Current Research Projects
Articulatory Compensation and Preservation Following Partial Glossectomy
Individuals diagnosed with lingual cancer oftentimes undergo partial glossectomy, whereby part of the tongue is surgically removed. This treatment typically gives rise to difficulties in speech production. This real-time MRI project (i) quantifies and compares lingual flexibility in typical and post-glossectomy speech using Principal Component Analysis (PCA); (ii) provides evidence for compensatory articulation in consonant production by investigating patterns exhibited in constriction location and degree; and (iii) compares patients' data to the synthesized output of a two-condition Task Dynamics model to provide evidence for or against articulatory compensation in vowel production.
Collaborators:
Shri Narayanan (USC Engineering)
Krishna Nayak (USC Engineering)
Louis Goldstein (USC Linguistics)
Yijing Lu (USC Linguistics)
Uttam Sinha (USC - Keck School of Medicine)
Non-native Sequence Learning [To participate, click here!]
While typical speakers produce consonant clusters that are phonotactically licit in their native language (e.g., /sn/, for a speaker of English) in somewhat predictable ways, the strategies used to produce consonant clusters that are not phonotactically licit (e.g., /fn/, for a speaker of English) are less understood. The findings of this electromagnetic articulography study help to characterize motor learning of complex speech sequences and the fine details of relative timing associated with their production.
Collaborators:
Adam Buchwald (NYU)
Hung-Shao Cheng (NYU)
Matthew Masapollo (UFL)
Articulatory Coordination in Individuals with Parkinson's Disease [To participate, click here!]
Individuals with Parkinson's disease have long been observed to exhibit hypokinetic dysarthria, affecting various kinds of somatic movement (including gait and speech), whereby the speed and magnitude of movements is reduced. Individuals with Parkinson's disease have also been observed to exhibit coordination difficulties in both limb movement and among laryngeal and supralaryngeal speech movements. This study uses electromagnetic articulography to investigate whether deficits in supralaryngeal articulatory coordination contribute to the perception of imprecise articulation and uncontrolled acoustic duration of speech sounds in individuals with Parkinson's disease.
Articulatory Kinematic Markers of Childhood Apraxia of Speech [To participate, click here!]
Childhood Apraxia of Speech (CAS) is a developmental motor speech disorder that affects the planning and programming of speech motor commands in a target sequence, and is characterized by the production of inconsistent errors across repetitions of the same word or phrase, impaired coarticulation patterns, and prosodic errors. This study uses electromagnetic articulography to identify differences in the kinematic and coordination patterns exhibited by individuals with CAS and by typical speakers.
Machine Intelligence for Cognitive and Communication Disorders
For individuals with communication disorders, speech and language intervention is provided by speech language pathologists. However, at present, most therapy programs and materials are limited to the English language and are therefore not accessible to multilingual societies such as India. This also brings forth a need for developing computational techniques to develop assistive devices for communication disorders focusing on challenges specific to India, given its diverse languages and cultures. This project aims to develop tools to deduce behavioral informatics from multi-modal signals which are representatives of verbal and nonverbal attributes. Such quantification of impaired speech and language characteristics can not only be used for detecting and diagnosing the severity of the disorder, but also to quantifiably monitor progress over the course of speech and language intervention.
Collaborators:
Prasanta Ghosh (Indian Institute of Science)
Shri Narayanan (USC)
B.K. Yamini (Indian Institute of Science)
Aravind Illa (Indian Institute of Science)
Individuals diagnosed with lingual cancer oftentimes undergo partial glossectomy, whereby part of the tongue is surgically removed. This treatment typically gives rise to difficulties in speech production. This real-time MRI project (i) quantifies and compares lingual flexibility in typical and post-glossectomy speech using Principal Component Analysis (PCA); (ii) provides evidence for compensatory articulation in consonant production by investigating patterns exhibited in constriction location and degree; and (iii) compares patients' data to the synthesized output of a two-condition Task Dynamics model to provide evidence for or against articulatory compensation in vowel production.
Collaborators:
Shri Narayanan (USC Engineering)
Krishna Nayak (USC Engineering)
Louis Goldstein (USC Linguistics)
Yijing Lu (USC Linguistics)
Uttam Sinha (USC - Keck School of Medicine)
Non-native Sequence Learning [To participate, click here!]
While typical speakers produce consonant clusters that are phonotactically licit in their native language (e.g., /sn/, for a speaker of English) in somewhat predictable ways, the strategies used to produce consonant clusters that are not phonotactically licit (e.g., /fn/, for a speaker of English) are less understood. The findings of this electromagnetic articulography study help to characterize motor learning of complex speech sequences and the fine details of relative timing associated with their production.
Collaborators:
Adam Buchwald (NYU)
Hung-Shao Cheng (NYU)
Matthew Masapollo (UFL)
Articulatory Coordination in Individuals with Parkinson's Disease [To participate, click here!]
Individuals with Parkinson's disease have long been observed to exhibit hypokinetic dysarthria, affecting various kinds of somatic movement (including gait and speech), whereby the speed and magnitude of movements is reduced. Individuals with Parkinson's disease have also been observed to exhibit coordination difficulties in both limb movement and among laryngeal and supralaryngeal speech movements. This study uses electromagnetic articulography to investigate whether deficits in supralaryngeal articulatory coordination contribute to the perception of imprecise articulation and uncontrolled acoustic duration of speech sounds in individuals with Parkinson's disease.
Articulatory Kinematic Markers of Childhood Apraxia of Speech [To participate, click here!]
Childhood Apraxia of Speech (CAS) is a developmental motor speech disorder that affects the planning and programming of speech motor commands in a target sequence, and is characterized by the production of inconsistent errors across repetitions of the same word or phrase, impaired coarticulation patterns, and prosodic errors. This study uses electromagnetic articulography to identify differences in the kinematic and coordination patterns exhibited by individuals with CAS and by typical speakers.
Machine Intelligence for Cognitive and Communication Disorders
For individuals with communication disorders, speech and language intervention is provided by speech language pathologists. However, at present, most therapy programs and materials are limited to the English language and are therefore not accessible to multilingual societies such as India. This also brings forth a need for developing computational techniques to develop assistive devices for communication disorders focusing on challenges specific to India, given its diverse languages and cultures. This project aims to develop tools to deduce behavioral informatics from multi-modal signals which are representatives of verbal and nonverbal attributes. Such quantification of impaired speech and language characteristics can not only be used for detecting and diagnosing the severity of the disorder, but also to quantifiably monitor progress over the course of speech and language intervention.
Collaborators:
Prasanta Ghosh (Indian Institute of Science)
Shri Narayanan (USC)
B.K. Yamini (Indian Institute of Science)
Aravind Illa (Indian Institute of Science)