functional roles of broca s area and smg: evidence from...

fstcuptgBstbetltial

gce

lciitsg

dP9

NeuroImage 10, 570–581 (1999)Article ID nimg.1999.0499, available online at http://www.idealibrary.com on

1CA

Functional Roles of Broca’s Area and SMG:Evidence from Cortical Stimulation Mapping in a Deaf Signer

David P. Corina,*,1 Susan L. McBurney,† Carl Dodrill,‡ Kevin Hinshaw,§ Jim Brinkley,¶ and George Ojemann\

*Department of Psychology, †Department of Linguistics, ‡Harborview Medical Center, §Department of Computer Science,¶Department of Biological Structure, and \Department of Neurosurgery, University of Washington, Seattle, Washington 98195

Received April 19, 1999

1esfimgaotsarobtgawsttt

c(ptsbpwTmcweola

The importance of the left hemisphere in languageunction has been firmly established and current worktrives to understand regional specializations withinhe perisylvian language areas. This paper reports aase study of a deaf user of American Sign Languagendergoing an awake cortical stimulation mappingrocedure. Patterns of sign errors accompanying elec-rical stimulation of Broca’s area and the supramar-inal gyrus (SMG) are reported. Our findings showroca’s area to be involved in the motor execution ofign language. These data demonstrate that the linguis-ic specificity of Broca’s area is not limited to speechehavior. In addition, unusual semantic–phonologicalrrors were observed with stimulation to the SMG;hese data may implicate the SMG in the binding ofinguistic features in the service of language produc-ion. Taken together, these findings provide importantnsight into the linguistic specificity of Broca’s areand the functional role of the supramarginal gyrus inanguage processing. r 1999 Academic Press

Key Words: American Sign Language; deafness; lan-uage localization; Broca’s area; supramarginal gyrus;ortical stimulation mapping (CSM); temporal lobepilepsy

INTRODUCTION

The role of the left hemisphere perisylvian areas inanguage function has been firmly established by de-ades of clinical research involving subjects who havencurred acute brain damage. Within the past 10 yearst has become clear that this left hemisphere specializa-ion for language is observed not only for users ofpoken languages, but for deaf users of signed lan-uages as well (Poizner et al., 1987; Hickok, et al.,

1 To whom correspondence and reprint requests should be ad-ressed at Laboratory for Cognitive Neuropsychology, Department ofsychology, 135 Guthrie Hall, University of Washington, Seattle WA

h8195. Fax: (206) 685-3157. E-mail: [email protected].

570053-8119/99 $30.00opyright r 1999 by Academic Pressll rights of reproduction in any form reserved.

996a; Corina, 1998). More recent work strives tolucidate regional specialization within the left hemi-phere perisylvian regions in relation to the multipleunctional systems involved in human language process-ng (e.g., phonological encoding, lexical access, gram-

atical parsing, motor implementation). In this re-ard, lesion studies have significant limitations; lesionsre often large and not circumscribed to specific regionsf interest and often impact several processing sys-ems, further obscuring functional–anatomical relation-hips. Thus, there is a growing emphasis on usinglternative techniques to ascertain specific corticalegional specialization within the left hemisphere. Somef these techniques include functional neuroimaging oflood flow (Petersen et al., 1988; Binder et al., 1997),he measurement of electrical and magnetic fieldsenerated by the brain (Osterhout, 1994; Lounasmaa etl., 1996), and electrical stimulation of the cortex,hich produces short-lasting reversible functional le-

ions (Flitman et al., 1998; Ojemann, 1995). Theseechniques are beginning to provide novel insight intohe complex functional–anatomical relationships withinhe left hemisphere systems subserving language.

The present paper reports data from a very rarelinical case of a deaf user of American Sign LanguageASL) undergoing an awake cortical stimulation map-ing (CSM) procedure during left temporal lobe resec-ion for treatment of a seizure disorder. Followingtandard techniques, temporal lobe resection is guidedy electrophysiological and cortical stimulation map-ing findings obtained during a part of the operation inhich the patient is awake but under local anesthesia.he CSM procedure identifies sites of language andotor function through application of a localized electri-

al current at specific cortical sites. Sites of stimulationithin the left-temporal, inferior-frontal, and parietalxposure are pseudo-randomly selected for the purposef identifying cortical regions responsible for motor andanguage function. During the mapping of motor areas,

subject is observed for movements of the face, jaw,
and, etc., and is asked to report any sensation he/she

etjwdr

ddActctrictpsvcg

F

ti(dssl1amfBrtcdaigcrrrc

cltlt

ottttcb

F

hsthlar1ttMgeoelao(estBstpima1btpal

Suotmttt

571FUNCTIONAL ROLES OF BROCA’S AREA AND SMG

xperiences under conditions of stimulation. Duringhe language-mapping portion of the procedure, a sub-ect may be required to name pictures or read writtenords. Disruption of the ability to perform the taskuring stimulation is taken as evidence of corticalegions integral to the language task (Whitaker, 1998).In the present case a standard motor mapping proce-

ure was used. However, the language mapping proce-ure was adapted for sign language assessment, asSL was this subject’s primary and preferred mode ofommunication. The subject (patient S.T.) was requiredo perform two tasks: name (in sign) line drawings ofommon objects and repeat signs presented on video-ape. Stimulation to several sites in the perisylvianegion evoked object naming disruption, and two sitesn particular resulted in repeated disruption withonsistent characteristics. These sites, an anterior fron-al site located within Broca’s area, and a posteriorarietal opercular site within the supramarginal gyrus,erve as the main focus of this paper. The data providealuable insight into the functional specificity of Bro-a’s area and the functional role of the supramarginalyrus (SMG).

unctional Roles of Broca’s Area

Broca’s area lies in the frontal opercular region and isypically defined as the posterior third of the leftnferior frontal gyrus, encompassing Brodmann’s areasBA) 44 and 45. Traditional accounts of aphasics withamage to Broca’s area describe patients with effortfulpeech production but with intact language comprehen-ion. However, the precise role of Broca’s area inanguage behaviors remains controversial (Mohr et al.,978; Goodglass, 1993; Zatorre et al., 1996). For ex-mple, the anatomical proximity of Broca’s area tootor and sensory cortex of the Rolandic fissure has

ueled speculation that speech difficulties evidenced inroca’s aphasics may simply reflect sensorimotor dis-uption of the speech articulators rather than a linguis-ic deficit per se. Moreover, patients described as Bro-a’s aphasics are known to have not only speech fluencyifficulties, but subtle language comprehension deficitss well (Goodglass, 1993). Consistent with these find-ngs, recent imaging studies have demonstrated re-ional subspecialization of Broca’s area. Price andolleagues have shown that while posterior frontalegions BA 44 and BA 6 (premotor mouth and faceegions) are active in tasks requiring repetition andeading aloud, the anterior extent (BA 45) is active inases of single word perception (Price et al., 1996).The current study provides valuable insight into the

ontribution of the posterior portion of Broca’s area toanguage function. To date, no studies have evaluatedhe role of Broca’s area in users of manual-gesturalanguages with the precision afforded by the CSM
echnique. Examining whether functional disruptions i
f sign language are observed during electrical stimula-ion to Broca’s area provides an opportunity to assesshe functional specificity of this region. The data ob-ained in the present study are pertinent to the ques-ion of whether this region’s function relates specifi-ally to speech behavior or more generally to languageehavior, both spoken and signed.

unctional Roles of Supramarginal Gyrus

The parietal lobe is a polymodal association area thatas been implicated in a large range of motor, sensory,patial, and attention functions. However, the func-ional specializations within this large cortical areaave been difficult to discern. The supramarginal gyrus

ies within the parietal opercular area and is identifieds the convolution that caps the posterior ascendingamus of the sylvian fissure (BA 40) (Steinmetz et al.,990). Early studies have suggested the importance ofhis association area in spatial behaviors, particularlyhose involving praxis and attention (Benton, 1969).ore recent studies have implicated the supramar-

inal gyrus in a variety of language-related behaviors,specially those involving the phonological componentf language structure (D’emonet et al., 1994). Forxample, speech errors have been observed with stimu-ation of the SMG in studies of oral naming (Ojemann etl., 1989), and aphasics with involvement of the SMGften show deficits in acoustic–phonetic processingCaplan et al., 1995). In addition, there is compellingvidence that the SMG plays an important role in theelection and/or activation of phonemic sequences inhe service of language-related tasks (Kertesz, 1993;ell et al., 1990). Interestingly, some studies haveuggested that the SMG may play a broader role inasks requiring the confluence of hand movements andhonemic selection. For example, the SMG has beenmplicated in some forms of acquired writing impair-

ents, specifically phonological agraphia (Penniello etl., 1995; Roeltgen and Heilman, 1984; Alexander et al.,992). The contributions of the SMG to sign languageehavior are not well understood. However, the inven-ory of functions associated with the SMG would, ariori, flag this anatomical convergence zone as servingn important function in the programming of manual-inguistic behaviors.

The disparate functions currently attributed to theMG are brought together in the present study in annusual and informative manner. Because the domainf interest, sign language, is a primary language sys-em that requires complex sequences of hand move-ents, this study provides an opportunity to examine

he interaction between linguistic, praxic, and atten-ional mechanisms involved in language. Furthermore,he anatomical specificity afforded by our image coreg-
stration techniques (Modayur et al., 1997) provides

ga

S

aradpssaic

aPopms

W

t(shmthoorn

C

2s6dipsItton1S

tbuee

L

gcr2banAadaet

tahSbtasact

A

iFAbbgmpak

b2pFiGpc

572 CORINA ET AL.

reat refinement in the localization of function associ-ted with this parietal opercular region.

METHODS

ubject Characteristics

Subject S.T. is a 50-year-old, right-handed male withprofound hearing loss (left ear, right ear . 90 dB) as a

esult of spinal meningitis at 18 months of age. Hettended day programs and residential schools for theeaf, has an 8th grade education, and has been em-loyed as a maintenance worker. WAIS-R performancecale testing estimates an I.Q. of 81. S.T. is a fluentigner of American Sign Language. ASL is his preferrednd only mode of effective communication; S.T.’s speechs functionally unintelligible, and he makes no effort toommunicate vocally.Complex partial seizures with some secondary gener-

lization began at approximately 38 years of age.reoperative evaluation identified left temporal onsetf S.T.’s medically refractory seizures. MRI scan andostoperative histological analysis of tissues revealededial temporal hippocampal sclerosis and cystic le-

ion in the hilus of the hippocampus.

ada Testing

Wada testing procedures adapted for ASL were usedo established memory and language lateralizationDodrill and Ojemann, 1997). On memory measures,odium amobarbital injections to the left and rightemispheres resulted in borderline and impaired perfor-ance, respectively, indicating bilateral memory func-

ions. In contrast, during language testing only the leftemisphere injection resulted in sign-blockage andbject naming errors; there were no language errorsbserved with injections to the right hemisphere. Theseesults are consistent with a left-hemisphere domi-ance for ASL in this subject.

ortical Stimulation Mapping

The cortex was mapped with 4-s trains of 60-Hz,.5-ms biphasic square waves from a constant currenttimulator delivered through electrodes 5 mm apart at

mA, the largest current that did not evoke after-ischarges from the sampled cortex. Language testtems were presented at 4-s intervals by either sliderojector or videotape. Stimulation occurred on everyecond or third item, with no site repeated successively.n the language mapping portion of the procedure, aotal of 24 sites were examined. The total number ofrials (stimulated and nonstimulated) was 106 forbject naming and 53 for sign repetition. The meanumber of stimulated trials per site was 4.6 (range–12) for naming and 2.8 (range 1–6) for repetition.
igned responses during the procedure were video- s
aped and scored with respect to their preoperativeaseline measures. Authors D.C. and S.M. (both fluentsers of ASL) and a native deaf research assistantvaluated signed responses with respect to well-stablished principles of ASL linguistic structure.

anguage Testing Procedures

Stimuli used for the object naming portion of lan-uage mapping were line drawings of 49 items that S.T.onsistently named with one-handed signs. For theepetition task, a deaf signer was videotaped producing0 signs and 20 nonsigns. Nonsigns were constructedy recombining existing ASL hand shapes, movements,nd locations into novel, sign-like constructions. Theonsign forms did not violate phonotactic constraints ofSL and may be considered the equivalent of pronounce-ble nonwords (e.g., ‘‘nust’’). Thus, the articulatoryemands that these nonsign forms place on the signerre identical to those present with actual signs. How-ver, the nonsigns are without lexical–semantic con-ent.

S.T. underwent extensive preoperative practice andesting with all stimuli. During preoperative sessions,ll stimuli were practiced with both left and rightands. Due to the constraints of the surgical procedure,.T. was instructed that in the operating room he woulde responding with his left hand. It should be notedhat handedness is not lexically contrastive in ASL. Inddition, routinely in the course of daily activities, deafigners often use their nondominant hand for signing,s in the case of carrying a bag of groceries or holding ahild, etc., while conversing; thus signing with eitherhe right or the left hand is not considered unnatural.

natomical Localization

During the CSM procedure, sites of stimulation weredentified by sterile number tickets laid on the cortex.ollowing testing, these locations were photographed.three-dimensional model of S.T.’s brain was created

y combining a T-1-weighted 3-D MR volume of therain with a 2-D MR venogram and a 3-D MR arterio-ram. Coregistration of the intraoperative numberarkers onto this three-dimensional model was accom-

lished through reference to patterns of cortical veinsnd arteries (Modayur et al., 1997; Hinshaw and Brin-ley, 1997).Imaging parameters were as follows: 3-D structural

rain image, SPGR (29/5/1/45°) (TR/TE/NEX/flip angle),2-cm FOV, 256 3 192 matrix, 124 1.2-mm sagittalartitions; 2-D TOF MR venogram (45/9/1/60°), 22-cmOV, 256 3 192 matrix, 100 contiguous 1.5-mm axial

mages; 3-D MOTSA MR angiogram (RAMP excitation,E Medical Systems, Milwaukee, WI), four overlap-ing slabs of 16 partitions each (36/6.9/1/25°), flowompensation, 22-cm FOV, 256 3 256 matrix, 64 axial
lices at 0.9-mm spacing.

M

rjwhTsRcc

L

seo

O

(

drenoslnta

tosm(ttara

ic

r


RESULTS

otor/Sensory Mapping

Stimulation of the cortical sites numbered 1 and 2esulted in mouth and lip movements, site 3 elicitedaw movement, and sensation in the lips was reportedith stimulation of sites 4 and 6. Sensation of the rightand was elicited with stimulation of site 5 (Fig. 1).here were no motor or sensory responses at any otherite. This mapping establishes the location of theolandic cortex, with its familiar pattern of sensoryortex lying posterior to motor cortex, and face/mouthortex lying inferior to hand representation.

anguage Mapping

Stimulation of several sites within the left hemi-phere exposure led to significant left-handed signingrrors. The anatomical locations of the sites and naturef the resulting errors are discussed below.

bject Naming

Nine of 23 sites tested led to object naming errorsTable 1). Two sites were particularly prone to naming

FIG. 1. Three-dimensional reconstruction of S.T.’s brain, show
epresentation.
isruption. An isolated frontal opercular site (B) evokedepeated object naming disruption. A site in the pari-tal opercular region (Po) also resulted in robust objectaming errors. Moderate naming disruption was alsobserved in sites surrounding site Po (Fig. 2). Occa-ional errors were also observed in the midtemporalobe region. No object naming errors occurred underonstimulation conditions. Importantly, the nature ofhe errors at the frontal opercular, parietal opercular,nd midtemporal lobe sites was qualitatively different.As seen in Fig. 1, site B lies at the posterior portion of

he third left frontal convolution, immediately in frontf sites evoking face motor responses. This site corre-ponds to the posterior aspect of Broca’s area, Brod-ann area (BA) 44. Site Po is the supramarginal gyrus

BA 40), as confirmed by an in-depth planar reconstruc-ion. S.T.’s parietal opercular area is consistent withhe anatomical type 1 pattern described in Steinmetz etl. (1990). Midtemporal sites stimulated included supe-ior and middle temporal gyri and correspond to BA 22nd BA 21.Stimulation of anterior site B resulted in errors

nvolving the motor execution of signs. These errors areharacterized by a laxed articulation of the intended

sites of motor (1, 2, 3), sensory (4, 5, 6), and language (B, Po)
ing

srtssfiasrlcstSmiswnf

qaiairswtUcplmtwefmoPt‘bhwsfianifu

l

574 CORINA ET AL.

ign, with nonspecific movements (repeated tapping orubbing) and a reduction in hand shape configurationso a laxed-closed fist hand shape. For example, withouttimulation S.T. correctly signed COW with a ‘‘Y’’ handhape (outstretched thumb and little finger, all otherngers closed), with the thumb anchored on the templend two downward twists of the wrist (Fig. 3a). Duringtimulation of site B, S.T.’s attempt at signing COWesulted in a closed fist hand shape with a repeatedaxed tapping of this incorrect hand shape at theheekbone (Fig. 3c). Without stimulation, S.T. correctlyigned SHIRT with the fingertips of the ‘‘F’’ hand shapeouching the chest twice (Fig. 3b). Under stimulation,.T. instead produced a closed fist hand shape andade small rubbing motions at the chest (Fig. 3d). As

llustrated by these examples, during stimulation ofite B the articulatory integrity of the intended signsas consistently compromised. Interestingly, there waso effort on the part of S.T. to self-correct these imper-ect forms.

TABLE 1

Language Mapping Results

Object naming Sign repetition

Site/ocation

Stimulustrials Errors

Site/location

Stimulustrials Errors

Frontal operculum

B 8 7 B 3 332a 5 0 32a 2 034 4 0 34 2 0

Parietal operculum

P0 12 10 P0 4 030s 2 1 30s 3 031 7 2 31 2 033 7 2 33 2 1

Posterior temporal lobe

27 3 0 27 2 027a 4 0 27a —27p 1 0 27p —28 3 0 28 2 029 7 2 29 3 029a 2 0 29a —

Midtemporal lobe

22 5 1 22 4 022p — 22p 1 023 5 1 23 6 024 4 1 24 3 025 6 0 25 4 025a 1 0 25a —25p 1 0 25p —26 5 0 26 2 026p 5 0 26p 2 0

Anterior temporal lobe

20 4 0 20 2 021 5 0 21 4 0
e
The sign errors observed with stimulation of Po areualitatively different. S.T. produced both formationalnd semantic errors. Formational errors are character-zed by repeated attempts by the subject to distinctlyrticulate the intended targets; often his signing exhib-ted successive formational approximations of the cor-ect sign. For example, the sign PEANUT is normallyigned with a closed fist and outstretched thumb, andith a movement composed of an outward wrist rota-

ion (the thumb flicking off the front of the teeth).nder stimulation, this sign began as an incorrect, but

learly articulated, ‘‘X’’ hand shape (closed fist with arotruding bent index finger) articulated at the correctocation, but with an incorrect inward rotation move-

ent. In two successive attempts to correct this error,he subject first corrected the hand shape and thenent on to correct the movement as well. A secondxample is illustrated in Figs. 4a and 4b. In its correctorm, the sign SCISSORS is iconic; the index andiddle fingers emulate the action of scissor blades

pening and closing (Fig. 4a). Under stimulation to siteo, S.T.’s attempts to correctly articulate the sign take

he following form (Fig. 4b): he starts with the correct‘V’’ hand shape but produces no scissoring movement,ends the fingers of this hand shape, switches to a ‘‘Y’’and shape, and executes a movement with a repeatedrist twist, switches back to the correct ‘‘V’’ hand

hape, but then proceeds to incorrectly bend these twongers downward repeatedly rather than ‘‘scissor’’ thems required, and ultimately, he gives up. Notably, we doot find the laxed and reduced articulations character-

stic of signing under conditions of stimulation torontal site B. Instead, as these examples illustrate,nder stimulation to site Po, the subject’s signing

FIG. 2. Object naming stimulation results.

xhibits problems involving the selection of the indi-

vmS

tlssssstosimS

admtsaob

otesafita

s nd

f


idual components of sign forms (i.e., hand shape,ovement, and to a lesser extent, location) (Corina andandler, 1993).Semantic errors were also observed under stimula-

ion to site Po, and the form of these errors is particu-arly noteworthy. Specifically, all of these errors involveemantic substitutions that are formationally quiteimilar to the intended targets. For example, thetimulus picture ‘‘pig’’ elicited the sign FARM, thetimulus picture ‘‘bed’’ was signed as SLEEP, and thetimulus picture ‘‘horse’’ was signed as COW. In ASL,hese semantic errors contain considerable formationalverlap with their intended targets; for example, theigns PIG and FARM differ in movement, but share andentical articulatory location (the chin) and each are

ade with similar hand shapes; the signs BED andLEEP share hand shape and are both articulated

FIG. 3. Site B object naming responses. (a) Correct articulationtimulation to site B, S.T.’s inaccurate articulation of the sign COW. (d) U

FIG. 4. Site Po object naming responses. (a) Correct articulat
ormationally inaccurate articulation of the sign SCISSORS.
bout the face; finally, the signs COW and HORSEiffer only in hand shape. In English these mistakesight be similar to uttering ‘‘lobster’’ when one in-

ended to say ‘‘oyster,’’ or ‘‘plane’’ when one intended toay ‘‘train’’; that is, these errors share both semanticnd formational properties. As discussed below, theccurrence of these unusual semantic–formationallends is of significant theoretical interest.Formational and semantic errors were occasionally

bserved at three sites proximal to site Po. Stimulationo sites 29, 30s, and 31 resulted in formational selectionrrors, while stimulation to sites 29 and 31 resulted in aemantic–formational error (ELEPHANT for RAT) and

semantic intrusion (CAT (correct) followed by thengerspelling of #D-O-G (intrusion)). Finally, at site 33wo additional errors were observed: in one error S.T.ppears to be groping to articulate the lexical sign

he sign COW. (b) Correct articulation of the sign SHIRT. (c) Underer stimulation to site B, S.T.’s inaccurate articulation of the sign SHIRT.

of the sign SCISSORS. (b) Under stimulation to site Po, S.T.’s

of t

ion

Ss

i(tGmr

efillTsiet

soafFfsPnmsp

S

fsf

576 CORINA ET AL.

UN; a second error resulted in a perseverative re-ponse from the previous trial.Three sites in the midtemporal region produced

solated naming errors. Stimulation of sites 23 and 22superior and middle temporal gyri) resulted in seman-ic substitutions (DRINK-LIQUOR for MILK, and CI-AR for CIGARETTE). Stimulation to site 24 (inferioriddle temporal gyrus) resulted in a perseverative

esponse from the previous trial.In summary, the analysis of these object naming

rrors shows that, following stimulation to anteriorrontal site B, we find an impairment in the globalmplementation of signs. Overall, these errors showedaxed articulation (not unlike ‘‘mumbling’’ in spokenanguage) and were not corrected by the signer (seeable 2). In contrast, stimulation at parietal opercularite Po resulted in errors affecting the selection of thendividual formational components of a sign, as well asrrors involving semantic–formational blends. Charac-eristically, successive approximations of the target

TAB

Summary of Naming Errors That Occ

ign were attempted. The movements and hand shapesbserved during these attempts tended to be complexnd varied (in contrast to the reduced inventory oforms observed with stimulation to B) (see Table 3).rom these findings it appears that stimulation to the

rontal site B has a global effect on the motor output ofigning, whereas stimulation to parietal opercular siteo disrupts the correct selection of the linguistic compo-ents (including both phonological and semantic ele-ents) required in the service of naming. Infrequent

emantic naming errors were also observed in midtem-oral lobe regions.

ign Repetition

The data from the sign repetition tasks permiturther insight into the specificity of function at corticalites B and Po. Strikingly, stimulation to anteriorrontal site B produced reliable errors in sign repetition

2

red with Stimulation to Broca’s Area

LE

ur

(spw

tsddsnst

clr

wnlnadpnantnrsop


3/3). No errors were observed in sign repetition withtimulation to any of the temporal lobe or posterior-arietal opercular sites. An isolated repetition erroras observed at site 33 (see Fig. 5).The sign repetition errors resulting from stimulation

o site B evidenced a laxed and imprecise articulationimilar to that observed in the naming task. For example,uring nonstimulated trials, the sign PUZZLED was pro-uced with an index finger hand shape that bends at theecond joint as it is brought toward the bridge of theose (Fig. 6a). Under stimulation to site B, the handhape is laxed and the movement is incorrect, reducedo rubbing and tapping on the nose (Fig. 6b).

The data from the nonsign repetition tasks areonsistent with the sign repetition data. During stimu-ation to site B, S.T. made clear errors on nonsign

TAB

Summary of Naming Errors That Occurred wit

epetition trials, while no nonsign repetition errors e

ere observed with stimulation to site Po. Unfortu-ately, during the intraoperative procedure S.T. ‘‘regu-

arized’’ many of the nonsign targets (i.e., he viewed aonsign trial and produced a formationally similarctual ASL sign). These performance errors occurreduring both stimulated and nonstimulated trials. Theresence of these regularization errors limited theumber of valid nonsign repetition data points avail-ble for analysis. Conservatively, scoring only thoseonsign trials in which S.T. correctly preformed theask, we found that stimulation of site B resulted in 4/4onsign repetition errors, while stimulation to site Poesulted in no errors (though we were limited to only aingle trial at Po using the available data). A clusteringf single nonsign repetition errors was also found atosterior temporal sites 27, 28, and 29 and a single

3

timulation to the Supramarginal Gyrus (SMG)

LE

h S

rror at anterior temporal site 21.

sooaactti

hcaoa

auca

L

Botoawl

oiegiosapAtuWsBtatsc

r

578 CORINA ET AL.

Focusing our discussion on the two regions thathowed robust and repeated naming errors, B and Po,ur data provide further evidence that the frontalpercular site is involved in the execution of the motoricspects of signs and is agnostic as to whether sign formsre derived from self-generation (as in naming) or fromopying another signer’s rendition of these forms. Thoughhe nonsign repetition data are limited, it would appearhat the motor output affects complex sign formationndependent of lexical–semantic content.

In contrast, the posterior opercular site appearsighly specialized; disruption was found only in theontext of a task in which S.T. was required to translatepicture of an object into a lexical sign name for that

bject. When the formational components of the signre available to the subject (as in the case of copying

FIG. 5. Sign repetition stimulation results.

FIG. 6. Site B sign repetition responses. (a) Correct repetition o
epetition of the sign PUZZLED.
nother individual’s sign and nonsign forms), evennder stimulation to site Po, the subject is able toombine these observed sign elements into coherentnd accurate gestures for reproduction.

DISCUSSION

inguistic Specificity of Broca’s Area

In our study, stimulation to the posterior portion ofroca’s area consistently affected the articulatory integrityf the sign and resulted in laxed and imperfect formation ofhe underlying sign target. The effects of motor executionn language output were observed for both object namingnd sign repetition tasks. In addition, nonsign repetitionas also compromised, thus ruling out the effects of

exical–semantic content in the observed disruptions.Our results are consistent with the characterization

f the posterior portion of Broca’s area as participatingn the motoric execution of complex articulatory forms,specially those underlying the phonetic level of lan-uage structure. Significantly, our sign language find-ngs provide new insights into the functional specificityf posterior Broca’s region. As noted, researchers haveuggested that speech difficulties associated with dam-ge to Broca’s area may be related to the anatomicalroximity of face and mouth cortex (Goodglass, 1993).s such, it is particularly informative that we observed

hese errors in sign language, a language that makesse of the hands and arms as the primary articulators.hen we examine the stimulation map for motor and

ensory function in this subject, it is evident that whileroca’s area shares the expected anatomical proximity

o mouth cortex, the site of hand representation isnatomically distant (Fig. 1). Thus, it is remarkablehat stimulation of Broca’s area leads to impairment inign language execution, as these sign language errorsannot be accounted for by the proximity of Broca’s area

e sign PUZZLED. (b) Under stimulation to site B, S.T.’s inaccurate
f th

tcsmasfiwfi

icdtaadfweott

almrbftm

tfcl(cgcradrrMiefsfw

wd

oiwdmttf

pfgmca

S

tptssaia

wcsrdddosaw

tti(teawws

fhttt


o the hand-motor representations within the Rolandicortex. Taken together with previous findings, our datauggest that Broca’s area may be participating in aspects ofotor control that underlie the complex movements of the

rticulators used in the service of language, bothpoken and signed. Though lacking the regional speci-city of the present study, case studies of deaf signersith lesions including Broca’s area corroborate thisnding (Poizner et al., 1987; Hickok et al., 1996b).It is interesting to note that recent functional imag-

ng studies in hearing individuals have reported Bro-a’s area activation during tasks outside the linguisticomain, specifically during tasks involving the percep-ion of hand movements (Shulag et al., 1994; Decety etl., 1997; Grafton et al., 1996). For example, Decety etl. (1997) reported significant activation of BA 45uring a PET study in which subjects watched meaning-ul pantomime. However, activation was not observedhen subjects watched meaningless gestures. Grafton

t al. (1996) reported activation of BA 45 when subjectsbserved the grasping of common objects, and activa-ion of BA 44 was prominent when subjects were askedo imagine themselves grasping objects.

These findings have led some to argue that Broca’srea function includes representational capacities re-ated to action/recognition of oro-facial and brachio-

anual behaviors (Rizzolatti and Arbib, 1998). Ouresults might be viewed as being consistent with thisroader, more general characterization of Broca’s areaunction. However, we believe the role of verbal media-ion during imaging tasks involving observation ofeaningful gestures requires further evaluation.While the involvement of Broca’s area in the percep-

ion of gestures has been explored, the contribution ofrontal opercular regions to the motor production ofomplex hand movements has not been well estab-ished. In an early PET imaging study, Frith et al.1991) reported activation in dorso-lateral prefrontalortex that was attributed to ‘‘willed-action.’’ This re-ion was found to be active in cases when a subjecthose to move one of two fingers on the right handelative to a stimulus specified finger movement. Inddition, this general region was found to be activeuring a verbal generation task relative to a verbalepetition task. Frith et al. (1991) interpreted thisegion as a modality-independent locus of willed action.ore recent studies have questioned the modality

ndependence of these data (Kapur et al., 1994; Hydert al., 1997). For example, Hyder et al. (1997), usingMRI, have reported a locus of activity for a willedensorimotor task (i.e., finger movement) in the middlerontal gyrus (BA 46) and a separate locus (BA 45) for ailled verbal task (i.e., verbal fluency).It is unlikely that the production problems observedith stimulation of Broca’s area in this subject reflect a
eficit in the initiation of action. Anatomically, the locus a
f stimulation that disrupts signing is posterior andnferior to middle frontal gyrus activations observed inilled sensorimotor tasks (see Hyder et al., 1997 foriscussion). In addition, rather than a cessation ofotor activity, S.T. produces goal directed movements

o locations associated with the correct target sign,hough, as reported, the articulatory details of the signormation are compromised.

Finally, it is noteworthy that the deficits in signroduction are evidenced in the left signing handollowing stimulation to the left frontal opercular re-ion; thus these output errors do not reflect primaryotor or sensory deficits. Taken together these data

learly implicate a role for Broca’s area in left-handrticulations of linguistic movement in this deaf signer.

MG and Linguistic Feature Binding

In our study we localized stimulation of the SMG tohe inferior portion of the sulcus lying between theosterior ascending ramus of the sylvian fissure andhe inferior portion of the postcentral sulcus. Undertimulation to this portion of the SMG, our subjecthowed great selectional difficulties with hand posturesnd movements associated with individual signs. Onemportant question concerns whether these errors arepraxic or aphasic in nature.Left-hemisphere parietal lesions often are associatedith limb apraxia—the impairment of the ability to

arry out voluntary movement in the absence of sen-ory loss, paresis, or motor weakness. There is a wideange of symptomology that is associated with apraxicisorders and the literature is replete with inconsistentefinitions. Apraxic symptoms can range from frankisorders of object use (which often fall under the rubricf ideational apraxia) to extremely subtle impairments,uch as the inability to imitate meaningless gesturesnd hand postures (an impairment often associatedith ideomotor apraxia).In the present case of S.T., there are several indica-

ions that the deficits in signing observed with stimula-ion to the SMG reflect linguistic rather than apraxicmpairment. For example, imitation of meaninglessand meaningful) movements is often used as a diagnos-ic measure of the ideomotor form of apraxia (De Renzit al., 1980). However, S.T. did not show impairments whensked to imitate nonsign or sign gestures—impairmentsere observed only during a task in which object namingas required. This precludes an analysis of these errors as

temming from an ideomotor apraxic deficit.In the context of the naming task the predominant

ormational errors involve selectional difficulties withand postures and movements. It is important to notehat the hand shapes and movements that comprisehese errors are chosen from the inventory of forma-ional elements of American Sign Language. These
ffected formational elements (hand shape and move-

msot1mpata

slhs1smiocwttanstcps

itnlbtlla

pftlts(wsapopt

tpbw

tttrppgvac(CsagTlvoa

tamfvot

tad

A

B

B

B

580 CORINA ET AL.

ent) correspond to two central parameters of ASLign phonology and can be characterized as collectionsf distinctive features within a phonological representa-ion (Corina and Sandler, 1993; Chomsky and Halle,968). These errors are not random selections of move-ents or hand shapes, but are constrained by the

honotactics of ASL. Taken together these data supportn interpretation of the deficit following stimulation tohe SMG as indicative of a linguistic, rather than anpraxic, impairment.The finding that ASL selectional difficulties are ob-

erved with stimulation of the SMG is interesting inight of recent imaging and behavioral studies thatave implicated parietal corticies in motor responseelection (Rushworth et al., 1997; Deiber et al., 1991,996). Rushworth et al. (1997), for example, haveuggested that impairment in selection of motor actionsay underlie the sequencing deficits characteristic of

deational apraxics. Ideational apraxics may misorder,mit, or perseverate components of an action that isomposed of a sequence of movements; for example,hen asked to post a letter, a patient may properly fold

he letter, but then seal the envelope before having puthe letter inside. The proper completion of a complexction requires careful orchestration of the subcompo-ents of the overall task. In the case of S.T., withtimulation to the SMG we observe selectional difficul-ies involving the phonological subcomponents of aomplex linguistic action. Thus, the SMG may bearticipating in a highly specialized form of responseelection, one that is specific to language function.In addition to the incorrect selection of phonological

nformation, we observed semantic errors. Importantly,hese semantic errors demonstrated considerable pho-ological overlap with their intended targets. In spoken

anguage speech error data, semantic–phonologicallends have been of significant theoretical interest inhat they have been analyzed as evidence for a model ofexical representations in which semantic and phono-ogical stages of lexical processing interact (Martin etl., 1996; Dell and O’Seaghdha, 1992).Taken together our data indicate that the SMG may

lay a critical role in the selection of phonologicaleature information and the association of this informa-ion with semantic representations in the service ofanguage production. Lexical–semantic representa-ions, like phonological representations, are often under-tood as collections of primitive semantic featuresKatz, 1972). Additionally, a stage of processing inhich lexical–semantic information (e.g., lemma repre-

entations) and phonological information are associ-ted is well attested in influential models of languageroduction (Garret, 1998; Levelt, 1989). The errorsbserved in S.T.’s signing during stimulation to thisarietal opercular region may indicate a disjunction in
he binding of lexical–semantic and phonological fea-
ures. Sign language semantic and phonological para-hasias (including semantic–formational blends) haveeen reported in case studies of aphasic deaf signersith lesions that include the SMG (Corina, 1998).It is interesting to note that the anatomical locus of

he SMG, the area that has given rise to these disjunc-ions in sign language, lies just above the posteriorhird of the superior temporal sulcus, a region ofteneferred to as Wernicke’s area. A recent MEG study oficture naming has implicated Wernicke’s area (inarticular the posterior third of the superior temporalyrus) and the tempo-parietal junction as being in-olved in phonological encoding for spoken languagend possibly participating in temporally adjacent pro-esses (e.g., lemma selection, articulatory encoding)Levelt et al., 1998). The convergence of the currentSM study and the MEG study is notable; both reveal atage of processing whereby phonological informationnd semantic information are associated during lan-uage production in response to a picture naming task.he question of whether the differences in anatomical

ocation implicated across these two studies (i.e., SMGersus Wernicke’s area) reflect normal variation, meth-dological differences, or language modality differenceswaits further study.

CONCLUSION

The unusual case of S.T. provides important datahat help to constrain and foster interpretation of thenatomical–functional relationships underlying hu-an language. Our findings suggest that, as we seek to

urther specify the general functional nature of perisyl-ian language areas, it will be worthwhile to make usef the differences in human language as tools to clarifyhe regional specificities common to all languages.

ACKNOWLEDGMENTS

We acknowledge the following individuals for their contributions tohis project: Nat Wilson and Deba Ackerman (stimulus developmentnd data analysis); Ecttore Lettich and Ramon Romero (videoocumentation); and Dr. Lea Bachman (WADA procedure).

REFERENCES

lexander, M. P., Friedman, R. B., Loverso, F., and Fischer, R. S. 1992.Lesion localization of phonological agraphia. Brain Lang. 43:83.

ell, W. L., Horner, J., Logue, P., and Radtke, R. A. 1990. Neologisticspeech automatisms during complex partial seizures. Neurology 40:49.

enton, A. L. 1969. Disorders of spatial orientation. In Handbook ofClinical Neurology (P. Vincken and G. Bruyn, Eds.), Vol. 3, pp.212–228. North-Holland, Amsterdam.

inder, J. R., Frost, J. A., Hammeke, T. A., Cox, R. W., Rao, S. M., andPrieto, T. 1997. Human brain language areas identified by func-
tional magnetic resonance imaging. J. Neurosci. 17:353–362.

C

C

C

C

D

D

D

D

D

D

D

F

F

G

G

G

H

H

H

H

K

KK

L

L

L

M

M

M

O

O

O

P

P

P

P

R

R

R

S

S

W

Z


aplan, D., Gow, D., and Makris, N. 1995. Analysis of lesions by MRIin stroke patients with acoustic-phonetic processing deficits. Neu-rology 45:293.

homsky, N., and Halle, M. 1968. The Sound Patterns of English.Harper & Row, New York.

orina, D. P. 1998. Aphasia in users of signed languages. In Aphasiain Atypical Populations (P. Coppens, Y. Lebrun, and A. Basso, Eds.),pp. 261–309. Erlbaum, Hillsdale, NJ.

orina, D. P., and Sandler, W. 1993. On the nature of phonologicalstructure in sign language. Phonology 10(2):165–207.ecety, J., Grezes, J., Costes, N., Perani, D., Jeannerod, M., Procyk,E., Grassi, F., and Fazio, F. 1997. Brain activity during observationof actions: Influence of action content and subject’s strategy. Brain120:1763–1777.eiber, M. P., Passingham, R. E., Colebatch, J. G., Friston, K. J.,Nixon, P. D., and Frackowiak, R. S. J. 1991. Cortical areas and theselection of movement: A study with positron emission tomography.Exp. Brain Res. 84:393–402.eiber, M. P., Ibanez, V., Sadato, N., and Hallet, M. 1996. Cerebralstructures participating in motor preparation in humans: A posi-tron emission tomography study. J. Neurophysiol. 75:233–246.ell, G. S., and O’Seaghdha, P. G. 1992. Stages of lexical access inlanguage production. Cognition 42:287–314.’emonet, J. F., Price, C., Wise, R., and Frackowiak, R. S. 1994. A PETstudy of cognitive strategies in normal subjects during languagetasks: Influence of phonetic ambiguity and sequence processing onphoneme monitoring. Brain 117:671.e Renzi, E., Motti, F., and Nichelli, P. 1980. Imitating gestures.Arch. Neurol. 37:6–10.odrill, C. B., and Ojemann, G. A. 1997. An exploratory comparison ofthree methods of memory assessment with the intracarotid amobar-bital procedure. Brain Cogn. 33(2):210–223.

litman, S. S., Grafman, J., Wassermann, E. M., Cooper, V., O’Grady,J., Pascual-Leone, A., and Hallett, M. 1998. Linguistic processingduring repetitive transcranial magnetic stimulation. Neurology50(1):175–181.

rith, C. D., Friston, K., Liddle, P. F., and Frackowiak, R. S. J. 1991.Willed action and the prefrontal cortex in man: A study with PET.Proc. R. Soc. London Ser. B 244:241–246.arret, M. 1988. Processes in language production. In Linguistics:The Cambridge Survey, Volume IV, Psychological and BiologicalAspects (F. J. Newmeyer, Ed.), pp. 69–96. Cambridge Univ. Press,Cambridge, MA.oodglass, H. 1993. Understanding Aphasia. Academic Press, SanDiego, CA.rafton, S. T., Arbib, M. A., Fadiga, L., and Rizzolatti, G. 1996.Localization of grasp representations in humans by positronemission tomography 2: Observation compared with imagination.Exp. Brain Res. 112:103–111.ickok, G., Bellugi, U., and Klima, E. S. 1996a. The neurobiology ofsign language and its implications for the neural basis of language.Nature 381:699–702.ickok, G., Kritchevsky, M., Bellugi, U., and Klima, E. S. 1996b. Therole of the left frontal operculum in sign language aphasia.Neurocase 2:373–380.inshaw, K. P., and Brinkley, J. F. 1997. Using 3-D shape models toguide segmentation of MR brain images. In Proceedings Annual FallSymposium, American Medical Informatics Association, pp. 469–473.Hanley and Belfus, Inc., Philadelphia, PA.yder, F., Phelps, E. A., Wiggins, C. J., Labar, K. S., Blamire, A. M.,and Shulman, R. G. 1997. ‘‘Willed action’’: A functional MRI studyof the human prefrontal cortex during a sensorimotor task. Proc.
Natl. Acad. Sci. USA 94:6989–6994.
apur, S., Rose, R., Liddle, P. F., Zipursky, R. B., Brown, G. M., Stuss,D., Houle, S., and Tulving, E. 1994. Neuroreport 5:2193–2196.atz, J. 1972. Semantic Theory. Harper and Row, New York.ertesz, A. 1993. Clinical forms of aphasia. Acta Neurochirurg.Suppl. 56:52.

evelt, W. J. 1989. The speaker as information processor. In Speaking:From Intention to Articulation, pp. 1–28. MIT Press, Cambridge, MA.

evelt, W. J. M., Praamstra, P., Myers, A. S., Helenius, P., andSalmelin, R. 1998. A MEG study of picture naming. J. Cogn.Neurosci. 10:553–567.

ounasmaa, O. V., Hamalainen, M., Hari, R., and Salmelin, R. 1996.Information processing in the human brain: Magnetoencephalo-graphic approach. Proc. Natl. Acad. Sci. USA 93(17):8809–8815.artin, N., Gagnon, D. A., Schwartz, M. F., Dell, G. S., and Saffran,E. M. 1996. Phonological facilitation of semantic errors in normaland aphasic speakers. Lang. Cogn. Proc. 11(3):257–282.odayur, B., Prothero, J., Ojemann, G., Maravilla, K., and Brinkley,J. 1997. Visualization-based mapping of language function in thebrain. NeuroImage 6(4):245–258.ohr, J. P., Pessin, M. S., Finkelstein, S., Funkenstein, H. H.,Duncan, G. W., and Davis, K. R. 1978. Broca’s aphasia: Pathologicaland clinical. Neurology 28:311–324.jemann, G. A., Ojemann, J., and Lettich, E. 1989. Cortical languagelocalization in left dominant hemisphere: An electrical stimulationmapping investigation in 117 patients. J. Neurosurg. 71:316–326.jemann, G. A. 1995. Investigating language during awake neurosur-gery. In Neuroscience, Memory, and Language: Decade of the Brain(R. D. Broadwell, Ed.), Vol. 1, pp. 117–131. U.S. Govt. PrintingOffice, Washington, DC.sterhout, L. 1994. Event related brain potentials as tools for comprehend-ing language comprehension. In Perspectives on Sentence Processing(C. Clifton, Jr., et al., Eds.), pp. 15–44. Erlbaum, Hillsdale, NJ.

enniello, M. J., Lambert, J., Eustache, F., Petit-Tabou’e, M. C.,Barr’e, L., Viader, F., Morin, P., Lechevalier, B., and Baron, J. C.1995. A PET study of the functional neuroanatomy of writingimpairment in Alzheimer’s disease. Brain 118:697.

etersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., and Raichle,M. E. 1988. Positron emission tomographic studies of the corticalanatomy of single-word processing. Nature 331:585–589.

oizner, H., Klima, E. S., and Bellugi, U. 1997. What the HandsReveal about the Brain. MIT Press, Cambridge, MA.

rice, C. J., Wise, R. J., Warburton, E. A., Moore, C. J., Howard, D.,Patterson, K., Frackowiak, R. S., and Friston, K. J. 1996. Hearingand saying. Brain 119:919–931.

izzolatti, G., and Arbib, M. A. 1998. Language within our grasp.Trends Neurosci. 21(5):188–194.

oeltgen, D. P., and Heilman, K. M. 1984. Lexical agraphia: Furthersupport for the two-system hypothesis of linguistic agraphia. Brain107:811.

ushworth, M. F., Nixon, P. D., Renowden, S., Wade, D. T., andPassingham, R. E. 1997. The left parietal cortex and motorattention. Neuropsychologia 35(9):1261–1273.

hulag, G., Knorr, U., and Setiz, R. 1994. Inter-subject variability ofcerebral activations in acquiring a motor skill: A study withpositron emission tomography. Exp. Brain Res. 98(3):523–534.

teinmetz, H., Eberling, U., Huang, Y. X., and Kahn, T. 1990. Sulcustopography of the parietal opercular region: An anatomic and MRstudy. Brain Lang. 38:515.hitaker, H. A. 1998. Electrical stimulation mapping of languagecortex. In Handbook of Neurolinguistics (B. Stemmer and H. A.Whitaker, Eds.), pp. 126–129, Academic Press, San Diego, CA.

atorre, R. J., Meyer, E., Gjedde, A., and Evans, A. C. 1996. PETstudies of phonetic processing of speech: Review, replication, and
reanalysis. Cerebral Cortex 6:21–30.

functional roles of broca s area and smg: evidence from...

Documents