The Neuroscience of Language
What is language? What is it for?
• Rapid efficient communication
– (as such, other kinds of communication might be
called language for our purposes and might share
underlying neural mechanisms)
• Two broad but interacting domains:
– Comprehension
– Production
Speech comprehension
• Is an auditory task
– (but stay tuned for the McGurk Effect!)
• Is also a selective attention task
– Auditory scene analysis
• Is a temporal task
– We need a way to represent both frequency
(pitch) and time when talking about language ->
the speech spectrogram
Speech comprehension
• Is also a selective
attention task
– Auditory scene analysis
• Which streams of sound
constiutute speech?
• Which one stream
constitutes the to-becomprehended speech
• Not a trivial problem
because sound waves
combine prior to reaching
the ear
Speech comprehension
• Is a temporal task
– Speech is a time-varying signal
– It is meaningless to freeze a word in time (like you
can do with an image)
– We need a way to consider both frequency (pitch)
and time when talking about language -> the
speech spectrogram
What forms the basis of spoken
language?
• Phonemes
• Phonemes strung together over time with
prosody
What forms the basis of spoken
language?
• Phonemes = smallest perceptual unit of sound
• Phonemes strung together over time with
prosody
What forms the basis of spoken
language?
• Phonemes = smallest perceptual unit of sound
• Phonemes strung together over time with
prosody = the variation of pitch and loudness
over the time scale of a whole sentence
What forms the basis of spoken
language?
• Phonemes = smallest perceptual unit of sound
• Phonemes strung together over time with
prosody = the variation of pitch and loudness
over the time scale of a whole sentence
To visualize these we need slick acoustic analysis
software…which I’ve got
What forms the basis of spoken
language?
• The auditory system is inherently tonotopic
Is speech comprehension therefore an
image matching problem?
• If your brain could just match the picture on the basilar
membrane with a lexical object in memory, speech would be
comprehended
Problems facing the brain
•Acoustic - Phonetic invariance
–says that phonemes should match one and only one pattern in the
spectrogram
–This is not the case! For example /d/ followed by different vowels:
Problems facing the brain
• The Segmentation Problem:
– The stream of acoustic input is not physically segmented into discrete
phonemes, words, phrases, etc.
– Silent gaps don’t always indicate (aren’t perceived as) interruptions in
speech
Problems facing the brain
• The Segmentation Problem:
– The stream of acoustic input is not physically segmented into discrete
phonemes, words, phrases, etc.
– Continuous speech stream is sometimes perceived as having gaps
How (where) does the brain solve these
problems?
– Note that the brain can’t know that incoming
sound is speech until it first figures out that it isn’t
!?
– Signal chain goes from non-specific -> specific
– Neuroimaging has to take the same approach to
track down speech-specific regions
Functional Anatomy of Speech
Comprehension
• low-level auditory pathway is not
specialized for speech sounds
• Both speech and non-speech
sounds activate primary auditory
cortex (bilateral Heschl’s Gyrus)
on the top of the superior
temporal gyrus
Functional Anatomy of Speech
Comprehension
• Which parts of the auditory pathway are specialized for
speech?
• Binder et al. (2000)
– fMRI
– Presented several kinds of stimuli:
•
•
•
•
•
white noise
These have non-word-like acoustical properties
pure tones
non-words
These have word-like acoustical properties but no
reversed words
lexical associations
real words word-like acoustical properties and lexical associations
Functional Anatomy of Speech
Comprehension
• Relative to “baseline” scanner
noise
– Widespread auditory cortex
activation (bilaterally) for all
stimuli
– Why isn’t this surprising?
Functional Anatomy of Speech
Comprehension
• Statistical contrasts reveal specialization for speech-like
sounds
– superior temporal gyrus
– Somewhat more prominent on left side
Functional Anatomy of Speech
Comprehension
• Further highly sensitive contrasts to identify specialization for
words relative to other speech-like sounds revealed only a
few small clusters of voxels
• Brodmann areas
– Area 39
– 20, 21 and 37
– 46 and 10
Next time we’ll discuss
• Speech production
• Aphasia
• Lateralization