7. Los servicios La gran ausencia del Estado
7.2. Otras asociaciones, otras terapias
Although Web conferences are multimodal and feature audio-, video- and text- based data, the present study examines only audio data. The audio data from each hour- long conference were transcribed according to the conventions set forth by Jefferson (2004). This system was chosen for its relative intuitive coding of paralinguistic information (e.g., pausing, overlapping speech, etc.) and the resulting accessibility and understandability of the end transcription. The transcribed data were then analyzed to find all instances in which either expert or novice participants attempted to engage in
requesting behavior. Each request was then rated to indicate its level of directness, frequency and kind of internal modifiers, frequency and kind of external modifiers, and level of appropriateness.
4.4.2 Transcription of Audio Data
Audio data from each Web conference were screened for participant request production using the freely available recording, editing and playback software Audacity (http://audacity.sourceforge.net/). After listening to the associated audio file and reading the transcription as many times as was necessary, the researcher established the boundary of each request sequence by documenting precisely when it began and when it ended, according to the time displayed on the Audacity playback stream. Only those audio file excerpts that contained request sequences were then transcribed into text-based data, using the time boundaries established in the previous step. The textual data also indicated the speakers involved in each request sequence. As such, each instance of requesting behavior was both preserved in its original audio form and documented as a written text.
4.4.3 Identification of Head Acts
In order to determine the head act of the transcribed request sequences in the data set, the researcher utilized the approach of Blum-Kulka et al. (1989, see also 1.3.1) and sought the “minimal unit” that conveyed the requestive force. Most head acts were
formulated at the sentence level, and in requests featuring much internal modification, the head acts often showed a high degree of syntactic complexity. Many head acts were formulated at the phrasal level (i.e., a noun phrase + a verb phrase), and the researcher also identified elided head acts that relied on contextual information to convey the
illocutionary force. Once the researcher identified the head act, it was highlighted in
green in the transcription (for information regarding transcription of audio data, see 4.4.2). To ensure reliability in identifying the head act of each request sequence, the researcher followed a designated set of procedures. First, the researcher read each transcribed request sequence several times and determined the sentence or set of sentences in which the minimal requestive unit was located. Next, the sentences were pared down by identifying the presence of external modifiers, which are a category of non-essential utterances occurring outside the boundary of the head act (see 1.3.2.3. and 4.2.3). After the external modifiers were excluded, the remaining utterance was then examined for a minimal unit occurring at the sentence or phrasal level. If such a minimal unit could be found, it was highlighted as the head act. (As noted above, the length of the head act often correlated with its syntactic complexity.) If such a unit could not be found at the sentence or phrasal level, the researcher proceeded to examine the data at the word level in order to determine which exact word(s) carried the requestive force. Once this element was identified, it was marked as the head act of the request sequence. By starting at a broad level and successively narrowing the focus, the researcher was able to
determine with confidence the boundary of the head act for each request sequence. 4.4.4 Further Data Coding and Rating by Primary Researcher
After identifying all requests in the data set, the researcher coded the transcribed request sequences according to participant as well as the experimental condition and Web conference in which they were produced. Finally, the researcher rated each instance of requesting behavior according to the level of directness of the head act, the frequency and
kind of internal modification, the frequency and kind of external modification, and the overall level of appropriateness.
4.4.5 Researcher Bias
It must be acknowledged that the procedures discussed thus far leave open the possibility for a subjective analysis and presentation of research results, for two main reasons. Firstly, the researcher devoted a great deal of time and energy to this project, both as a pedagogue and the principle investigator, and there existed a natural hope that pragmatic development had taken place among the learners. Secondly, and more broadly, the primary researcher is the product of his own ontological development, and may have exhibited an unconscious tendency to interpret the data subjectively (Merriam et al., 2002), as influenced by his position as an American speaker of German. However, the researcher decided against conducting an inter-coder reliability study of the process of identifying head acts for two reasons. First, by establishing and following a prescribed protocol, the researcher was able to approach each request sequence systematically and objectively, thereby helping to ensure a high degree of intra-coder reliability.
Additionally, the researcher felt that the risk of researcher bias was more likely to occur in rating the illocutionary force of the data than in identifying head acts. It was therefore decided to utilize the researcher’s limited resources conservatively and to focus on establishing inter-rater reliability, as is described in the next section 4.5.