State-of-the-art technologies in neural speech decoding utilize data collected from microwires or... more State-of-the-art technologies in neural speech decoding utilize data collected from microwires or microarrays implanted directly into the cerebral cortex. Yet as a tool accessible only to individuals with implanted electrodes, speech decoding from devices of this nature is severely limited in its implementation, and cannot be considered a viable solution for widespread application. Speech decoding from non-invasive EEG signals can achieve relatively high accuracy (70-80%), but only from very small classification tasks, with more complex tasks typically yielding a limited (20-50%) classification accuracy. We propose a novel combination of technologies in which transcranial magnetic stimulation (TMS) is first applied to augment the neural signals of interest, producing a greater signal-to-noise ratio in the EEG data. Next, delay differential analysis (DDA) – a cutting-edge computational method based on nonlinear dynamics – is implemented to capture the widest range of information avai...
ABSTRACTElectroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as ... more ABSTRACTElectroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-invasive measure of neural activity. With increased attention to EEG-based BCI systems, publicly available datasets that can represent the complex tasks required for naturalistic speech decoding are necessary to establish a common standard of performance within the BCI community. Effective solutions must overcome various kinds of noise in the EEG signal and remain reliable across sessions and subjects without overfitting to a specific dataset or task. We present two validated datasets (N=8 and N=16) for classification at the phoneme and word level and by the articulatory properties of phonemes. EEG signals were recorded from 64 channels while subjects listened to and repeated six consonants and five vowels. Individual phonemes were combined in different phonetic environments to produce coarticulated variation in forty consonant-vowel pairs, twenty real words, and twenty pseudowords...
One of the main goals in Brain-Computer Interface (BCI) research is to help patients with falteri... more One of the main goals in Brain-Computer Interface (BCI) research is to help patients with faltering communication abilities due to neurodegenrative diseases produce text or speech output using their neural recordings. However, practical implementation of such a system has proven difficult due to limitations in the speed, accuracy, and training time of existing interfaces. In this paper, we contribute to this endeavour by isolating appropriate input features from speech-producing neural signals that will feed into a machine learning classifier to identify target phonemes. Analysing data from six subjects, we discern frequency bands that encapsulate differential information regarding production of vowels and consonants broadly, and more specifically nasals and semivowels. Subsequent spatial local-ization analysis reveals the underlying cortical regions responsible for different phoneme categories. Anatomical locations along with their respective frequency bands act as prospective feature sets for machine learning classifiers. We demonstrate this classification ability in a preliminary language reconstruction task and show an average word classification accuracy of 30.6% (p<0.001).
Meta-analyses are a method by which to increase the statistical power and generalizability of neu... more Meta-analyses are a method by which to increase the statistical power and generalizability of neuroimaging findings. In the neurolinguistics literature, meta-analyses have the potential to substantiate hypotheses about L1 and L2 processing networks and to reveal differences between the two that may escape detection in individual studies. Why then is there so little consensus between the reported findings of even the most recently published and most highly powered meta-analyses? Limitations in the literature, such as the absence of a common method to define and measure descriptive categories (e.g., proficiency level, degree of language exposure, age of acquisition, etc.) are often cited. An equally plausible explanation lies in the technical details of how individual meta-analyses are conducted. This paper provides a review of recent meta-analyses, with a discussion of their methodological choices and the possible effect those choices may have on the reported findings.
Meta-analyses are a method by which to increase the statistical power and generalizability of neu... more Meta-analyses are a method by which to increase the statistical power and generalizability of neuroimaging findings. In the neurolinguistics literature, meta-analyses have the potential to substantiate hypotheses about L1 and L2 processing networks and to reveal differences between the two that may escape detection in individual studies. Why then is there so little consensus between the reported findings of even the most recently published and most highly-powered meta-analyses? Limitations in the literature, such as the absence of a common method to define and measure descriptive categories (e.g., proficiency level, degree of language exposure, age of acquisition, etc.) are often cited. An equally plausible explanation lies in the technical details of how individual meta-analyses are conducted. This paper provides a review of recent meta-analyses, with a discussion of their methodological choices and the possible effect those choices may have on the reported findings.
Electroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-in... more Electroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-invasive measure of neural activity. With increased attention to EEG-based BCI systems, publicly available datasets that can represent the complex tasks required for naturalistic speech decoding are necessary to establish a common standard of performance within the BCI community. Effective solutions must overcome various kinds of noise in the EEG signal and remain reliable across sessions and subjects without overfitting to a specific dataset or task. We present two validated datasets (N=8 and N=16) for classification at the phoneme and word level and by the articulatory properties of phonemes. EEG signals were recorded from 64 channels while subjects listened to and repeated six consonants and five vowels. Individual phonemes were combined in different phonetic environments to produce coarticulated variation in forty consonant-vowel pairs, twenty real words, and twenty pseudowords. Phoneme pairs and words were presented during a control condition and during transcranial magnetic stimulation targeted to inhibit or augment the EEG signal associated with specific articulatory processes.
Author(s): Comstock, Lindy B | Advisor(s): Schumann, John H | Abstract: Prosody and formulaic phr... more Author(s): Comstock, Lindy B | Advisor(s): Schumann, John H | Abstract: Prosody and formulaic phrases are phenomena that bridge gradient and categorical classification. They carry systematic linguistic meaning, but may also act as a pragmatic resource. While linguistic meaning is invariant, pragmatic resources tolerate idiosyncratic use, through which speaker intent is revealed. My dissertation investigates how second language and heritage speakers bridge this distinction between gradient and categorical implementations of prosody and formulaic phrases, challenging studies that predict prosody is one of the most difficult skills for second language learners to acquire and problematizing the assumptions of speech accommodation within intercultural interactions.Speech accommodation and sociolinguistic theory predict that when speakers affiliate, they will mirror socially salient features of their interlocutor's speech in their own production. Yet a speaker’s ability to accurately ...
This paper investigates what form journalistic questioning takes within the international press c... more This paper investigates what form journalistic questioning takes within the international press corps when representatives of different press systems work in close proximity. Within the U.S. context, adversarial questioning is valued as a key resource to ensure an independent press. Yet independent journalism is not universally upheld in media systems worldwide; Russian officials have explicitly criticized adversarial reporting by Western journalists in their coverage of Russian affairs. Questions posed to Russian Presidents Putin and Medvedev in G8/G20 press conferences 2000-2015 were assessed for two indicators: (a) initiative, and (b) critical content, with the aim to determine whether journalistic practice will converge in an international context, and whether the rise in adversarialness documented within U.S. presidential press conferences parallels a more general international phenomenon. Findings show a significant increase in both indicators over time and by presidential term. Questioning practices exhibited by Russian, non-Western, and Western journalists are discussed.
This paper investigates a new method for analyzing single-subject event-related potentials (ERPs)... more This paper investigates a new method for analyzing single-subject event-related potentials (ERPs) from pediatric epilepsy patients. The substantial inter-subject variability observed in the amplitude and latency of ERPs often complicates the analysis of subject-level results. We present a series of analyses that compare the single-subject results from an adult non-psychiatric subject and an adult schizophrenic patient-both of whom contributed to successful group-level findings-to the results of two epilepsy patients for whom no group-level analyses are available. We first reveal the absence of definitive results from traditional signal processing metrics for all clinical conditions. Then we illustrate how a nonlinear signal processing technique, delay differential analysis (DDA), may discern greater information about network-level processes in the brain of individuals prone to chaotic neural organization and psychopathological information processing, and how these network-level processes appear to confirm an underlying evoked potential that is compatible with the auditory deviant response (ADR). Differences in the findings observed for each epilepsy patient are discussed in light of their etiologies and the level of attention they displayed toward the task.
The functional organization of first (L1) and second (L2) language processing in bilinguals remai... more The functional organization of first (L1) and second (L2) language processing in bilinguals remains a topic of great interest to the neurolinguistics community. Functional magnetic resonance imaging (fMRI) studies report meaningful differences in the location and extent of hemodynamic changes between tasks performed in the L1 and L2, yet there is no consensus on whether these networks can be considered truly distinct. In part, this may be due to the multiplicity of task designs implemented in such studies, which complicates the interpretation of their findings. This paper compares the results of previous bilingual meta-analyses to a new ALE meta-analysis that categorizes neuroimaging studies by task design. Factors such as the age of L2 acquisition (AoA) and the L2 language proficiency level of participants are also considered. The findings support previous accounts of the effect of participant characteristics on linguistic processing, while at the same time revealing dissociable differences in fMRI activation for L1 and L2 networks within and across tasks that appear independent of these external factors.
State-of-the-art technologies in neural speech decoding utilize data collected from microwires or... more State-of-the-art technologies in neural speech decoding utilize data collected from microwires or microarrays implanted directly into the cerebral cortex. Yet as a tool accessible only to individuals with implanted electrodes, speech decoding from devices of this nature is severely limited in its implementation, and cannot be considered a viable solution for widespread application. Speech decoding from non-invasive EEG signals can achieve relatively high accuracy (70-80%), but only from very small classification tasks, with more complex tasks typically yielding a limited (20-50%) classification accuracy. We propose a novel combination of technologies in which transcranial magnetic stimulation (TMS) is first applied to augment the neural signals of interest, producing a greater signal-to-noise ratio in the EEG data. Next, delay differential analysis (DDA)-a cutting-edge computational method based on nonlinear dynamics-is implemented to capture the widest range of information available in the neural signal, by incorporating both linear and nonlinear dynamics.
This article investigates the intercultural competence and identities of under-graduate heritage ... more This article investigates the intercultural competence and identities of under-graduate heritage language learners (HLLs) studying abroad under the auspices of the U.S. Flagship Program. Based on survey responses and interviews with HLLs from Flagship Programs across the country, the article aims to assess the preparation of HLLs for study abroad and provides recommendations on the advisability of adopting an IC Level 3 standard that HLLs must meet prior to departure. We argue HLLs in study abroad programs require a specialized approach to achieve intercultural competence, with attention to cultural differentiation, development of an intercultural identity, and appreciation of the culturally-specific nature of language use.
Language plays a complex role in mediating implicit associations. This study shows that the degre... more Language plays a complex role in mediating implicit associations. This study shows that the degree to which these associations are a state-dependent or stable construct may be the product of life experiences. Mexican-American bilinguals were cued regarding the social acceptability of mixing languages ('codeswitching'). A subset of them-according to a latent profile analysis revealing five response patterns that significantly correlated with participant background characteristics-shifted or even reversed their implicit attitudes toward mixing languages. Thus, language does not "predetermine" bias, as previously claimed, but may evoke contradictory associations within the same individual, indexing shifting context-dependent relationships. The varied response to the same linguistic manipulation suggests that implicit attitudes reflect deeper conceptual processing rather than shallow linguistic associations.
1. DISSERTATION ABSTRACT Prosody and formulaic phrases are phenomena that bridge gradient and cat... more 1. DISSERTATION ABSTRACT Prosody and formulaic phrases are phenomena that bridge gradient and categorical classification. They carry systematic linguistic meaning, but may also act as a pragmatic resource. While linguistic meaning is invariant, pragmatic resources tolerate idiosyncratic use, through which speaker intent is revealed. My dissertation investigates how second language and heritage speakers bridge this distinction between gradient and categorical implementations of prosody and formulaic phrases, challenging studies that predict prosody is one of the most difficult skills for second language learners to acquire and problematizing the assumptions of speech accommodation within intercultural interactions. Speech accommodation and sociolinguistic theory predict that when speakers affiliate, they will mirror socially salient features of their interlocutor's speech in their own production. Yet a speaker's ability to accurately reproduce phonological phenomena may be linked to the critical period of language acquisition. Native-like articulation of prosody has been associated with age of acquisition, whereas the ability to learn lexical items continues to grow into adulthood. Thus, prosody and lexical items are theorized to differ in their degree of perceptual salience for late second language and heritage speakers. When attempting speech accommodation, the perceptual abilities of the two classes of speakers may render the former a preferred resource for heritage speakers, and the latter for late second language speakers. Political interviews often center around polarizing issues that evoke a display of stance through pragmatic cues. Therefore, this genre serves as an ideal setting for the study of intercultural speech accommodation. Russian-American political discourse shows how ready and able political
State-of-the-art technologies in neural speech decoding utilize data collected from microwires or... more State-of-the-art technologies in neural speech decoding utilize data collected from microwires or microarrays implanted directly into the cerebral cortex. Yet as a tool accessible only to individuals with implanted electrodes, speech decoding from devices of this nature is severely limited in its implementation, and cannot be considered a viable solution for widespread application. Speech decoding from non-invasive EEG signals can achieve relatively high accuracy (70-80%), but only from very small classification tasks, with more complex tasks typically yielding a limited (20-50%) classification accuracy. We propose a novel combination of technologies in which transcranial magnetic stimulation (TMS) is first applied to augment the neural signals of interest, producing a greater signal-to-noise ratio in the EEG data. Next, delay differential analysis (DDA) – a cutting-edge computational method based on nonlinear dynamics – is implemented to capture the widest range of information avai...
ABSTRACTElectroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as ... more ABSTRACTElectroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-invasive measure of neural activity. With increased attention to EEG-based BCI systems, publicly available datasets that can represent the complex tasks required for naturalistic speech decoding are necessary to establish a common standard of performance within the BCI community. Effective solutions must overcome various kinds of noise in the EEG signal and remain reliable across sessions and subjects without overfitting to a specific dataset or task. We present two validated datasets (N=8 and N=16) for classification at the phoneme and word level and by the articulatory properties of phonemes. EEG signals were recorded from 64 channels while subjects listened to and repeated six consonants and five vowels. Individual phonemes were combined in different phonetic environments to produce coarticulated variation in forty consonant-vowel pairs, twenty real words, and twenty pseudowords...
One of the main goals in Brain-Computer Interface (BCI) research is to help patients with falteri... more One of the main goals in Brain-Computer Interface (BCI) research is to help patients with faltering communication abilities due to neurodegenrative diseases produce text or speech output using their neural recordings. However, practical implementation of such a system has proven difficult due to limitations in the speed, accuracy, and training time of existing interfaces. In this paper, we contribute to this endeavour by isolating appropriate input features from speech-producing neural signals that will feed into a machine learning classifier to identify target phonemes. Analysing data from six subjects, we discern frequency bands that encapsulate differential information regarding production of vowels and consonants broadly, and more specifically nasals and semivowels. Subsequent spatial local-ization analysis reveals the underlying cortical regions responsible for different phoneme categories. Anatomical locations along with their respective frequency bands act as prospective feature sets for machine learning classifiers. We demonstrate this classification ability in a preliminary language reconstruction task and show an average word classification accuracy of 30.6% (p<0.001).
Meta-analyses are a method by which to increase the statistical power and generalizability of neu... more Meta-analyses are a method by which to increase the statistical power and generalizability of neuroimaging findings. In the neurolinguistics literature, meta-analyses have the potential to substantiate hypotheses about L1 and L2 processing networks and to reveal differences between the two that may escape detection in individual studies. Why then is there so little consensus between the reported findings of even the most recently published and most highly powered meta-analyses? Limitations in the literature, such as the absence of a common method to define and measure descriptive categories (e.g., proficiency level, degree of language exposure, age of acquisition, etc.) are often cited. An equally plausible explanation lies in the technical details of how individual meta-analyses are conducted. This paper provides a review of recent meta-analyses, with a discussion of their methodological choices and the possible effect those choices may have on the reported findings.
Meta-analyses are a method by which to increase the statistical power and generalizability of neu... more Meta-analyses are a method by which to increase the statistical power and generalizability of neuroimaging findings. In the neurolinguistics literature, meta-analyses have the potential to substantiate hypotheses about L1 and L2 processing networks and to reveal differences between the two that may escape detection in individual studies. Why then is there so little consensus between the reported findings of even the most recently published and most highly-powered meta-analyses? Limitations in the literature, such as the absence of a common method to define and measure descriptive categories (e.g., proficiency level, degree of language exposure, age of acquisition, etc.) are often cited. An equally plausible explanation lies in the technical details of how individual meta-analyses are conducted. This paper provides a review of recent meta-analyses, with a discussion of their methodological choices and the possible effect those choices may have on the reported findings.
Electroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-in... more Electroencephalography (EEG) holds promise for brain-computer interface (BCI) devices as a non-invasive measure of neural activity. With increased attention to EEG-based BCI systems, publicly available datasets that can represent the complex tasks required for naturalistic speech decoding are necessary to establish a common standard of performance within the BCI community. Effective solutions must overcome various kinds of noise in the EEG signal and remain reliable across sessions and subjects without overfitting to a specific dataset or task. We present two validated datasets (N=8 and N=16) for classification at the phoneme and word level and by the articulatory properties of phonemes. EEG signals were recorded from 64 channels while subjects listened to and repeated six consonants and five vowels. Individual phonemes were combined in different phonetic environments to produce coarticulated variation in forty consonant-vowel pairs, twenty real words, and twenty pseudowords. Phoneme pairs and words were presented during a control condition and during transcranial magnetic stimulation targeted to inhibit or augment the EEG signal associated with specific articulatory processes.
Author(s): Comstock, Lindy B | Advisor(s): Schumann, John H | Abstract: Prosody and formulaic phr... more Author(s): Comstock, Lindy B | Advisor(s): Schumann, John H | Abstract: Prosody and formulaic phrases are phenomena that bridge gradient and categorical classification. They carry systematic linguistic meaning, but may also act as a pragmatic resource. While linguistic meaning is invariant, pragmatic resources tolerate idiosyncratic use, through which speaker intent is revealed. My dissertation investigates how second language and heritage speakers bridge this distinction between gradient and categorical implementations of prosody and formulaic phrases, challenging studies that predict prosody is one of the most difficult skills for second language learners to acquire and problematizing the assumptions of speech accommodation within intercultural interactions.Speech accommodation and sociolinguistic theory predict that when speakers affiliate, they will mirror socially salient features of their interlocutor's speech in their own production. Yet a speaker’s ability to accurately ...
This paper investigates what form journalistic questioning takes within the international press c... more This paper investigates what form journalistic questioning takes within the international press corps when representatives of different press systems work in close proximity. Within the U.S. context, adversarial questioning is valued as a key resource to ensure an independent press. Yet independent journalism is not universally upheld in media systems worldwide; Russian officials have explicitly criticized adversarial reporting by Western journalists in their coverage of Russian affairs. Questions posed to Russian Presidents Putin and Medvedev in G8/G20 press conferences 2000-2015 were assessed for two indicators: (a) initiative, and (b) critical content, with the aim to determine whether journalistic practice will converge in an international context, and whether the rise in adversarialness documented within U.S. presidential press conferences parallels a more general international phenomenon. Findings show a significant increase in both indicators over time and by presidential term. Questioning practices exhibited by Russian, non-Western, and Western journalists are discussed.
This paper investigates a new method for analyzing single-subject event-related potentials (ERPs)... more This paper investigates a new method for analyzing single-subject event-related potentials (ERPs) from pediatric epilepsy patients. The substantial inter-subject variability observed in the amplitude and latency of ERPs often complicates the analysis of subject-level results. We present a series of analyses that compare the single-subject results from an adult non-psychiatric subject and an adult schizophrenic patient-both of whom contributed to successful group-level findings-to the results of two epilepsy patients for whom no group-level analyses are available. We first reveal the absence of definitive results from traditional signal processing metrics for all clinical conditions. Then we illustrate how a nonlinear signal processing technique, delay differential analysis (DDA), may discern greater information about network-level processes in the brain of individuals prone to chaotic neural organization and psychopathological information processing, and how these network-level processes appear to confirm an underlying evoked potential that is compatible with the auditory deviant response (ADR). Differences in the findings observed for each epilepsy patient are discussed in light of their etiologies and the level of attention they displayed toward the task.
The functional organization of first (L1) and second (L2) language processing in bilinguals remai... more The functional organization of first (L1) and second (L2) language processing in bilinguals remains a topic of great interest to the neurolinguistics community. Functional magnetic resonance imaging (fMRI) studies report meaningful differences in the location and extent of hemodynamic changes between tasks performed in the L1 and L2, yet there is no consensus on whether these networks can be considered truly distinct. In part, this may be due to the multiplicity of task designs implemented in such studies, which complicates the interpretation of their findings. This paper compares the results of previous bilingual meta-analyses to a new ALE meta-analysis that categorizes neuroimaging studies by task design. Factors such as the age of L2 acquisition (AoA) and the L2 language proficiency level of participants are also considered. The findings support previous accounts of the effect of participant characteristics on linguistic processing, while at the same time revealing dissociable differences in fMRI activation for L1 and L2 networks within and across tasks that appear independent of these external factors.
State-of-the-art technologies in neural speech decoding utilize data collected from microwires or... more State-of-the-art technologies in neural speech decoding utilize data collected from microwires or microarrays implanted directly into the cerebral cortex. Yet as a tool accessible only to individuals with implanted electrodes, speech decoding from devices of this nature is severely limited in its implementation, and cannot be considered a viable solution for widespread application. Speech decoding from non-invasive EEG signals can achieve relatively high accuracy (70-80%), but only from very small classification tasks, with more complex tasks typically yielding a limited (20-50%) classification accuracy. We propose a novel combination of technologies in which transcranial magnetic stimulation (TMS) is first applied to augment the neural signals of interest, producing a greater signal-to-noise ratio in the EEG data. Next, delay differential analysis (DDA)-a cutting-edge computational method based on nonlinear dynamics-is implemented to capture the widest range of information available in the neural signal, by incorporating both linear and nonlinear dynamics.
This article investigates the intercultural competence and identities of under-graduate heritage ... more This article investigates the intercultural competence and identities of under-graduate heritage language learners (HLLs) studying abroad under the auspices of the U.S. Flagship Program. Based on survey responses and interviews with HLLs from Flagship Programs across the country, the article aims to assess the preparation of HLLs for study abroad and provides recommendations on the advisability of adopting an IC Level 3 standard that HLLs must meet prior to departure. We argue HLLs in study abroad programs require a specialized approach to achieve intercultural competence, with attention to cultural differentiation, development of an intercultural identity, and appreciation of the culturally-specific nature of language use.
Language plays a complex role in mediating implicit associations. This study shows that the degre... more Language plays a complex role in mediating implicit associations. This study shows that the degree to which these associations are a state-dependent or stable construct may be the product of life experiences. Mexican-American bilinguals were cued regarding the social acceptability of mixing languages ('codeswitching'). A subset of them-according to a latent profile analysis revealing five response patterns that significantly correlated with participant background characteristics-shifted or even reversed their implicit attitudes toward mixing languages. Thus, language does not "predetermine" bias, as previously claimed, but may evoke contradictory associations within the same individual, indexing shifting context-dependent relationships. The varied response to the same linguistic manipulation suggests that implicit attitudes reflect deeper conceptual processing rather than shallow linguistic associations.
1. DISSERTATION ABSTRACT Prosody and formulaic phrases are phenomena that bridge gradient and cat... more 1. DISSERTATION ABSTRACT Prosody and formulaic phrases are phenomena that bridge gradient and categorical classification. They carry systematic linguistic meaning, but may also act as a pragmatic resource. While linguistic meaning is invariant, pragmatic resources tolerate idiosyncratic use, through which speaker intent is revealed. My dissertation investigates how second language and heritage speakers bridge this distinction between gradient and categorical implementations of prosody and formulaic phrases, challenging studies that predict prosody is one of the most difficult skills for second language learners to acquire and problematizing the assumptions of speech accommodation within intercultural interactions. Speech accommodation and sociolinguistic theory predict that when speakers affiliate, they will mirror socially salient features of their interlocutor's speech in their own production. Yet a speaker's ability to accurately reproduce phonological phenomena may be linked to the critical period of language acquisition. Native-like articulation of prosody has been associated with age of acquisition, whereas the ability to learn lexical items continues to grow into adulthood. Thus, prosody and lexical items are theorized to differ in their degree of perceptual salience for late second language and heritage speakers. When attempting speech accommodation, the perceptual abilities of the two classes of speakers may render the former a preferred resource for heritage speakers, and the latter for late second language speakers. Political interviews often center around polarizing issues that evoke a display of stance through pragmatic cues. Therefore, this genre serves as an ideal setting for the study of intercultural speech accommodation. Russian-American political discourse shows how ready and able political
Uploads
Papers by Lindy Comstock