Search

Family dogs are exposed to a continuous flow of human speech throughout their lives. However, the extent of their abilities in speech perception is unknown. Here, we used functional magnetic resonance imaging (fMRI) to test speech detection and language representation in the dog brain. Dogs (n = 18) listened to natural speech and scrambled speech in a familiar and an unfamiliar language. Speech scrambling distorts auditory regularities specific to speech and to a given language, but keeps spectral voice cues intact. We hypothesized that if dogs can extract auditory regularities of speech, and of a familiar language, then there will be distinct patterns of brain activity for natural speech vs. scrambled speech, and also for familiar vs. unfamiliar language. Using multivoxel pattern analysis (MVPA) we found that bilateral auditory cortical regions represented natural speech and scrambled speech differently; with a better classifier performance in longer-headed dogs in a right auditory region. This neural capacity for speech detection was not based on preferential processing for speech but rather on sensitivity to sound naturalness. Furthermore, in case of natural speech, distinct activity patterns were found for the two languages in the secondary auditory cortex and in the precruciate gyrus; with a greater difference in responses to the familiar and unfamiliar languages in older dogs, indicating a role for the amount of language exposure. No regions represented differently the scrambled versions of the two languages, suggesting that the activity difference between languages in natural speech reflected sensitivity to language-specific regularities rather than to spectral voice cues. These findings suggest that separate cortical regions support speech naturalness detection and language representation in the dog brain.

Background Effectively utilizing disease-relevant text information from unstructured clinical notes for medical research presents many challenges. BERT (Bidirectional Encoder Representation from Transformers) related models such as BioBERT and ClinicalBERT, pre-trained on biomedical corpora and general clinical information, have shown promising performance in various biomedical language processing tasks. Objectives This study aims to explore whether a BERT-based model pre-trained on disease-related clinical information can be more effective for cerebrovascular disease-relevant research. Methods This study proposed the StrokeBERT which was initialized from BioBERT and pre-trained on large-scale cerebrovascular disease related clinical text information. The pre-trained corpora contained 113,590 discharge notes, 105,743 radiology reports, and 38,199 neurological reports. Two real-world empirical clinical tasks were conducted to validate StrokeBERT's performance. The first task identified extracranial and intracranial artery stenosis from two independent sets of radiology angiography reports. The second task predicted the risk of recurrent ischemic stroke based on patients’ first discharge information. Results In stenosis detection, StrokeBERT showed improved performance on targeted carotid arteries, with an average AUC compared to that of ClinicalBERT of 0.968 ± 0.021 and 0.956 ± 0.018, respectively. In recurrent ischemic stroke prediction, after 10-fold cross-validation on 1,700 discharge information, StrokeBERT presented better prediction ability (AUC±SD = 0.838 ± 0.017) than ClinicalBERT (AUC±SD = 0.808 ± 0.045). The attention scores of StrokeBERT showed better ability to detect and associate cerebrovascular disease related terms than current BERT based models. Conclusions This study shows that a disease-specific BERT model improved the performance and accuracy of various disease-specific language processing tasks and can readily be fine-tuned to advance cerebrovascular disease research and further developed for clinical applications.

To provide explainable and accurate aspect terms and the corresponding aspect–sentiment detection, it is often useful to take external domain-specific knowledge into consideration. In this work, we propose a knowledge-enabled language representation model BERT for aspect-based sentiment analysis. Specifically, our proposal leverages the additional information from a sentiment knowledge graph by injecting sentiment domain knowledge into the language representation model, which obtains the embedding vectors of entities in the sentiment knowledge graph and words in the text in a consistent vector space. In addition, the model is capable of achieving better performance with a small amount of training data by incorporating external domain knowledge into the language representation model to compensate for the limited training data. As a result, our model is able to provide explainable and detailed results for aspect-based sentiment analysis. Experimental results demonstrate the effectiveness of the proposed method, showing that the knowledge-enabled BERT is an excellent choice for solving aspect-based sentiment analysis problems.

The paradigm of representation learning through transfer learning has the potential to greatly enhance clinical natural language processing. In this work, we propose a multi-task pre-training and fine-tuning approach for learning generalized and transferable patient representations from medical language. The model is first pre-trained with different but related high-prevalence phenotypes and further fine-tuned on downstream target tasks. Our main contribution focuses on the impact this technique can have on low-prevalence phenotypes, a challenging task due to the dearth of data. We validate the representation from pre-training, and fine-tune the multi-task pre-trained models on low-prevalence phenotypes including 38 circulatory diseases, 23 respiratory diseases, and 17 genitourinary diseases. We find multi-task pre-training increases learning efficiency and achieves consistently high performance across the majority of phenotypes. Most important, the multi-task pre-training is almost always either the best-performing model or performs tolerably close to the best-performing model, a property we refer to as robust. All these results lead us to conclude that this multi-task transfer learning architecture is a robust approach for developing generalized and transferable patient language representations for numerous phenotypes., Comment: Journal of Biomedical Informatics (in press)

In the last few years, people started to share lots of information related to health in the form of tweets, reviews and blog posts. All these user generated clinical texts can be mined to generate useful insights. However, automatic analysis of clinical text requires identification of standard medical concepts. Most of the existing deep learning based medical concept normalization systems are based on CNN or RNN. Performance of these models is limited as they have to be trained from scratch (except embeddings). In this work, we propose a medical concept normalization system based on BERT and highway layer. BERT, a pre-trained context sensitive deep language representation model advanced state-of-the-art performance in many NLP tasks and gating mechanism in highway layer helps the model to choose only important information. Experimental results show that our model outperformed all existing methods on two standard datasets. Further, we conduct a series of experiments to study the impact of different learning rates and batch sizes, noise and freezing encoder layers on our model.

This study analyzes lexical representation of G. Hofstede's gender values illustrating the opposition of masculine and feminine national cultures. The lexical material of the research is English and Russian phrasecons. One of the values “maximum / minimum emotional and social role differentiation between the genders” is analyzed in more detail including quantitative (the number of representations) and qualitative aspects (their semantics). The results show that Russian phraseological units representing the masculine value dominate in number but they are more metaphorical than English phraseological units. The findings correlate with the sociological study results obtained by G. Hofstede.

Introduction: The description of bilingual aphasic patients goes back to the 19th century. Since then, the study of the relationship between bilingualism and the brain has questioned whether the neuroanatomical representation of two languages in the same brain is similar or different. The answer to this question has generated erroneous interpretations based on the results of the investigations carried out on this topic. Development: The present paper will try to distinguish between myth and reality of the following statements: a) the neuroanatomical organization of language in bilinguals is different from that of monolinguals, and b) language is less lateralized in bilingual speakers. Results: There is no reason to believe in the existence of qualitative differences in the cerebral organization of language between bilinguals and monolinguals. It is most likely that two languages are represented as different microanatomical subsystems in the same cerebral regions. Conclusions: The differences are quantitative rather than qualitative, that is, the degree of participation of the different neurofunctional mechanisms involved in the use of language, such as metalinguistic knowledge and implicit linguistic competence. Resumen: Introducción: La descripción de pacientes bilingües con afasia se remonta al siglo xix. Desde entonces, el estudio de la relación bilingüismo-cerebro se ha preguntado si la representación neuroanatómica de las lenguas que habla una misma persona es similar o diferente. La respuesta a esta pregunta ha generado interpretaciones erróneas de los resultados obtenidos en las investigaciones realizadas al respecto. Material y métodos: En el presente trabajo, se describirá lo que hay de mito y de realidad en las siguientes afirmaciones: a) la organización neuroanatómica del lenguaje en el bilingüe es diferente del monolingüe, y b) la lateralización del lenguaje en el bilingüe es menor. Resultados: No hay motivos para creer que haya diferencias cualitativas en la organización cerebral del lenguaje en bilingües y monolingües. Lo más probable es que las lenguas que habla una misma persona estén representadas como subsistemas microanatómicos distintos en las mismas regiones cerebrales. Conclusiones: Las diferencias serían más bien de tipo cuantitativo, es decir, con relación al grado de participación de los diferentes mecanismos neurofuncionales implicados en el uso del lenguaje, entre ellos el conocimiento metalingüístico y la competencia lingüística implícita. Keywords: Neurolinguistics, Bilingualism, Aphasia, Language representation, Language lateralization, Palabras clave: Neurolingüística, Bilingüismo, Afasia, Representación del lenguaje, Lateralización del lenguaje

This review addresses language function and reorganization associated with various forms of epilepsy. Longstanding epilepsy, particularly types with onset early in life, may be associated with changes in the representation of language function in the brain. As a result of this reorganization, language function may be relatively spared despite injury to areas of the brain that normally subserve these functions. We examine the changes seen in language function in two types of epilepsy: hemispheric epilepsy of childhood and focal epilepsies. Findings from behavioral studies, intracarotid amytal testing, intraoperative cortical testing, and more recent functional imaging studies are reviewed. Studying changes in the representation of language function seen in some forms of epilepsy provides information about brain plasticity with implications for other neurologic diseases, as well as for the neuroscientific understanding of how and when functional reorganization may occur.

To address the impact of differences in language lateralization on joke comprehension, event-related brain potentials (ERPs) were recorded as 16 left- and 16 right-handed adults read one-line jokes and non-funny control stimuli (“A replacement player hit a home run with my girl/ball,”). In right-handers, jokes elicited a late positivity 500–900 ms post-stimulus onset that was largest over right hemisphere (RH) centro-parietal electrode sites, and a slow sustained negativity over anterior left lateral sites. In left-handers, jokes elicited a late positivity 500–900 ms post-onset that was larger and more broadly distributed than in the right-handers' ERPs. In right-handed women, the late positivity was larger over RH electrode sites. In left-handed women, the late positivity was bilaterally symmetric. The highly asymmetric slow sustained negativity over left anterior electrode sites was absent from left-handers' ERPs to jokes. Differences may reflect more efficient inter-hemispheric communication in the left-handers, as they are reputed to have relatively larger corpus callosal areas than right-handers. Results support the portrait of more bilateral language representation among left-handers, and suggest language lateralization affects high-level language comprehension tasks such as joke comprehension.

Determining the language-dominant hemisphere is essential for planning epilepsy surgery. A 60-year-old right-handed woman with epilepsy since age 16 failed a partial right anterior lobectomy at age 21. Later, a brain MRI found extensive right-sided cortical dysplasia and periventricular heterotopia. Subsequently, prolonged video-EEG monitoring localized her seizures to the right temporoparietal region. Functional MRI was inconclusive in lateralizing her language, prompting a Wada test, which strongly lateralized language to the right. This unique case of atypical language representation in a right-handed individual with an extensive right-hemispheric congenital malformation and seizure focus illustrates the important thorough presurgical language assessment.

Hemispheric language dominance was established for 368 epilepsy surgery candidates based on results of objective language performance during the intracarotid amobarbital procedure. Eighty-three percent of patients were found to be left hemisphere dominant, while six percent were right dominant, and eleven percent demonstrated some degree of bilateral language representation. Bilateral patients were grouped according to modality-specific patterns of language performance in each hemisphere, with a majority demonstrating left hemisphere superiority in over all language processing. Variables affecting atypical language development are reviewed and results are considered in the context of previous research.

Many studies have reported and analyzed the results of the intracarotid amobarbital procedure (IAP). Our study objective was to investigate bihemispheric language representation. We retrospectively reviewed the results of 165 IAPs performed in the evaluation of candidates for epilepsy surgery. Language function was assessed primarily by the duration of speech arrest. Three different sets of criteria were used: absolute duration, side-to-side difference, and laterality index. Only subjects whose dominance for language was classified as bilateral by two of the three methods were considered bilateral. There were 126 (76%) patients with language representation only in the left hemisphere, and 20 (12%) had language representation only in the right. Nineteen patients (12%) had evidence of speech in both hemispheres. Of these 19, 10 demonstrated return of speech soon after each injection; the remaining 9 patients demonstrated prolonged speech arrest after both injections. We conclude that two distinct patterns of bilateral language representation may exist: one in which each hemisphere is dependent on the other ("bilateral dependent") and one in which each hemisphere can support language independently ("bilateral autonomous").

Published 100 years ago, Freud's monograph on aphasia, Zur Auffassung der Aphasien, is an incisive analysis of the so-called diagram-makers, those aphasiologists who would reduce brain and language to simple schemes of circumscribed cortical centers linked by unidirectional subcortical pathways. Chief architects of the diagram-maker school were Wernicke and Lichtheim, and their formulations of conduction aphasia and the transcortical aphasias were particular targets of Freud's trenchant criticism. Freud argued against functional differentiation within left hemisphere perisylvian language regions, and he suggested that traditional aphasic syndromes were artifacts of lesions that included cortical language regions and deeper white matter pathways not directly involved with language. He provided an innovative classification of aphasia based on types of associations presumed to be affected. Freud, however, failed to adduce new clinical or pathological observations; he was perhaps unnecessarily harsh in challenging authorities in the field of aphasia; and his theory had to compete with a straightforward and already entrenched anatomical view of language representation. Despite Freud's lucid argumentation, his carefully crafted monograph was never widely noted.

The technique of direct stimulation mapping of the cortex is used to identify regions of language representation in the dominant cerebral hemosphere and the Rolandic cortex in either hemisphere. The use of electrocorticography to remove epileptogenic zones in patients with difficult-to-control seizures is described. The article presents the authors' experience with brain mapping during glioma surgery to maximize the extent of tumor resection and minimize surgical morbidity.

Five neuropsychological tests designed to measure the presence and severity of visual hemineglect were given to 101 patients with clearly denned unilateral lesions. Of these, visual hemineglect was present in 28 out of a total of 56 left brain damaged (LBD) patients and in 20 out of 45 right brain damaged (RBD) patients. The incidence of hemineglect in these groups did not differ significantly, but the severity of hemineglect was greater in the RBD group than in the LBD group. However, the two groups were found to differ significantly with respect to the loci of the lesions most likely to result in hemineglect. In the RBD group most patients with hemineglect had posterior lesions, and in the LBD group most patients with hemineglect had anterior lesions. Possible reasons for this were discussed first from the point of view of an inattention hypothesis and second in terms of the effects the lateralization of language representation might have on the representation of spatial functions.

The evidence from bilingual-crossed aphasia which has been used to support a bilingualism-right hemisphere effect for language representation and/or processing is critically reevaluated. Data gathered by the case survey method are shown to have been misrepresented in the literature. The argument based on a single group study of right-lesioned subjects is shown to rest upon a fundamental misinterpretation of those data. The consequences of such misrepresentation and misinterpretation as reflected in recent research are discussed.