Sobre la unidad de análisis para el estudio de los deseos en los actos del habla con el Algoritmo David Liberman (ADL)

this paper aims at discussing methodological problems of reliability related to the application of the dla instrument to the analysis of wishes in the patient′s and therapist′s speech acts developed during the session. the paper is composed by five parts. in the first one we describe some generalities about the method (dla) and in particular about the instrument for the analysis of speech acts, and we present the problem related to the criterion to decide which is the type of unit of analysis (speaking turns, sentences, utterances) more appropriate from the methodological point of view. in the second part we describe 5 inter-judge reliability tests: 1) establishment of the sample, 2) segmentation and analysis of utterances, 3) segmentation and analysis of sentences, 4) segmentation and analysis of the patient′s speech, 5) segmentation and analysis of the therapist′s speech. the same pair of raters participated in all the tests. due to the fact that the results of the analysis of wishes with each type of segmentation are different one from the other, in the third part we discuss which is the type of unit of analysis that has the is preferable in the quantitative research. in order to respond to this problem we compare the sensitivity of the different types of segmentation of the sample. the results of this comparison indicate that when the unit of analysis is the utterance, the research may capture the nuances of the speech acts in a finer way than in the other units of analysis. thereafter, the results of this test suggest that in the quantitative research it is better to segment the sample into utterances. in the fourth part we describe two additional reliability tests between independent judges about the segmentation and the analysis of the two samples, consisting respectively in a session of a different patient. first study, kappa .776, sig. < .001, and second study kappa .738, sig. < .001. in the fifth part we show two strategies of presentation of the outcomes