used for communication between those with a previous real-life relationship (via an editable, personal list of friends). The semiotic mode of split-window ICQ chat, at least for the participants in the present study, is therefore different from that in IRC, i.e. the language exchanged plays different roles for participants. In public IRC (or in web chat), the interface is used to initiate or maintain mostly fleeting relationships, and in ICQ (or, in IM) it is used to maintain or further existing relationships. These fundamental properties of the two conversational writing genres studied, their different semiotic tenor and mode, contributing to a typically lower signal-to-noise ratio in multiparty IRC (section 6.3), more than the genres’ respective synchronicity of communication (synchronous vs. supersynchronous) have been found to have a bearing on the results in this study. Whereas the public IRC chats are disjointed, superficial, rapid-fire exchanges between multiple (i.e. >2) participants, the ICQ discourse, just like most of the spoken conversational material, consists of intimate, personal, conversations, ranging from adversarial to affective, between two or three previously acquainted individuals. Differences in the lexico-grammatical make-up of the IRC and ICQ texts are thus due more to the situational, client-imposed, cultural and semiotic factors associated with the respective genres than to the sheer difference in synchronicity (synchronous vs. supersynchronous communication) between the two. More precisely, although a potential supersynchronicity effect is vaguely discernible in the more common use of the inserts “response forms” and “hesitators” in the split-window ICQ chats (as seen in section 4.6), no all-round supersynchronicity effect is evidenced in the material that would liken the supersynchronous chats, more than the synchronous, to spoken conversations.119 ← 277 | 278 → Rather, the marginally “higher” degree of orality in split-window ICQ found in section 6.2 (in the quantitative assessment) and the adherence of split-window ICQ chat to text type 1 (“intimate interpersonal interaction”; see section 6.3) are more likely due to the similarity in semiotic tenor and mode of the supersynchronous chats and the oral conversations. That is not to say that a supersynchronicity effect is ruled out. Observing and establishing the existence of such an effect, however, would require not only access to high-quality video recordings of the supersynchronous chats for an analysis of overlapping sequences (which was not the case here, as mentioned in section 4.6), but also that the synchronous material (i.e. the control group material) be acquired from channels, or chat clients, that allow only two or three participants, participants who also preferably are previous acquaintances, to make for maximally comparable situational settings. In hindsight, such a research design might have been preferable.
That said, there still remains a substantial synchronicity effect evident in conversational writing, which likens conversational writing to conversational speech, and distinguishes both from the medium of writing, as well as from ACMC. That conversational writing resembles oral conversations to a great degree is due to the related synchronicities of the two; conversational writing and conversational speech are both carried out in real time, which gives rise to a number of linguistic features typical of immediate, interpersonal interaction, while at the same time restricting the number of linguistic traits typically associated with edited asynchronous writing, or the elaborated writing produced for one-way communication (cf. table 1.1). The following lists survey the relationships found in the present study between conversational writing (SCMC as represented by IRC and SSCMC as represented by split-window ICQ chat), writing and speech, at the level of medium (cf. figure 1.2 and the results in chapters 4 and 5 and in the present chapter), i.e. they sum up some of the answers to the first and second research questions.
Compared to writing, conversational writing (SCMC and SSCMC) has
- lower lexical density
- shorter clause length
- shorter word length
- more explicitly involved, interpersonal content, as reflected in e.g. more frequent use of first and second person pronouns, present tense verbs, direct WH-questions and contractions ← 278 | 279 →
- fewer prepositional phrases, reflecting very limited clausal elaboration
- more situation-dependent reference, as reflected in e.g. frequent use of time adverbials and infrequent use of WH-relative constructions
- inherently non-abstract, non-impersonal content (unlike writing), with e.g. few conjuncts and passive constructions
- less informational elaboration by way of e.g. THAT verb complements and THAT relatives in object position
- more analytic than synthetic negation (whereas the opposite obtains in writing)
- paralinguistic features encoded in the script, e.g. graphic imagery, repeated question marks, uppercase words and repeated letters for acoustic effects etc. (rare or absent in writing)
- more exophoric reference to the extra-linguistic context, for instance to the shared virtual room and to web content and files shared
- emotives (unlike writing), signaling the interlocutor’s sentiment or the sentiment in which a message is to be received (including ironic or tongue-in-cheek intention)
In addition to the traits above, Internet relay chat (SCMC) in comparison to writing has
- fewer third person pronouns, reflecting hardly any third person reference to participants in the same virtual room other than by way of nicknames
In addition to the traits noted in the 12 bullet points for conversational writing, split-window ICQ (SSCMC) in comparison to writing has
- more possibility and prediction modals, signaling a high degree of involvement and sensitivity to developing relationships
- more predicative adjectives, most of which reflect evaluative and/or supportive discourse content
Compared to speech, conversational writing (SCMC and SSCMC) has
- lexical density similar to that of face-to-face conversations
- slightly shorter clause length
- slightly shorter word length
- similar explicitly involved, interpersonal content as oral conversations, reflected in e.g. similar or more frequent use of first and second person pronouns and present tense verbs
- fewer prepositional phrases, reflecting very limited clausal elaboration ← 279 | 280 →
- situation-dependent reference similar to that of oral conversations, reflected in e.g. frequent use of time adverbials and infrequent use of WH-relative constructions
- similar non-abstract, non-impersonal content as in oral conversations, with e.g. few conjuncts and passive constructions
- less real-time informational elaboration by way of e.g. THAT verb complements and THAT relatives in object position
- a similar ratio of analytic to synthetic negation (with more of the former)
- graphemic script (unlike speech), which enables the encoding of paralinguistic features, e.g. graphic imagery, repeated exclamation marks and uppercase words
- nearly similar reference to extra-linguistic content (even though conversational writing lacks options for accompanying the exophoric reference with e.g. glances, pointing and actions in a shared physical space)
- more limited paralinguistic means for expressing emotions, attitudes and sentiments
In addition to the 12 traits above, Internet relay chat (SCMC) in comparison to speech has
- fewer third person pronouns, reflecting hardly any third person reference to participants in the same virtual room other than by way of nicknames
In addition to the traits noted in the 12 bullet points for conversational writing (compared to speech), split-window ICQ (SSCMC) in comparison to speech has
- slightly more possibility and prediction modals, signaling a high degree of involvement and sensitivity to developing relationships
- more predicative adjectives, most of which reflect evaluative and/or supportive discourse content
The third research question, as to how the genres of SCMC, SSCMC and ACMC studied relate to oral conversations on Biber’s dimensions, was addressed in the previous chapter and above in the present chapter. The requisite data for addressing the question was illustrated in the dimension graphs of chapter 5 and discussed, and a conclusive analysis of the data was provided in section 6.3. First, the analysis in this chapter (section 6.2) appeared to support the initial hypotheses of the study, whereby the CMC genres were assumed to display a declining degree of orality (i.e. similarity to oral conversations) in the order SSCMC > SCMC > ACMC. Secondly, however, the analysis in this chapter (section 6.3) approached the dimension scores of the conversational genres and ACMC on Biber’s (1988) dimensions of linguistic variation from a different perspective. By moving away from ← 280 | 281 → genre boundaries to find similarities in text types across the genres (cf. Biber 1989, 1995), invoking the multidimensional characterization of texts, it was possible to determine, in a complementary way, the relationships between the supersynchronous and synchronous texts, the ACMC genre, and oral conversations. The results showed that both conversational writing genres contain texts of the same text types as most oral conversations. Whereas the split-window ICQ texts belong to the text type “intimate interpersonal interaction” (text type 1), the IRC texts are more difficult to analyze but nevertheless belong to text type 2, “informational interaction,” which for IRC entails just as “oral,” but less intimate, interaction.120 Collot’s (1991) genre of ACMC, however, was deemed by its more peripheral relationship to the multidimensional characterizations of text types 1 and 2, to be more distant than conversational writing from face-to-face and telephone conversations.
The fourth research question, as to whether conversational writing constitutes a modality of its own, brings us back to figure 1.2, illustrating the working relationship between modalities, media and genres/modes in the present study. Modalities are means of production/reception of linguistic content, of which three are regularly recognized in linguistics: speech (language conveyed via acoustic signals), writing (language encoded/decoded in written or typed characters) and sign language (language encoded/decoded in manual and non-manual signs). It was seen in section 1.1, and passim, that linguists have recurrently characterized computer-mediated communication as a hybrid variety of communication that occupies the middle ground between the first two modalities (leaving sign language out of the account, as in the present study) or as a variety different from all three. Figure 1.2 illustrates the latter relationship, not assuming a priori a hybrid status for conversational writing (SCMC and SSCMC), but nevertheless a tentative status as a fourth modality (ACMC being subsumed under the written modality). It is now time to address the question raised in connection with the figure, in the light of insights gained in the study.
Conversational writing has been put to the test repeatedly throughout this study, in chapter 4 relating the genres of SCMC and SSCMC to the media of writing and speech, and in chapter 5 more expressly contrasting the conversational writing genres to the written and spoken genres. Discussions of textual examples, illustrating structural patterns (e.g. a similar low lexical density) and recurring lexico-grammatical features, have consistently found a great similarity ← 281 | 282 → between the conversational writing texts and oral conversational texts, and a noticeable dissimilarity between the former and texts of traditional writing. From the text-linguistic point of view, then, conversational writing and oral conversations are strikingly similar. Bear in mind, however, that in the consideration of spoken texts here, the prosodic mark-up was either removed (cf. section 3.4) or largely ignored. Variation in intonation (pitch), stress (loudness), pauses, pace and rhythm, as well as other vocal traits of speech, have thus not been taken into account in the comparisons. In speaking, such vocal traits and, furthermore, gestures, facial expressions, shrugs and glances, as well as, for instance, conventions of body posture, all critically contribute to the encoding of messages with the interlocutor’s intention, stance and attitude. In conversational writing, chatters tend to exploit paralinguistic devices (repeated letters, capitals, exclamation marks, emotives etc., the latter arguably linguistic, as seen in section 4.5) to substitute for the lack of acoustic means and facial expressions. However, even though these devices add some expressiveness to messages, they are not capable of carrying all the nuances typically encoded in speech (cf. Crystal 2004b). Conversational writing is rather inherently different from speech, especially from face-to-face conversations, in this respect.
Conversational writing differs from oral conversations in several more respects. For one thing, its production and reception is slower than in speech, owing to the relatively slow pace of typing. For another, written conversations persist for some time on the screen (typically in a scrollable window), or optionally in a log file, whereas spoken conversations are genuinely ephemeral. The textual persistence makes, for instance, response elicitors (e.g. Okay?, Pardon?, What) rare in conversational writing, even though they are common in conversational speech (as seen in section 4.6). The textual persistence of conversational writing, furthermore, enables interlocutors in IRC to participate in several conversations at once, in one or multiple windows – a peculiarity unparalleled in conversational speech (Crystal 2004b). What is more, IRC messages differ from speech in that they are delivered upon their completion, whereas in speech, and in split-window ICQ chat, messages are decoded while under construction. This makes interlocutors in IRC unable to signal their understanding or puzzlement, or any feedback equivalent to a nod or a backchannel (such as mhm), while the conversational partner is composing their turn, even though such signals (response forms; see section 4.6) do appear in IRC upon the interlocutor’s receipt of the sender’s full turn.
In sum, it is evident that conversational writing is dependent on the encoding and decoding of typed characters and that the set of keys on the keyboard determines what kind of information can be conveyed (just as traditional writing is ← 282 | 283 → confined within the bounds of the alphabet). This kinship to writing, however, is challenged when the synchronicity of the communication is taken into account; for, whereas traditional writing is communicated one way or in asynchronous exchanges, produced in one context and received later in another, conversational writing is communicated in real time, synchronously or supersynchronously, in two-way exchanges (cf. table 1.1). Authors of traditional writing, accordingly, rarely refer exophorically to the extra-linguistic context or situation in which their text is produced. Rather, cohesion in writing must be lexicalized and the state of things made explicit (cf. sections 2.2 and 4.5). In conversational writing, by contrast, exophoric reference is possible in principle (when interlocutors share, for instance, audio files, images, web content and the like). The two factors mentioned, the synchronicity of communication and the degree of shared context (the co-spatiality), in fact, appear to be fundamental for the distinction between writing and conversational writing. To begin to relate the two factors, and their influence on texts, the factors are here combined into a matrix, figure 6.1, in which the x-axis divides the genres studied by their synchronicity of communication and the y-axis determines their position in terms of degree of shared context (“no” shared context, “limited” degree of shared context and “high” degree of shared context). Just as in the figures of chapter 5, the written genres in figure 6.1 are represented by black bullets (for a list of the genres, see Appendix I), spoken genres by gray, and the conversational writing genres by white bullets.
Figure 6.1 clearly illustrates the divide between the written and spoken genres, and the intermediate position of the conversational writing genres. All genres of traditional writing studied reside in the leftmost bottom sector, as they represent asynchronous communication in which the producer and recipient do not share the same context. At the same time, most of the spoken genres (prepared and spontaneous speeches, interviews and face-to face-conversations) reside in the upper third of the matrix, characterized by a high degree of shared context, except for broadcasts, in which the situation of production and reception are separate contexts, and telephone conversations, in which interlocutors share only the auditory context.121 The spoken and conversational writing genres are all represented in the synchronous and supersynchronous parts, with face-to-face and telephone conversations symbolically positioned on the dividing line between the synchronicities, as oral conversations may contain limited stretches of complete overlap (cf. sections 1.2 and 1.3).122 Internet relay chat is exclusively synchronous communication, whereas split-window ICQ chat admits extensively overlapping, supersynchronous, turns, which assigns the genre a position in the rightmost third of the matrix in figure 6.1, to set it apart from the limited supersynchronicity in oral conversations (cf. table 1.1).
As mentioned, participants in computer-mediated conversational writing share a limited semiotic field. The field is defined by features of the software window, the ongoing interaction and the surrounding information shared on participants’ screens (e.g. web content and shared files). Although the chatters’ discourse is heavily influenced by the synchronicity of their interaction and chatters’ occasional sense of shared context, it is nevertheless restricted to the ← 284 | 285 → modality of writing, reduced to the characters on the keyboard. Telephone conversationalists also share a limited context, confined as they are to the auditory context, but their communication shares the richness of the face-to-face genres in that interlocutors are able to use prosody to convey meaning.123 The semiotic richness of face-to-face communication, its high degree of shared context, is not easily recreated in computer chatting. Conversational writing, despite its resemblance to oral conversations, is, after all, still writing.
On the other hand, conversational writing texts differ markedly from traditional writing. For one, traditional writing is bound to static and permanent representations, prototypically on sheets of paper. For another, the asynchronous character of traditional writing enables authors to carefully plan, redraft and edit their texts and to compose elaborate constructions. As there is no recipient simultaneously present (as in conversational writing), there is no pressure on authors to communicate rapidly. Moreover, most traditional writing consists of one-way texts with the character of a monolog, whereas chatted texts by default are dialogs, or even “polylogues” (conversations between ≥3 people, Kerbrat-Orecchioni 2004). Despite all these differences, authors of traditional writing and the writers in computer-mediated communication all rely on the same means of representation for the production/reception of language, the graphemes, in themselves abstractions of phonemes. All in all, this reliance on the same means of representation makes conversational writing a variety subsumed under the modality of writing, although owing to its resemblance to oral conversations, the variety is pulled a long way in the direction of conversations. Figure 6.2 sums up and illustrates the relationships found in the present study between the modalities writing and speech, their respective media and the genres investigated. The genres of traditional writing are the same 17 genres represented by black bullets in figure 6.1 (as well as in chapter 5; see Appendix I for a list of these), and, again, the spoken genres are represented by gray and the conversational writing genres by white bullets. ← 285 | 286 →
Figure 6.2 thus illustrates the answer to the fourth research question, as to whether conversational writing (SCMC and SSCMC) constitutes a modality of its own (raised in section 1.2 in connection with figure 1.2). The findings in the present study have not evidenced that conversational writing is a new modality. Rather, the qualitative assessment of conversational writing in the present section, based on the combined results, indicates that the genres of conversational writing share the same modality as traditional writing and ACMC (the modality of writing), while at the same time being extreme offshoots of traditional writing. In fact, conversational writing is so spoken-like that its genres functionally, structurally and lexico-grammatically most fittingly are represented among the oral conversational genres in figure 6.2. The interspersing of the conversational writing genres among the spoken genres is justified, among other things, by the adherence of the conversational writing texts to Biber’s (1989, 1995) text types 1 and 2 (cf. section 6.3). The discussion in section 6.2, and the combined in-depth analyses of the genres’ positions on Biber’s (1988) dimensions in chapter 5, also lead to the same conclusion – Internet relay chat and split-window ICQ texts are about as “oral” as oral conversations, despite being conveyed via graphemes. Figure 6.2 thus expressly evolves the hybrid character categorizations of CMC made in previous studies (cf. section 1.1) by illustrating how the written and spoken genres intertwine in linguistic space. The figure, of course, is highly synoptic; the relative positions of the conversational writing genres – indeed all genres – are most accurately specified in a continuum of multiple dimensions, those explored in chapter 5 and above.
A number of authors and linguistic scholars who pioneered the exploration of computer-mediated communication asserted that, while displaying some similarities with traditional notions of spoken and written discourse, the linguistic ← 286 | 287 → character of online interaction is something entirely new and unique (e.g. Ferrara et al. 1991, Davis & Brewer 1997, Crystal 2001). Crystal (2001) went so far as to propound that linguistic accounts after the advent of the Internet must comprise four discourse varieties (cf. modalities): written language, spoken language, sign language and now computer-mediated language (as mentioned in section 1.2). Later, investigators attempted to defuse the initial insinuations of a fourth modality (or “fourth medium,” in Crystal’s 2001: 238 terms) by reverting to explanations of genre/activity variation, pointing to written character aspects and users’ creative adaptation of written language to the confined semiotic field(s) of CMC (e.g. Hård af Segerstad 2002). Regardless of their perspectives, most scholars agree that the electronic media “facilitate and constrain our ability to communicate in ways that are fundamentally different from those found in other semiotic situations” (Crystal 2004b: 68). As the novelty of the CMC media is wearing off, however, the adduction of explanations relating to genre/activity variation appears increasingly level-headed and progressive. Genres of speech, by all means, range from the most written-like genres, such as news broadcasts or prepared speeches (cued by written props or manuscripts), to intimate face-to-face conversations. Similarly, written genres, as seen here, range from the most prototypical, information-dense, elaborated pieces of text to the most oral-like pieces of conversational writing. At the same time, there is no simple dichotomy of writing vs. speech at the levels of medium and genre (see figure 6.2); rather, written and spoken genres intertwine. As early as 1988, Biber demonstrated this interspersing of written and spoken genres in linguistic space on his dimensions of linguistic variation (Biber 1988). The present study has only extended the range of written genres on the same dimensions and added to the complexity of linguistic variation in the English language. In future accounts of the complete textual variation in a language, conversational writing texts, with all their peculiarities, cannot be ignored.
The present section has revisited, discussed and summarized some of the answers to the four research questions addressed in this study. Before concluding the section, the promised review of the definition of conversational writing will be tended to. At the beginning of this section, the tentative definition (from section 1.1) was reiterated, by which conversational writing is written communication 1) for social interaction 2) which requires the simultaneous presence (physical or virtual) of producer and recipient, 3) in which interlocutors expect immediate feedback (i.e. within seconds) and 4) during which the discourse may be reconfigured by the participants while under construction (e.g. as interlocutors are able to influence each other’s line of thought). The discussion in this ← 287 | 288 → section was not intended to address the definition per se, nor did it give reason to question the four criteria for identifying conversational writing. The discussion, nevertheless, prompted a consideration of the situational features of the individual conversational writing genres, parameters in light of which conversational writing genres may be seen to differ from each other. In the consideration, at the beginning of the section, it was found that computer-mediated conversational writing, besides points 1–4, is also written communication in which participants adopt oral linguistic strategies that reflect the semiotic field (e.g. participants’ degree of shared context, potential reference to web content and shared files), the semiotic tenor (i.e. the parameters of participants’ social relationship, previous acquaintance and future off-line relations) and the semiotic mode (e.g. the configuration of the conversational writing client, its options for selecting the number and identity of participants, public and private space etc.) of their interaction. Conversational writing genres may be identified on the basis of the four-fold definition and described, and possibly grouped, by means of the parameters (the parameters are found in the parentheses here; the semiotic elements field, tenor and mode, of course, are at play in any communicative act). In the present study, the conversational writing genres have both essentially been found to be lexico-grammatically correspondent to oral conversations. Functionally, the dialogic configuration of split-window ICQ chat serves better for participants to further personal, real-life relationships, whereas the extremely fragmented, polylogic structure of IRC, rarely found in typical face-to-face and telephone conversations, invites more cursory acquaintance. Structural differences notwithstanding, both chat genres are inherently conversational; just as oral conversations, they involve real-time communication between interlocutors sharing features of the same situation and the ability to immediately affect each other’s contributions to the discourse.
It should be borne in mind that the four-fold definition of conversational writing indeed draws on the properties of the two conversational writing genres studied here, but equally well applies to, for instance, web chat and recent IM applications such as Facebook chat (when used for SCMC). A close functional, structural and/or lexico-grammatical examination of more genres (for instance, textual conversations in virtual worlds, with a higher degree of shared context, or the occasionally two-way synchronous communication in SMS, e-mail and Twitter; cf. table 1.1) is likely to occasion refinements of the definition. The definition offered, all the same, may serve as a starting point for the identification of conversational writing in existing and emergent modes of CMC and telecommunications. Conversational writing is likely to stay relevant well into the future, but ← 288 | 289 → also to evolve along unforeseeable paths. In this constant state of flux, the field of CMC linguistics rarely allows long-standing definitions, but continues to afford ample opportunity for scholars to explore emergent modes against the backdrop of previously described communications.
The purpose of the present chapter has been to synthesize and discuss the results of the full investigation. After a few introductory remarks, the chapter opened with the consideration of quantitative results pertaining to the two hypotheses underlying the study. By relating the positions of the conversational writing genres and the genre of ACMC to the oral conversational genres on Biber’s (1988) dimensions, it was possible to begin the assessment of the orality in conversational writing and asynchronous CMC. Initially, both hypotheses appeared to be supported; more precisely, the quantitative findings indicated that the highest degree of orality (i.e. lexico-grammatical similarity to oral conversations) was observed in supersynchronous conversational writing, followed by that in synchronous conversational writing, and showed that the asynchronous CMC genre was the least oral of the three CMC genres investigated. The chapter proceeded, however, via a close examination of the overall picture afforded by all dimensions, to assess the multidimensional character of the three CMC genres to identify the genres’ most prevalent text types (Biber 1989, 1995), in order to address the hypotheses further. The texts of the conversational writing genres studied, like most oral conversational texts in Biber’s studies, were found to belong to text types 1, “intimate interpersonal interaction,” and 2, “informational interaction,” whereas Collot’s (1991) genre of BBS conferencing adhered to the latter, but more for its similarity to, for instance, personal letters than to conversations. The quantitative and qualitative assessments in combination thus supported the first and rejected the second hypothesis; that is, in short, the study has found conversational writing to be more “oral” than asynchronous CMC, but the SCMC genre to be no less “oral” than the SSCMC genre. Next, the chapter revisited the research questions posed at the beginning of the study, and addressed throughout, to find and synthesize the answers provided to these. By way of a semiotic analysis of the communication in the material studied, it was possible, among other things, to relate the higher degree of orality initially found in split-window ICQ than in IRC to the more similar semiotic tenor and mode of the former and oral conversations, rather than to any supersynchronicity effect. Although no supersynchronicity effect was evidenced in the material, a substantial synchronicity effect was found, which likens the conversational writing texts, of both chat genres, to ← 289 | 290 → oral conversations. A selective bullet-point overview of the relationships found between conversational writing, writing and speech was then provided, before the last research question, as to whether conversational writing constitutes a modality of its own, was addressed and answered. In short, conversational writing was found to rely on the modality of writing but to convey discourse most closely akin to oral conversations. Finally, the definition of conversational writing was revisited in light of the findings. The final chapter, below, offers a concluding summary of the full study and some suggestions for further research.
110 The dimension scores of BBS conferencing in table 5.5 are based on the standard scores (called FDS, “feature deviation scores”) of “ELC other” reported in Collot (1991: 69–70), and are not those erroneously arrived at in Collot (1991: 77–79). Corrected dimension scores for the genre were computed in this study; see section 5.1.
111 The distances in table 6.1 are given in absolute values to enable the comparison of totals. Thus, the difference between the Dimension 1 score of IRC (25.6) and face-to-face SBC (43.7), for instance, is indicated as the positive value 18.1, i.e. as the interval of 18.1 standard deviation units on the dimension. See table 6.2 for the mean dimension scores (“mean”) of the conversational genres (derived from table 5.1 and from Appendix VIII) and table 5.5 for those of BBS conferencing.
112 The totals in the rightmost column of table 6.1 are the sums of the standard deviation units separating the dimension scores of the relevant genres. The totals in table 6.1 are provided only to enable the surveying of all the relationships at once and must not be confused with dimension scores, as dimension scores cannot be summed across dimensions to provide an overview of the character of genres.
113 The stepwise presentation of the t-tests here (via t-values) serves two purposes, both of which pertain to the replicability of the present study: 1) The t-value calculation explains how it was possible to carry out the tests (to obtain p-values) even without access to the dimension scores of Biber’s (1988) individual texts. As explained, this was done via the computation of the standard error of the mean (SEM) of Biber’s texts; the formulae and table 6.2 serve to clarify the procedure and the data involved. (By contrast, the same calculation was not feasible for the ACMC genre, as no standard deviation is specified in Collot 1991.) 2) The t-values in table 6.2 show discernible trends for the data, which might be of interest in future studies (even if the corresponding multiplicity adjusted p-values in table 6.3 indicate no difference for the genres in question).
114 In table 6.4, the genres in each text type are listed according to the centrality of their texts (cf. Biber 1989), i.e. from genres with more central texts in the cluster to genres with more peripheral texts (except for in text type 1, in which the texts of both genres are essentially equally central in the cluster).
115 The plotting of texts discussed here was carried out by hand onto Biber’s (1989: 15) graph of cluster distributions with respect to Dimensions 1 and 3. The resulting graph is not produced here, as no numeral dimension score values are available for Biber’s (1988) individual texts, rendering the production of a new graph for print here unfeasible.
116 As the dimension scores of Collot’s (1991) individual texts are unavailable, the plotting and multidimensional characterization of BBS conferencing here are based on the mean dimension scores of the genre (table 5.5) rather than on the scores of individual texts.
117 Carrying out a new comprehensive cluster analysis of Biber’s (1988) texts and those investigated in the present study was not feasible, as the dimension scores for Biber’s (1988) individual texts are unavailable. (Regardless, it is unlikely that adding a few dozen texts to Biber’s 481 texts in the analysis would much alter the cluster patterns even if this were done.) Instead, to inform and substantiate the qualitative account given here, the dimension scores of texts in the present study, and those of the ACMC genre, were related to the cluster centroids of Biber’s (1988, 1989, 1995) texts via a computation of the Euclidean distance obtaining between the scores and the centroids. The results of the computation are provided in Appendix IX.
118 To facilitate further research, the dimension scores of individual texts of the corpora annotated in the present study are given in Appendix X.
119 Inserts of the categories mentioned are not included among Biber’s (1988) features and, consequently, have no bearing on the quantitative results of the MD analysis. Upon inclusion of conversational writing genres in future MD analyses, the consideration of various inserts (cf. Biber et al. 1999: 1082ff) is recommended, as well as the consideration of lexical density and emotives (cf. sections 4.3 and 4.6).
120 In the discussion of the text type of the Internet relay chats, the disjointed character of the discourse was elaborated on and further research into text type 2 was eventually suggested, such that its label might encompass texts of the IRC kind.
121 The genre of broadcasts is extremely diverse and eludes simple classification in figure 6.1, as it may contain, for instance, synchronous texts of all three degrees of shared context (depending on the studio setting and persons involved, etc.). The position opted for to denote the genre here indicates a live broadcast in which the producer of the broadcast discourse and the final recipient are in different locations, as is the case in, for instance, a radio news broadcast. The vast majority of the LLC texts in the genre derive from radio broadcasts (Greenbaum & Svartvik 1990).
122 The “interviews” genre contains interviews, public conversations and debates (Biber 1988, 1995: 87) and may, like conversations, contain limited stretches of overlapping speech, motivating the same position as face-to-face conversation. Its position in the synchronous sector here, however, serves to illustrate that the typical turns produced in the genre are very long (rather monologic) and that the genre contains significantly fewer overlaps than do the face-to-face and telephone conversations.
123 Neither of the axes in figure 6.1 represents a continuum; rather, the genres are represented in sectors to which they conceptually belong (cf. table 1.1). If the vertical axis were a continuum, telephone conversations might be positioned above conversational writing, but this would entail that the matrix indicates variable degrees of shared context in the traditional writing genres as well, which it is not intended to do.