Illness Narrative Complexity in Right and Left-Hemisphere Lesions

Introduction

The analysis of the disruption of narrative reasoning in brain damaged people can lead to a deeper understanding of the nature of the disability, and to tailor rehabilitation plans and follow-up evaluations to the patient's needs (Tabor Connor, 2002; Harter, Japp, & Beck, 2005; Ingram, 2007; Hamilton, 2008). In fact, right brain damaged patients are unable to retain and elaborate the details of a story because they do not succeed in structuring the statements into a cohesive and consistent way, even if the phonological and morpho-syntactical levels are not deeply impaired (Butterworth, 1993; Tompkins, & Lehman, 1998; Harris Wright, & Capilouto, Wagovich, Cranfill & Davis, 2005; Lê, Coelho, Mozeiko, Krueger & Grafman, 2011; Ulatowska, Reyes, Olea Santos & Worle, 2011). Conversely, left brain damaged patients generally show poor capabilities of linguistic elaboration (e.g. functor and/or verb omission), but good capabilities to select contextual information (Hickok, Love-Geffren & Klima, 2002). Although the main linguistic functions are elaborated in the left hemisphere, the right hemisphere plays a role in prosody, abstract and figurative language comprehension. Usually, the analysis of narratives in brain lesions is carried out by means of formal linguistic approaches. Cohesion and coherence are among the most common linguistic measures of discourse. Cohesion refers to the continuity in the surface structure (word and sentence) of the discourse, and its analysis is carried out by counting cohesive ties such as references, conjunctions, lexical markers, and so on (Liles, 1985; Ellis, Rosenbek, Rittman, & Boylstein, 2005). Standardized measures (Laska, Hellblom, Murray, Kahan, & Von Arbin, 2001) have been proposed. Coherence refers to continuity in meaning or the overall interrelatedness of the discourse (Louwerse, & Graesser, 2005), and can be further divided into global and local (Agar, & Hobbs, 1982): local coherence refers to the interrelatedness, or topic maintenance, across adjacent utterances while global coherence refers to the interrelatedness, or topic maintenance, even across an entire conversation or narrative. However, these approaches do not grasp the meaning embedded into the narrative plot and its structure. In this respect, innovative models of formal analysis of illness narratives, i.e. the way in which patients spontaneously construct and organize the mental representation of their experience, might be valuable tools to complement the linguistic analysis, monitor the evolution of the disability, and evaluate the efficacy of the rehabilitative interventions. The present study is an attempt to develop and apply a qualitative-quantitative approach to the study of the complexity of narrative reasoning in patients observed one year after the initial stroke episode. It is a pilot experiment aimed at exploring the potentialities of a multi-disciplinary approach based upon Latent Dirichlet Allocation (LDA) (Blei, Ng, & Jordan, 2003) and Semantic Network Analysis (SNA) (Giani, Martone, 1998;Giani, Brascio, 2006; Zhuge, 2011; Chakraborty, S., Samanta, S., Mukherjee, A., Dey, N., Chaudhuri, S.S., 2013; Singh, A., Sharma, A., Dey, N., 2015). SNA is a bottom-up approach, because it allows to mathematically infer the episodes of an illness narrative, while LDA is a top-down approach, because it starts from an initial set of episodes from which the latent classes (topics) that generate the narrative are extracted. LDA has been widely used and adapted in computer vision, where the inference algorithms are applied to natural images in the service of image retrieval, classification, and organization (Dey, N., Karâa, W.B.A., 2015; Dey, N., Bhateja, V., Hassanien A.E., 2015; Nandi, D., Ashour, A.S., Samanta, S., Chakraborty, S., Salem, M., Dey, N., 2015; Chakraborty, S., Karâa, W.B.A., Dey, N., Banerjee, S., Salem, M., Azar, A.T., 2015). Likewise documents, images exhibit a combination of visual patterns, the “visual words”, and the same visual patterns recur throughout a collection of images. Two different sets of the complexity of the narratives of patients affected by left and right brain lesions were computed. The results showed that right hemisphere lesions entail a reduction of the narrative complexity, and suggested that the qualitative-quantitative analysis of this reduction can be a guide for the formulation of personalized rehabilitative planes.