Evaluating Sentence Segmentation in Different Datasets of Neuropsychological Language Tests in Brazilian Portuguese

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 2605–2614 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 2605 Evaluating Sentence Segmentation in Different Datasets of Neuropsychological Language Tests in Brazilian Portuguese Edresson Casanova1, Marcos V. Treviso2*, Lilian C. Hübner3, Sandra M. Aluísio1 1University of São Paulo, 2Instituto de Telecomunicações, 3Pontifical Catholic University of Rio Grande do Sul edresson@usp.br, marcos.treviso@lx.it.pt, lilian.hubner@pucrs.br, sandra@icmc.usp.br Abstract Automatic analysis of connected speech by natural language processing techniques is a promising direction for diagnosing cognitive impairments. However, some difficulties still remain: the time required for manual narrative transcription and the decision on how transcripts should be divided into sentences for successful application of parsers used in metrics, such as Idea Density, to analyze the transcripts. The main goal of this paper was to develop a generic segmentation system for narratives of neuropsychological language tests. We explored the performance of our previous single-dataset-trained sentence segmentation architecture in a richer scenario involving three new datasets used to diagnose cognitive impairments, comprising different stories and two types of stimulus presentation for eliciting narratives — visual and oral — via illustrated story-book and sequence of scenes, and by retelling. Also, we proposed and evaluated three modifications to our previous RCNN architecture: (i) the inclusion of a Linear Chain CRF; (ii) the inclusion of a self-attention mechanism; and (iii) the replacement of the LSTM recurrent layer by a Quasi-Recurrent Neural Network layer. Our study allowed us to develop two new models for segmenting impaired speech transcriptions, along with an ideal combination of datasets and specific groups of narratives to be used as the training set. Keywords: Sentence Segmentation, Impaired Speech, Neuropsychological Language Tests 1. Introduction Language assessment has been shown to be an efficient complementary tool for detecting cognitive and neuropsy- chological disorders, therefore present in most tests, tasks and batteries that evaluate cognitive processes. For ex- ample, neuropsychological language tests are an important tool for diagnosing individuals with significant depression in Alzheimer’s disease (AD) (Fraser et al., 2016), to differ- entiate between Mild Cognitive Impairment (MCI) and AD (Drummond et al., 2015), to differentiate between AD and other neurodegenerative dementias (Yancheva et al., 2015; Beltrami et al., 2018) and to differentiate variants of neu- rodegenerative dementias, such as in Primary Progressive Aphasia (PPA) (Fraser et al., 2014). Language assessment has been performed mainly by using discursive production in which narratives are largely used, since they are a natural form of communication and favor the observation of the patient’s functionality in everyday life (Tillas, 2015). The discourse tasks used to assess the narrative productions of elder individuals are often based on: (i) an illustrated story book without a text (e.g. Cin- derella), (ii) an immediate and delayed retelling of a story orally presented, or (iii) a single scene or a sequence of scenes, presented on pictures, of a common event that oc- curs in daily life. With regard to specific batteries used to evaluate language in discourse tasks, we can cite …

Content truncated to 500 words