Parkinson's disease is a neurodegenerative disorder involving speech and language deficits. Often, these are separately studied as proxies of motor and non-motor (e.g., cognitive) symptoms, respectively. Conversely, links between both dimensions remain virtually uncharted. This paper introduces a methodology that enables the synchronous study of acoustic and linguistic patterns in Parkinson's speech. Our findings show that verbs and nouns provided relevant acoustic and linguistic information not only to model motor impairments but also to understand non-motor symptoms like those that appear when Parkinson's disease patients develop mild cognitive impairment.

Pre-trained models generate speech representations that are used in different tasks, including the automatic detection of Parkinson’s disease (PD). Although these models can yield high accuracy, their interpretation is still challenging. This paper used a pre-trained Wav2vec 2.0 model to represent speech frames of 25ms length and perform a frame-by-frame discrimination between PD patients and healthy control (HC) subjects. This fine granularity prediction enabled us to identify specific linguistic segments with high discrimination capability. Speech representations of all produced verbs were compared w.r.t. nouns and the first ones yielded higher accuracies. To gaina deeper understanding of this pattern, representations of motor and non-motor verbs were compared and the first ones yielded better results, with accuracies of around 83% in an independent test set. These findings support well-established neurocognitive models about action-related language highlighted as key drivers of PD. Index Terms: computational paralinguistics, interpretability of pre-trained models, action verbs, Parkinson’s disease

Abstract Chatbots enable the automation of several components in customer service and allow the support of multiple users. Despite their multiple advantages, due to the large amount of conversations generated by a chatbot, it is difficult to determine whether customer requests are well-addressed. For practical reasons, chatbot’s effectiveness is evaluated manually based upon a small sample (randomly chosen) of conversations or through self-reported user satisfaction. This procedure does not guarantee the correct evaluation of the service because the sample is generally not large enough and self-reports might be influenced by different external factors not directly associated to the chatbot’s functioning. This study proposes a methodology for automatic evaluation of chatbot effectiveness in real production environments. The analysis considers convolutional neural networks adapted for natural language processing, using two parallel convolutional layers to evaluate questions and answers independently. The proposed model also incorporates filters to extract features with multiple temporal resolution. This methodology is tested upon real conversations of chatbots that provide service to two different companies. The results are compared to baseline models based on classical techniques with different pre-trained word embedding models. According to our results, the proposed approach provides accuracies between 78.95% and 80.18%, which outperforms the best result of the baseline models by 2.9%.

Dialect recognition is useful in many industrial sectors, par-ticularly with the aim of allowing a better interaction between customers and providers. The core idea is to improve or customize marketing and customer service strategies, de-pending on the geographic location, birthplace and culture. This study proposes different models to automatically dis-criminate between two Colombian dialects: “Antioqueño” and “Bogotano”, to the best of our knowledge this is the first work of Colombian dialect recognition based on real conver-sations from customer service centers. The proposed strategy consists of independent analyses, using information from speech recordings and their corresponding transliterations. On the one hand, classical approaches are used to model speech including prosody features, Mel frequency cepstral coefficients and the mean Hilbert envelope coefficients. For text models, Word2Vec and bidirectional encoding represen-tations from transformer embeddings are considered. On the other hand, a deep learning approach is applied by considering convolutional neural networks, which are trained using spectrograms and embedding matrices for speech and text, respectively. The implemented deep learning models seem to be more promising than the classical ones for the addressed problem. Further experiments will be considered to validate this claim in a wider spectrum of methods.