Proceedings Fonetik 2009 - Institutionen för lingvistik

More documents

Recommendations

Info

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm UniversityUhm… What’s going on? An EEG study on perceptionof filled pauses in spontaneous Swedish speechSebastian Mårback 1 , Gustav Sjöberg 1 , Iris-Corinna Schwarz 1 and Robert Eklund 2, 31 Dept of Linguistics, Stockholm University, Stockholm, Sweden2 Dept of Clinical Neuroscience, Karolinska Institute/Stockholm Brain Institute, Stockholm, Sweden3 Voice Provider Sweden, Stockholm, SwedenAbstractFilled pauses have been shown to play asignificant role in comprehension and longtermstorage of speech. Behavioral andneurophysiological studies suggest that filledpauses can help mitigate semantic and/orsyntactic incongruity in spoken language. Thepurpose of the present study was to explorehow filled pauses affect the processing ofspontaneous speech in the listener. Brainactivation of eight subjects was measured byelectroencephalography (EEG), while theylistened to recordings of Wizard-of-Oz travelbooking dialogues.The results show a P300 component in thePrimary Motor Cortex, but not in the Broca orWernicke areas. A possible interpretation couldbe that the listener is preparing to engage inspeech. However, a larger sample is currentlybeing collected.IntroductionSpontaneous speech contains not only wordswith lexical meaning and/or grammaticalfunction but also a considerable number ofelements, commonly thought of as not part ofthe linguistic message. These elements includeso-called disfluencies, some of which are filledpauses, repairs, repetitions, prolongations,truncations and unfilled pauses (Eklund, 2004).The term ‘filled pause’ is used to describe nonwordslike “uh” and “uhm”, which are commonin spontaneous speech. In fact they make uparound 6% of words in spontaneous speech(Fox Tree, 1995; Eklund, 2004).Corley & Hartsuiker (2003) also showedthat filled pauses can increase listeners’attention and help them interpret the followingutterance segment. Subjects were asked to pressbuttons according to instructions read out tothem. When the name of the button waspreceded by a filled pause, their response timewas shorter than when it was not preceded by afilled pause.Corley, MacGregor & Donaldson (2007)showed that the presence of filled pauses inutterances correlated with memory andperception improvement. In an event-relatedpotential (ERP) study on memory, recordingsof utterances with filled pauses before targetwords were played back to the subjects.Recordings of utterances with silent pauseswere used as comparison. In a subsequentmemory test subjects had to report whethertarget words, presented to them one at a time,had occurred during the previous session or not.The subjects were more successful inrecognizing words preceded by filled pauses.EEG scans were performed starting at the onsetof the target words. A clearly discernable N400component was observed for semanticallyunpredictable words as opposed to predictableones. This effect was significantly reducedwhen the words were preceded by filled pauses.These results suggest that filled pauses canaffect how the listener processes spokenlanguage and have long-term consequences forthe representation of the message.Osterhout & Holcomb (1992) reported froman EEG experiment where subjects werepresented with written sentences containingeither transitive or intransitive verbs. In someof the sentences manipulation produced agarden path sentence which elicited a P600wave in the subjects, indicating that P600 isrelated to syntactic processing in the brain.Kutas & Hillyard (1980) presented subjectswith sentences manipulated according to degreeof semantic congruity. Congruent sentences inwhich the final word was contextuallypredictable elicited different ERPs thanincongruent sentences containing unpredictablefinal words. Sentences that were semanticallyincongruent elicited a clear N400 whereascongruent sentences did not.We predicted that filled pauses evoke eitheran N400 or a P600 potential as shown in thestudies above. This hypothesis has explanatoryvalue for the mechanisms of the previously92
Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm Universitymentioned attention-enhancing function offilled pauses (Corley & Hartsuiker, 2003).Moreover, the present study is explorative innature in that it uses spontaneous speech, incontrast to most previous EEG studies ofspeech perception.Given the present knowledge of the effectof filled pauses on listeners’ processing ofsubsequent utterance segments, it is clear thatdirect study of the immediate neurologicalreactions to filled pauses proper is of interest.The aim of this study was to examinelisteners’ neural responses to filled pauses inSwedish speech. Cortical activity was recordedusing EEG while the subjects listened tospontaneous speech in travel booking dialogs.MethodSubjectsThe study involved eight subjects (six men andtwo women) with a mean age of 39 years andan age range of 21 to 73 years. All subjectswere native speakers of Swedish and reportedtypical hearing capacity. Six of the subjectswere right-handed, while two consideredthemselves to be left-handed. Subjects werepaid a small reward for their participation.ApparatusThe cortical activation of the subjects wasrecorded using instruments from ElectricalGeodesics Inc. (EGI), consisting of a HydrocelGSN Sensor Net with 128 electrodes. Thesehigh impedance net types permit EEGmeasurement without requiring gel applicationwhich permits fast and convenient testing. Theamplifier Net Amps 300 increased the signal ofthe high-impedance nets. To record and analyzethe EEG data the EGI software Net Station 4.2was used. The experiment was programmed inthe Psychology Software Tools’ software E-Prime 1.2.StimuliThe stimuli consisted of high-fidelity audiorecordings from arranged phone calls to a travelbooking service. The recordings were made in a“Wizard-of-Oz” setup using speakers (twomales/two females) who were asked to maketravel bookings according to instructions (seeEklund, 2004, section 3.4 for a detaileddescription of the data collection).The dialogs were edited so that only the partybooking the trip (customer/client) was heardand the responding party’s (agent) speech wasreplaced with silence. The exact times for atotal of 54 filled pauses of varying duration(200 to 1100 ms) were noted. Out of these, 37were utterance-initial and 17 were utterancemedial.The times were used to manuallyidentify corresponding sequences from theEEG scans which was necessary due to thenature of the stimuli. ERP data from a period of1000 ms starting at stimulus onset wereselected for analysis.ProcedureThe experiment was conducted in a soundattenuated, radio wave insulated and softly litroom with subjects seated in front of a monitorand a centrally positioned loud speaker.Subjects were asked to remain as still aspossible, to blink as little as possible, and tokeep their eyes fixed on the screen. Thesubjects were instructed to imagine that theywere taking part in the conversation − assumingthe role of the agent in the travel bookingsetting − but to remain silent. The total durationof the sound files was 11 min and 20 sec. Theexperimental session contained three shortbreaks, offering the subjects the opportunity tocorrect for any seating discomfort.Processing of dataIn order to analyze the EEG data for ERPs,several stages of data processing were required.A band pass filter was set to 0.3−30 Hz toremove body movement artefacts and eyeblinks. A period of 100 ms immediately prior tostimulus onset was used as baseline. The datasegments were then divided into three groups,each 1100 ms long, representing utteranceinitialfilled pauses, utterance-medial filledpauses and all filled pauses, respectively. Datawith artefacts caused by bad electrode channelsand muscle movements such as blinking wereremoved and omitted from analysis. Badchannels were then replaced with interpolatedvalues from other electrodes in their vicinity.The cortex areas of interest roughlycorresponded to Broca’s area (electrodes 28,34, 35, 36, 39, 40, 41, 42), Wernicke’s area(electrodes 52, 53, 54, 59, 60, 61, 66, 67), andPrimary Motor Cortex (electrodes 6, 7, 13, 29,30, 31, 37, 55, 80, 87, 105, 106, 111, 112).Theaverage voltage of each session was recordedand used as a subjective zero, as shown inFigure 1.93
Page 1 and 2:
Department of LinguisticsProceeding
Page 3 and 4:
Proceedings, FONETIK 2009, Dept. of
Page 5 and 6:
Page 7 and 8:
Page 9 and 10:
Page 11 and 12:
Page 13 and 14:
Page 15 and 16:
Page 17 and 18:
Page 19 and 20:
Page 21 and 22:
Page 23 and 24:
Page 25 and 26:
Page 27 and 28:
Page 29 and 30:
Page 31 and 32:
Page 33 and 34:
Page 35 and 36:
Page 37 and 38:
Page 39 and 40:
Page 41 and 42: Proceedings, FOETIK 2009, Dept. of
Page 43 and 44: Proceedings, FONETIK 2009, Dept. of
Page 91: Proceedings, FONETIK 2009, Dept. of
Page 143 and 144:
Page 145 and 146:
Page 147 and 148:
Page 149 and 150:
Page 151 and 152:
Page 153 and 154:
Page 155 and 156:
Page 157 and 158:
Page 159 and 160:
Page 161 and 162:
Page 163 and 164:
Page 165 and 166:
Page 167 and 168:
Page 169 and 170:
Page 171 and 172:
Page 173 and 174:
Page 175 and 176:
Page 177 and 178:
Page 179 and 180:
Page 181 and 182:
Page 183 and 184:
Page 185 and 186:
Page 187 and 188:
Page 189 and 190:
Page 191 and 192:
Page 193 and 194:
Page 195 and 196:
Page 197 and 198:
Page 199 and 200:
Page 201 and 202:
Page 203 and 204:
Proceedings, FOETIK 2009, Dept. of
Page 205 and 206:
Proceedings, FOETIK 2009, Dept. of
Page 207 and 208:
Page 209 and 210:
Page 211 and 212:
Page 213 and 214:
Page 215 and 216:
Page 217 and 218:
Page 219 and 220:
Page 221 and 222:
Page 223 and 224:
Page 225 and 226:
Page 227:
Department of LinguisticsPhonetics
show all

Proceedings Fonetik 2009 - Institutionen för lingvistik

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?