An fMRI investigation of covertly and overtly produced mono- and multisyllabic words.
Shuster LI, Lemieux SK.
Brain and Language 93 (2005):20-31.
The Insula: what is it? An forgotten island
of cortex hiding behind the lateral fissure
Ill-defined functions may include Visceral sensory Vestibular Motor Supplementary motor
Speech movements
Background: Speech and the Insula
Left anterior insular damage linked to acquired apraxia of speech (AOS) (Dronkers, 1996)
AOS characterized as, "a disorder in the motor planning of articulatory movements”
Lesion-overlap method 25 chronic stroke patients
with AOS 19 patients non-AOS
aphasics Hence, “this area seems to
be specialized for the motor planning of speech“
AOS Non-AOS
Background: Speech and the Insula Insular damage-AOS correlation does
not appear to hold for acute-stage stroke patients (Hillis et al., 2004)
Insula currently thought to be involved in motor movements of speech E.g. coordination of muscle movements
rather than planning (Ackermann & Riecker, 2004)
Overview Goal:
Examine the role of the insula in overt and covert speech production
Methods: fMRI Blood Oxygenation Level Dependent (BOLD)
responses Overtness contrast Syllabic-length contrast
Logic: Overtness contrast should identify general overt
speech areas Syllable-length contrast should show specific areas
with graded activation in motor planning of utterances
Shuster and Lemieux (2005)
Behavioral task: Participants (n=10) auditorily
prompted to overtly or covertly “say” individual words 30 monosyllabic nouns 30 tetrasyllabic nouns Words frequency-matched à la
Thorndike and Lorge (1944)
(No additional behavioral tasks; no behavioral measures taken)
Shuster and Lemieux (2005): Methods
Repeat the word to yourself as quickly as you can, without moving any part of your mouth. Just hear
yourself saying the word inside your head. Be careful not to move any part of your mouth and try not to swallow during the response time
Thorndike and Lorge (1944) and Kucera and Francis (1967) frequency ratings for non-zero frequency one- and four- syllable words.
R2 = 0.6511
0
50000
100000
150000
200000
250000
300000
350000
0 1000 2000 3000 4000 5000 6000 7000 8000
kucera and francis (1967)
thor
ndik
e an
d lo
rge
(194
4)
But do either of these criticisms really matter here?
-Length effects may be difficult to interpret.
Criticisms of Materials Length likely confounded with
concreteness Thorndike-Lorge?
Criticisms of the tasks Image-naming or stem-completion
would have been a better task Phonological vs. articulatory rehearsal
Participants may have been ‘sleeping’ in covert trials A behavioral measure was needed
Covert-subtraction-based differences difficult to interpret
Shuster and Lemieux (2005): Methods
Design details: Event-related design
Allows isolation of speech/motion artifacts Requires long ISIs (1250 ms!)
Overt and covert response blocks of 12.5 minutes each All words presented in each block Randomized word orders Block order counterbalanced between subjects
Shuster and Lemieux (2005): Methods
Imaging details: 1.5T scanner Anatomical scans
.86 x .86 x 1.2mm voxels Functional scans
20 interleaved axial slices acquired 3.43 x 3.43 x 5 mm voxels No baseline
BOLD responses analyzed via AFNI Subtraction methods for overtness and length
Shuster and Lemieux (2005): Methods
An interesting analysis detail… “the stimulus time
series was convolved with a gamma variate function”
Translation: speech-motion artifacts removed by correlating observed data with an idealized ‘motion artifact’ function
Overview of results More overall activity in overt than
covert speech Included more activity in general
cognitive regions More overall activity in multisyllabic
than monosyllabic word production
Shuster and Lemieux (2005): Results
Shuster and Lemieux (2005):Results
Overtness effects
Covert > Overt
Overt > Covert
Mono- vs. Multisyllable effects
Mono- > Multisyllabic
Multi- > Monosyllabic
Shuster and Lemieux (2005):Results
Left Insula
More active in overt speech
No more active with longer words
Active in speech production but maybe not in sequencing speech movements
Shuster and Lemieux (2005): Discussion
Left inferior parietal lobule (and the left parietal cortex in general)
More active with longer (overt) words
Consistent with previous production and apraxia studies
Indicates involvement in word-length dependent processes (e.g. articulatory planning, sequencing, monitoring)
Shuster and Lemieux (2005): Discussion
Left Superior Temporal Sulcus (STS) More active in overt
speech No more active with
longer words Activity may reflect
perception of self-generated speech Confounded by
auditory stimulus presentation
Shuster and Lemieux (2005): Discussion
Conclusions: Covert and overt speech produced
qualitatively different response patterns, including more activation of general cognitive regions, suggesting that covert production may not be a good substitute for overt production
Shuster and Lemieux (2005)
Conclusions: The left insula is active in overt
speech production, but shows no utterance-length-dependent effects, suggesting a revised role in overt speech production
The left parietal cortex, however, does show more length-related activity
Shuster and Lemieux (2005)
Questions for discussion Any thoughts on the role of the insula in
speech? Monitoring in the left inferior parietal cortex? What might have been happening in the
“inner speech condition? How might this change our interpretation of overt
vs. covert results? Occipital activation?
Shuster and Lemieux (2005)
“Clearly, as Bennett and Netsell (1999) noted, further studies are required before the specific role of the left insula in speech production can be established”
--Shuster and Lemieux (2005)