In this chapter, the findings of the three research questions presented at the end of the first chapter are presented and discussed. The three research questions in the present study include:
1. How is reporting results, the most important communicative purpose of RAs, presented in moves across the sections of Abstract, Results, and Discussion?
2. How are the rhetorical moves of reporting results realized linguistically in these sections?
3. To what degree is reporting results presented and realized differently in RAs of hard sciences and soft sciences?
Therefore, to provide detailed illustrations and answers to the research questions, this chapter is presented in the following sequence. First of all, a move analysis, including occurrences and frequencies of moves as well as move patterns in the three sections are illustrated and discussed. After that, a content analysis, including the investigation of various moves to report research findings in single research articles are presented. After that, a section illustrating the linguistic features in terms of use of main verbs, modal verbs, lexical bundles, and use of voice are entailed to provide a more detailed examination of reporting research findings in RAs. Finally, this chapter ends with a section elaborating on the similarities as well as differences across RAs of the two disciplines.
51
Move Analysis
To investigate how reporting results in the major sections of an RA is presented in moves, and whether and how these sections are different in presenting results, it is first necessary to identify and count the occurrences of the moves related to the presentation of results. Since Abstracts, as mentioned in previous chapters, provide an overview of the whole study, only the sentences related to and after presentation of results and findings are included in the Abstract corpus. The Results corpus in the present study entails all the illustrations of data but excludes tables, figures, as well as other non-verbal presentation of data. Finally, the Discussion corpus includes the Discussion and possible Conclusion sections that end an RA. Based on the modified coding scheme of moves, as shown and illustrated in the previous chapter, the results from a move analysis of the various sections in the 48 RA samples are shown in Table 4.1.
Table 4.1. Frequency of moves in the 48 RAs in the present study.
Abstract Results Discussion
freq % freq % freq %
(S) summarizing results 48 46 122 11 146 25
(L) locating data 0 0 168 15 5 1
(R) reporting findings 0 0 373 34 37 6
(I) interpreting results/findings 17 16 293 27 106 18 (P) providing reasons/explanations 0 0 27 2 25 4
(E) evaluating results 4 4 28 2 50 9
(C) comparing results to literature 3 3 75 6 101 17 (B) indicating limitations/weaknesses 0 0 0 0 19 3 (A) indicating implication/application 32 31 4 0.4 60 10 (F) suggestions for future studies 0 0 3 0.3 37 6
Total 104 1,093 586
Note. freq = frequency of moves, %: percentage of moves within various sections
52
From Table 4.1, it can be observed that the three moves that occurred most frequently in the Abstract section were moves AS (summarizing results), AA (indicating implications and/or applications), and AI (interpreting results and findings), with the first move—AS as an obligatory move that occurs in all the 48 RAs after the range of occurrences is checked. In addition, it is also essential to note that in Abstract, the shortest and most concise section of an RA, writers tend to not only highlight the major findings of the whole study but also support them with possible interpretations and/or further indicate their implications or applications, as shown in the following examples:
[4.1] //AS// In this paper, we undertake a cross-linguistic analysis of collocation, semantic prosody, and near synonymy, drawing upon data from English and Chinese (pu3tong1hua4). //AA// The implications of the findings for language learning are also discussed. (AL05, Abstract)
[4.2] //AS// This alternative web implies the use of a simplified web browser and an adequate web design. //AI// Discussion of the need to have several levels of cognitive accessibility, equivalent (although not identical) content for this collective and the need for testable protocols of accessibility that support these people‘s needs is also included. //AA// This article finishes with conclusions about the potential impact of accessible pages in the daily life of people suffering from cognitive deficits, outlining the features to be considered within a user profile specification that support cognitive difficulties and with reflections about the suitability of SemanticWeb Technologies for future developments in this field.
(CH11, Abstract)
Example 4.1 shows the organization of the paper provided by the writers and an illustration of what has been found and what will be discussed in the final section of the paper. Similarly, in example 4.2, authors provided an illustration of the main findings, followed by the layout of the RA so that potential readers can clearly know what are being discussed in the final Discussion section.
53
However, several moves in the coding scheme do not occur in the Abstract at all, including AL (locating data), AR (reporting findings), AB (indicating limitation/weaknesses), and AF (need/suggestions to future studies), which can be explained by the very nature of Abstract to concentrate only on the most crucial findings as a result of the limitation of space.
Analysis of the moves in the Results section shows that the three moves with the highest frequency in the Results corpus are RR (reporting findings), RI (interpreting results and findings), and RL (locating data), indicating that the results section, compared to Abstract section, focuses more on presenting as well as interpreting the data obtained from the study although the move of summarizing results (RS) also occur frequently. In addition, these four major moves of this section constitute nearly 90 percent (87%) of all the moves in this section. Moves that rarely occur, those with a frequency as low as 4 or lower, or even do not occur in the Results include RB (indicating limitations/weaknesses), RA (indicating implications/applications), and RF (need/suggestions for future studies). This finding can be related to the major communicative purposes of the Results section, that is, to present and interpret the results instead of indicating limitations or implications/applications of the study, or suggesting future research.
Taking a closer look at the moves in Results, the top high-frequency moves in this section follow a possible cyclic pattern of RR followed by RI, which is then followed by another cycle of RR and RI, suggesting that the data and/or findings of a study are usually presented in the form of a move pair—first presenting the data, and then interpreting the data. A second pair of finding together with its interpretation will follow the first pair. The cyclic patterns that occur frequently are either RL→RR→RI→RR→RI or RR→RI→RR→RI, as shown in the following example:
54
[4.3] //RL// Table 7 displays the descriptive statistics for the language analysis test. //RR// The direct meta group had the highest mean score and the control group the lowest. //RI// However, a one-way ANOVA revealed that these differences were not significant, F(2, 88) = 1.09, ns. The results of the repeated measures ANCOVA showed that there was a significant effect for aptitude as the covariate, F(2, 88) = 4.95, p < 0.05 … //RR// However, the ANCOVA also showed that there was still a significant effect for CF after controlling for the effect of aptitude, F(2, 88) = 2.45, p < 0.001. //RI// That is, the learners‘ test performance changed over time after removing the effect of their language analysis ability. (TQ10, Results)
Example 4.3 illustrates the move pattern of RL→RR→RI→RR→RI, an example illustrating the move sequence of directly interpreting data before presenting a new set of data. In this way, readers may understand writers‘ interpretation as well as explanation of data before a new set of information is presented and discussed. A common 3-move pattern observed in the Results section is RS→RR→RI, which demonstrates that RA writers tend to provide an overview of the main, or general, findings before going into the specific presentation and interpretation of data, thus showing a cyclic pattern of RS→RR→RI→RS→RR→RI, as shown in the following example:
[4.4] //RS// To address this question, two separate sets of analyses were conducted for each dependent measure: proportion of correct recall, and proportion of correct and modified recall … //RR// This pattern on the first dependent measure (the mean proportion of correct recalls) approached but did not reach significance at p < .05 for several tests according to recast type, with the test for correct recall of all recasts and the test for correct recall of short recasts showing differences due to literacy level that fall between p = 0.05 and p
= 0.10. //RI// However, the impact of literacy level on the second dependent measure (the mean proportion of correct and modified recall) produced much lower p values, two of which reached significance at the p < .05 level … //RS//
To address this question, two new dependent measures were calculated by subtracting the proportion of recall (correct, or correct and modified) for short recasts from the proportion of recall for long recasts. The average difference in
55
proportion of correct recall was – 0.037 (SD = 0.21). //RR// An exact permutation test (Good, 2001) was conducted by computing the mean difference score for all 256 possible assignments of negative and positive signs to the absolute values of the difference scores. //RI// The resulting one-tailed p value (p
= 0.344) suggests there is no statistical evidence that, for the group as a whole, the participants‘ proportion of correct recall was dependent on the length of the recast. (TQ09, Abstract)
[4.5] //RS// The completion-time results are summarized in Figure 8 for trials ending in successful completion. //RR// An analysis of variance revealed that the number of objects in each set (n) contributed significantly to task completion time (F (5, 56) = 72.41; p < 0.001). Most relevant to our model however, was an interaction between the number of objects and the navigation mechanism (n × m) that also contributed significantly to task completion time (F (5, 56) = 12.16; p <
0.001). //RI// As predicted by the model, there was a crossover in efficiency between the two navigation methods between 3 and 4 items per set. //RS// This was substantiated by individual analyses of variance for each level of n as summarized in Table I. //RR// There was a small but significant interaction between blocking of verbal working memory and the navigation mechanism (F (1, 26) = 10.91; p < 0.01) … //RI// This interaction suggests that verbal working memory is used as an additional resource in the zoom condition, but not in the multi-window condition. (CH06, Results)
Example 4.4 and 4.5 show a three-move patterns in which writers first provide an overview of the data to be presented, followed by specific and factual data, which is eventually succeeded by interpretation of writers. Therefore, it could be concluded that the four most common moves in the Results section not only occur frequently but are also presented in a cyclic pattern that are presented similarly in various disciplines.
In contrast to Abstract, no cyclic patterns of moves were observed in the Abstract corpus.
In the Discussion section of RAs, as shown in Table 4.1, it can be observed that the major moves of this section are still DS (summarizing results) and DI (interpreting results/findings). However, the frequencies of four other moves, DE (evaluating results), DC (comparing results to literature), DA (indicating implications/
56
applications), and DF (need/suggestions for future studies) are also high, especially when compared to their frequencies the previous two sections discussed above. This suggests that in the Discussion section, writers, after summarizing and interpreting the results of their study, often want to go further, evaluating their findings in a larger research context. Therefore, to evaluate the results, compare them with what has been found in other studies, indicate their implications and applications, or show possible future research would become possible moves in this section. On a brief look, it can be observed that the Discussion section is more similar to the Abstract section as the former not only provides an overview of the main findings but draws possible implications and/or applications from the retrieved data. However, as the Abstract is much shorter in length, only the most crucial aspects could be presented, while in the Discussion section, authors have more flexibility in presenting the values as well as the contributions of their studies.
When taking a closer look at the move patterns in the Discussion section, it can be observed that common move patterns in this section include DS→DI→DS→DI and DS→DC→DS→DC, with the former similar to the most common move pattern in the Abstract section and the latter similar to the patterns found in the Results section, as shown in the following examples:
[4.6] //DS// In Experiment 1, participants were presented with sentences of the kind often used in previous research on implicit causality, for example ‗John defied Ted.‘ Only the verb is informative in these sentences … //DI// Specifically, readers consistently preferred adjectives matching the verb in evaluative valence to describe stimuli or agents, and adjectives that were less positive or less negative to describe experiencers or patients in the same sentences … //DS// In the second experiment, participants chose verbs given information about nouns in the sentences … //DI// Evidently, the use of these adjectives led most participants to choose experiencer verbs rather than action verbs for most sentences. (AL10, Discussion)
57
[4.7] //DS// Aggregating the results across all scenes tested, we found that awareness cues were first identifiable between blur levels 3 and 5 … //DC// The levels we found for providing awareness are somewhat more filtered (2 to 3 levels) than those found by Boyle et al. [2000]; thus, in our study, participants were able to garner awareness cues from blurrier scenes. We believe this difference is a result of using videos of a greater fidelity than Boyle et al. [2000].
//DS// Blur levels 1 and 2 are the only levels that adequately preserve privacy for all scenes … //DC// This is consistent with the Boyle et al. [2000] result, which found overlap for what we consider here to be mundane scenes. (CH05, Discussion)
In examples 4.6 and 4.7, writers first provide an overview of what has been found in the study before interpreting or comparing to previous studies to indicate the values as well as significance of their studies. It, therefore, seems plausible to assert that the rhetorical purpose of the Discussion is, in nature, between the Abstract and Results as it not only provides an overview of what has been found but also provides authors‘ interpretation as well as evaluation of the obtained data.
After investigating the occurrences of moves in the three sections, Table 4.1 further illustrates the frequencies as well as percentage of moves in the whole RA corpus. It can be observed that the three most frequently-used moves constitute more than half (64%) of the moves in the three sections. In other words, the moves of -S (summarizing results), -R (reporting findings), and -I (interpreting results and findings) are not only the most dominant moves in reporting results across the three sections of Abstract, Results, and Discussion, but they are usually presented in a cyclic pattern in the latter two sections with high number of occurrence. The move analysis in this section indicates the use of moves in reporting results, and in the next section, Content Analysis, features of reporting research findings in the same research article, is carried out to provide a more detailed interpretation to realize the features of reporting results in RAs.
58
Content Analysis
Although move analysis in the previous section reveals how Abstract, Results, and Discussion differ from one another in terms of moves related to reporting results, it is limited in providing detailed comparison across the three sections within a single article. Therefore, content analysis of the three sections in a single RA was further conducted to showcase how research findings are presented differently in the three sections in terms of level of generality and language use. To accomplish this goal, each RA in the present study was carefully examined to retrieve sentences in each of the three sections that report similar research results. In the following, examples in both RAs from applied linguistics and computer science are given for demonstration and interpretation.
Levels of Generality
It has been pointed out that various sections in an RA present information in different manners as these sections not only differ in length but also in their communicative purposes. As RA writers only present the most crucial findings in Abstract, data is presented in the most concise and general manner due to the limitation of text length. The Results section, which aims to present as well as interpret specific findings, would include the most specific illustration of data. And finally, in the Discussion section, RA writers not only give an overview of what they have found in the study but also provide possible implications as well as applications and suggestions for future studies. Therefore, the level of generality in this section is between the prior two sections. The following example from a single RA in the field of applied linguistics illustrates this variation in generality:
59
[4.8.1] //AS// Results indicated that when listening to speech with correct primary stress, the participants recalled significantly more content. (TQ03, Abstract)
[4.8.2] //RR// Subjects listening to Version A remembered a significantly greater number of ideas than subjects listening to Version B (p = .001) or to Version C (p
= .02). Subjects listening to Version A remembered significantly more main ideas than subjects listening to Version B (p = .001) or to Version C (p = .05) …Results (see Table 5) indicated a significant overall effect for condition (version of the text) [F(2, 87) = 2.442, p = .001]. (TQ03, Results)
[4.8.3] //DS// The mean scores for each experimental group on the recall task … : Group A scored higher than Group C, which in turn scored higher than Group B … //DI// The differences among the three groups of subjects were statistically significant for the recall (comprehension) data and nearly all of the ICES items … //DA// The results of this study complement current literature advocating the inclusion of primary stress in an ESL pronunciation curriculum … For example, students could practice identifying the primary stress in recordings of spoken discourse. Comparing speech samples with correct, misplaced, and missing primary stress may also help learners perceive primary stress and its meaning. (TQ03, Discussion)
From the example above, it can be observed that when reporting the research finding about listening to speech with correct primary stress, the writers use the most general expression ―recalled significantly more content‖ in example 4.8.1 in Abstract without providing any specific data to support or any justification or interpretation. In contrast, in Discussion, for the corresponding finding, the writers discuss the scores of different experimental groups for the purpose of discussing how the findings of the study could be applied in classroom settings in an ESL pronunciation curriculum—entailing a classroom activity that enables students identify and compare use of stress to perceive meanings, as shown in example 4.8.3. The same result in the Results section, presented in example 4.8.2, is again presented in a different manner.
The Results section presents it in the most detailed manner: different versions of listening materials and statistical tests are given to show the performance of the
60
participants in the three versions. By doing this, readers of this research article are able to know the details of the experiment, that is, the research procedure from which the results are obtained and whether the results, in this case, various versions, are significant statistically. Thus, the illustration of the same finding in the Results section is more specific than in Abstract or Discussion. The various levels of generality, as demonstrated above, reflects the different communicative purposes of the three sections in the research article, with Abstract providing a summary of the findings,
participants in the three versions. By doing this, readers of this research article are able to know the details of the experiment, that is, the research procedure from which the results are obtained and whether the results, in this case, various versions, are significant statistically. Thus, the illustration of the same finding in the Results section is more specific than in Abstract or Discussion. The various levels of generality, as demonstrated above, reflects the different communicative purposes of the three sections in the research article, with Abstract providing a summary of the findings,