Improving the workflow to crack Small, Unbalanced, Noisy, but Genuine (SUNG) datasets in bioacoustics: The case of bonobo calls.

Despite the accumulation of data and studies, deciphering animal vocal communication remains challenging. In most cases, researchers must deal with the sparse recordings composing Small, Unbalanced, Noisy, but Genuine (SUNG) datasets. SUNG datasets are characterized by a limited number of recordings, most often noisy, and unbalanced in number between the individuals or categories of vocalizations. SUNG datasets therefore offer a valuable but inevitably distorted vision of communication systems. Adopting the best practices in their analysis is essential to effectively extract the available information and draw reliable conclusions. Here we show that the most recent advances in machine learning applied to a SUNG dataset succeed in unraveling the complex vocal repertoire of the bonobo, and we propose a workflow that can be effective with other animal species. We implement acoustic parameterization in three feature spaces and run a Supervised Uniform Manifold Approximation and Projection (S-UMAP) to evaluate how call types and individual signatures cluster in the bonobo acoustic space. We then implement three classification algorithms (Support Vector Machine, xgboost, neural networks) and their combination to explore the structure and variability of bonobo calls, as well as the robustness of the individual signature they encode. We underscore how classification performance is affected by the feature set and identify the most informative features. In addition, we highlight the need to address data leakage in the evaluation of classification performance to avoid misleading interpretations. Our results lead to identifying several practical approaches that are generalizable to any other animal communication system. To improve the reliability and replicability of vocal communication studies with SUNG datasets, we thus recommend: i) comparing several acoustic parameterizations; ii) visualizing the dataset with supervised UMAP to examine the species acoustic space; iii) adopting Support Vector Machines as the baseline classification approach; iv) explicitly evaluating data leakage and possibly implementing a mitigation strategy.

Toward a phenomenology of taking care.

CONTEXT AND PURPOSE: From nurses to dentists and doctors, caregivers undergo significant initial and life-long training. This training, however, rarely addresses the subjective side of their practice, especially the lived experience of caregiving. Better understanding this experience can nevertheless help to build fruitful relationships with patients. We focus on what it is like to take care of someone else and attempt to outline an encompassing "phenomenology of care". METHODS: We investigate the lived experience of caregivers during their first meeting with a patient. We rely on micro-phenomenological interviews, which offer fine-grained, first-person descriptions of someone's holistic experience in a given situation. RESULTS: We show how the subjective experience of meeting a new patient can be structured with i) categories of micro-experiential acts (gathering information, assessing and performing actions), ii) the scopes of these acts, which involve inner and outer perceptions, various elaborations, regulations and interventions and iii) a range of experiential modalities. CONCLUSION: We highlight the richness of lived experience, and what all caregivers intimately share beyond the frame of their respective professions and practices. We discuss our results in terms of methodology, finalized and productive activities, pre-reflective aspects, and reflexive practice.

Time-synchronic comments on video streaming website reveal core structures of audience engagement in movie viewing.

To what extent movie viewers are swept into a fictional world has long been pondered by psychologists and filmmakers. With the development of time-synchronic comments on online viewing platforms, we can now analyze viewers' immediate responses toward movies. In this study, we collected over 3 million Chinese time-synchronic comments from a video streaming website. We first assessed emotion and cognition-related word rates in these comments with the Simplified Chinese version of the Linguistic Inquiry and Word Count (SCLIWC) and applied time-series clustering to the word rates. Then Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) was conducted on the text to investigate the prevalent topics among the comments. We found different commenting behaviors in front of various movies and prototypical diachronic trajectories of the psychological engagement of the audience. We further identified how topics are discussed through time, and tried to account for viewer's engagement, considering successively movie genres, topics and movie content. Among other points, we finally discussed the challenge in explaining the trajectories of engagement and the disconnection with narrative content. Overall, our study provides a new perspective on using social media data to answer questions from psychology and film studies. It underscores the potential of time-synchronic comments as a resource for detecting real-time human responses to specific events.

Different languages, similar encoding efficiency: Comparable information rates across the human communicative niche.

Language is universal, but it has few indisputably universal characteristics, with cross-linguistic variation being the norm. For example, languages differ greatly in the number of syllables they allow, resulting in large variation in the Shannon information per syllable. Nevertheless, all natural languages allow their speakers to efficiently encode and transmit information. We show here, using quantitative methods on a large cross-linguistic corpus of 17 languages, that the coupling between language-level (information per syllable) and speaker-level (speech rate) properties results in languages encoding similar information rates (~39 bits/s) despite wide differences in each property individually: Languages are more similar in information rates than in Shannon information or speech rate. These findings highlight the intimate feedback loops between languages' structural properties and their speakers' neurocognition and biology under communicative pressures. Thus, language is the product of a multiscale communicative niche construction process at the intersection of biology, environment, and culture.

Assessing sound symbolism: Investigating phonetic forms, visual shapes and letter fonts in an implicit bouba-kiki experimental paradigm.

Classically, in the bouba-kiki association task, a subject is asked to find the best association between one of two shapes-a round one and a spiky one-and one of two pseudowords-bouba and kiki. Numerous studies report that spiky shapes are associated with kiki, and round shapes with bouba. This task is likely the most prevalent in the study of non-conventional relationships between linguistic forms and meanings, also known as sound symbolism. However, associative tasks are explicit in the sense that they highlight phonetic and visual contrasts and require subjects to establish a crossmodal link between stimuli of different natures. Additionally, recent studies have raised the question whether visual resemblances between the target shapes and the letters explain the pattern of association, at least in literate subjects. In this paper, we report a more implicit testing paradigm of the bouba-kiki effect with the use of a lexical decision task with character strings presented in round or spiky frames. Pseudowords and words are, furthermore, displayed with either an angular or a curvy font to investigate possible graphemic bias. Innovative analyses of response times are performed with GAMLSS models, which offer a large range of possible distributions of error terms, and a generalized Gama distribution is found to be the most appropriate. No sound symbolic effect appears to be significant, but an interaction effect is in particular observed between spiky shapes and angular letters leading to faster response times. We discuss these results with respect to the visual saliency of angular shapes, priming, brain activation, synaesthesia and ideasthesia.

Modeling Linguistic Variables With Regression Models: Addressing Non-Gaussian Distributions, Non-independent Observations, and Non-linear Predictors With Random Effects and Generalized Additive Models for Location, Scale, and Shape.

As statistical approaches are getting increasingly used in linguistics, attention must be paid to the choice of methods and algorithms used. This is especially true since they require assumptions to be satisfied to provide valid results, and because scientific articles still often fall short of reporting whether such assumptions are met. Progress is being, however, made in various directions, one of them being the introduction of techniques able to model data that cannot be properly analyzed with simpler linear regression models. We report recent advances in statistical modeling in linguistics. We first describe linear mixed-effects regression models (LMM), which address grouping of observations, and generalized linear mixed-effects models (GLMM), which offer a family of distributions for the dependent variable. Generalized additive models (GAM) are then introduced, which allow modeling non-linear parametric or non-parametric relationships between the dependent variable and the predictors. We then highlight the possibilities offered by generalized additive models for location, scale, and shape (GAMLSS). We explain how they make it possible to go beyond common distributions, such as Gaussian or Poisson, and offer the appropriate inferential framework to account for 'difficult' variables such as count data with strong overdispersion. We also demonstrate how they offer interesting perspectives on data when not only the mean of the dependent variable is modeled, but also its variance, skewness, and kurtosis. As an illustration, the case of phonemic inventory size is analyzed throughout the article. For over 1,500 languages, we consider as predictors the number of speakers, the distance from Africa, an estimation of the intensity of language contact, and linguistic relationships. We discuss the use of random effects to account for genealogical relationships, the choice of appropriate distributions to model count data, and non-linear relationships. Relying on GAMLSS, we assess a range of candidate distributions, including the Sichel, Delaporte, Box-Cox Green and Cole, and Box-Cox t distributions. We find that the Box-Cox t distribution, with appropriate modeling of its parameters, best fits the conditional distribution of phonemic inventory size. We finally discuss the specificities of phoneme counts, weak effects, and how GAMLSS should be considered for other linguistic variables.