First steps in Language Acquisition: Evidence from ERPs Angela D. Friederici Max Planck Institute for Human Cognitive and Brain Sciences Leipzig, Germany Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Method of Event Related Potential (ERP) • High temporal resolution as it measures the brain‘s activity millisecond by millisecond. • No behavioral reaction necessary. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences ERP method ONGOING EEG Amplifier S S S S one sec AUDITORY EVENT-RELATED POTENTIAL -6 µV N400 Signal averager ELAN AUDITORY STIMULUS (S) P600 P200 +6 µV STIMULUS ONSET Trieste, 07.05.2006 200 400 600 800 1000 Time (ms) Max Planck Institute for Human Cognitive and Brain Sciences Infants' first Steps into Language are Based on Prosodic Information The identification of word boundaries (word onset and offset) is eased by information about word stress. In a language like English or German, in which the stress of two syllable words lies predominantly on the first syllable, stress information can certainly help to identify word onsets. The identification of syntactic phrase boundaries is eased by prosodic information as each intonational phrase boundary is a syntactic phrase boundary. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Syllable Length and Word Stress Discrimination • Word stress in German is predominantly (› 90% of two syllable words) on the first syllable. • Word stress is indicated by a number of acoustic parameters with syllable duration (length) being the most prominent one. • In order to investigate the infants' ability to discriminate words stressed on the first syllable versus words stressed on the second syllable we decided to initially investigate the infants' ability to discriminate between a long and a short syllable. We did so by using the MMN paradigm. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Mismatch Negativity (MMN) ... is a pre-attentive response. ... is elicited by any discriminable change in repetitive auditory stimulation (oddball design): sta sta sta dev sta sta ... is the result of deviance detection. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 1 Discrimination of Syllables of Different Duration /ba:/ 341 ms /ba/ 202 ms Standard 5/ 6 Deviant 1/ 6 1. Block baa 2. Block ba Trieste, 07.05.2006 baa ba ba baa baa ba baa ba baa ba ba baa ba baa baa ba ba ... ba ... Max Planck Institute for Human Cognitive and Brain Sciences The Mismatch Negativity MMN Adults: Difference wave between DeviantStandard Source: Friedrich, Weber & Friederici, Psychophysiology, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 1 Discrimination of Syllables of Different Duration ERP and difference waves (Deviant-Standard) of 2-month-old infants Source: Friederici, Friedrich & Weber, NeuroReport, 2002 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 1 Discrimination of Syllables of Different Duration Difference waves for long syllable in 2-monthold infants and adults Source: Friederici et al., NeuroReport, 2002; Friederici et al., 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 1 • Infants by the age of 2 months are able to discriminate long syllables from short syllables. • Infants more easily discriminate a long syllable in a stream of short syllables than vice versa. • Infants show a positivity as MMN response. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Hypothesis for SLI One of the hypotheses we persued was the notion that a major underlying cause of Specific Language Impairment (SLI) is a deficiency in processing prosodic information. If so, infants with risk for SLI may be deficient already in discriminating long from short syllables at the age of 2 months. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 1b Difference waves for 2-month-olds with and without risk for SLI Source: Friedrich, Weber & Friederici, Psychophysiology, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 1b Infants with risk for SLI differ from those with no risk already at the age of 2 months in their perceptual ERP parameters for duration discrimination. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 2 Discrimination of Different Stress Patterns /ba:ba/ 750 ms /baba:/ 750 ms Standard 5/ 6 Deviant 1/ 6 1. Block baaba baaba baaba babaa baaba baaba 2. Block babaa babaa babaa babaa baaba babaa Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 2 Discrimination of Different Stress Patterns Difference waves 4- and 5-month-old infants MMN Source: Weber, Hahne, Friedrich & Friederici, Cognitive Brain Research, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 2 • Infants by the age of 5 months are able to discriminate word stress of two syllable words. • Discrimination is evident as indicated by MMN for the trochaic pattern (stress on first syllable) which is the predominant pattern in German. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 2b Discrimination of Different Stress Patterns Difference waves 5-month-old infants MMN Source: Weber, Hahne, Friedrich & Friederici, Cognitive Brain Research, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 2b Infants with risk for SLI demonstrate a reduced discrimination ability for the language-specific trochaic pattern at the age of 5 months. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Stress Pattern Discrimination Novel Analysis: 5-month-olds At-risk group classification based on word production at 2 years. MMN Source: Weber, Hahne, Friedrich & Friederici, Cognitive Brain Research, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Processing Intonational Phrase Boundaries As each intonational phrase boundary is a syntactic phrase boundary, the identification of an intonational phrase boundary provides an easy entrance into the syntax of a given language. How can we investigate this identification process using ERPs? Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 3 Prosodically correct conditions (A) correct: Late Closure [IP1 Peter verspricht Anna zu ARBEITEN ] # [IP2 und ... ('Peter promises Anna to work and ... (B) correct: Early Closure [IP1 Peter verspricht ] # [IP2 ANNA zu entlasten ] [IP3 und ... ('Peter promises to support Anna and ... Source: Steinhauer, Alter & Friederici, Nature Neuroscience, 1999 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Effect of Intonational Phrase Boundary Spoken sentences: Adults Source: Steinhauer, Alter & Friederici, Nature Neuroscience, 1999 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Spoken Sentences: 8-month-old Infants Source: Pannekamp, Weber & Friederici, NeuroReport, in press Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 3 Infants by the age of 8 months show a brain response to intonational phrase boundary similar to that of adults. adults Trieste, 07.05.2006 8-month-old infants Max Planck Institute for Human Cognitive and Brain Sciences Development of Prosodic Perception basic mechanisms underlying the identification of intonational (syntactic) boundaries basic mechanisms underlying the identification of word boundaries ERPcorrelates discrimination of syllable duration Birth 2 discrimination word stress pattern sensitivity to intonational phrase boundary 5 MMN 8 Months of age CPS Source: Friederici, Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Processing Lexical-Semantic Information To understand mechanisms of early word learning, it is important to know whether the infant’s brain acts in an adult-like manner when processing meaningful words in meaningful contexts. The N400, a negative component in the ERP of adults, reflects neural mechanisms of semantic integration into context. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 4: Lexical-Semantic Processing In the present study, we investigated whether the neural mechanisms observed for adults are already working during early language acquisition when infants know only a few words. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 4: Lexical-Semantic Processing Using a picture-word design we recorded the ERP response of 12-, 14- and 19-month-old infants on slowly spoken basic level words. While infants were looking at sequentially presented pictures they were acoustically presented with words that were either congruous or incongruous to the picture contents. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Lexical-Semantic Processing Paradigm “Ente” (duck) congruous “Ball” (ball) incongruous “Lampe” (lamp) time Trieste, 07.05.2006 incongruous “Schaf” (sheep) congruous Max Planck Institute for Human Cognitive and Brain Sciences Adults Adapted from Friederici,Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 4: Lexical-Semantic Processing ERP of 19-month-old infants N400 Source: Friedrich & Friederici, Journal of Cognitive Neuroscience, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 4: Lexical-Semantic Processing ERP of 19-month-old infants: normals and at risk (low scores on word & sentence production) Normals At risk Source: Friedrich & Friederici, Psychophysiology, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences 19-month-olds Adults Adapted from Friederici,Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences 14-month-olds 19-month-olds Adults Adapted from Friederici,Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences 12-month-olds 14-month-olds 19-month-olds Adults Adapted from Friederici,Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences 12-month-olds 14-month-olds 19-month-olds 400 – 700 ms 400 – 700 ms 400 – 700 ms Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 4: Lexical-Semantic Processing ERP of 12-month-old infants early neg. early neg. Source: Friedrich & Friederici, Journal of Cognitive Neuroscience, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Summary Experiment 4 Lexical-Semantic Processing Early starting ERP differences in the infants’ temporal brain region have previously been linked to processing differences between known and unknown words (Mills et al., 1993). • In our study, developmental changes in the early time range reflect increased familiarity with the presented words. The early negativity for congruous words in 12-month-old infants is interpreted as facilitation of phonological processing by lexical priming. • The later centro-parietal negativity for incongruous words in 14- and 19-month-old infants is regarded as infant N400. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Conclusion Experiment 4 The occurrence of a phonological-lexical priming effect in both age groups implies that not only 14-, but also 12month-old infants already create lexical expectations from picture contents. This result suggests the existence of some lexical-semantic knowledge even in 12-month-old infants. Infants at that age, however, do not display a N400 semantic incongruity effect that is present in 14-month-old infants. From that we conclude that N400 maturates during the first half of the second year of life. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences ERP correlates Developmental stages Development of Auditory Language Perception Basic stages and ERP correlates Identification of intonational boundaries Identification of word boundaries Discrimination of phonemes Birth 2 3 Discrimination word stress pattern 4 5 6 Sensitivity to intonational phrase boundary 7 8 Lexical processing Lexical form Lexical semantics 12 14 Months of age MMN CPS N400 months of age Adapted from Friederici, Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences ERP correlates Developmental stages Development of Auditory Language Perception Basic stages and ERP correlates Identification of intonational boundaries Identification of word boundaries Discrimination of phonemes Birth 2 3 Discrimination word stress pattern 4 5 6 Sensitivity to intonational phrase boundary 7 8 Lexical processing Sentence processing Lexical form Lexical semantics Selectional restriction of verbs Local phrase structure building 12 14 19 32 Months of age MMN CPS N400 months of age ELAN-P600 Adapted from Friederici, Trends in Cognitive Sciences, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences MAX PLANCK INSTITUTE FOR HUMAN COGNITIVE AND BRAIN SCIENCES LEIPZIG Manuela Friedrich Anja Hahne Birgit Herold Claudia Männel Regine Oberecker Ann Pannekamp Christiane Weber Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 5: Semantic Processes in Sentences The next question is: Do infants at that age also process semantic relations in sentential context? Experiment 5 examines this question. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Semantic Violation N400 Cz correct: -5 µV "Das Baby wurde gefüttert" "The baby was fed" incorrect: "Das Lineal wurde gefüttert" "The ruler was fed" 5 0 0.5 correct incorrect 1 sec CZ Source: Hahne & Friederici, Cognitive Brain Research, 2002 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Stimulus Material correct Die Katze trinkt die Milch. The cat drinks the milk. semantically incorrect Die Katze trinkt den Ball. The cat drinks the ball. Source: Friedrich & Friederici, NeuroReport, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 5: ERP of 24-month-olds N400 Source: Friedrich & Friederici, Journal of Cognitive Neuroscience, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 5: ERP of 19-month-olds N400 Source: Friedrich & Friederici, Journal of Cognitive Neuroscience, 2004 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Conclusion Experiment 5 The data show that children at the age of 19 months can process semantic relations between words in a sentence quite similar to adults. The N400 effect, however, is longer lasting than in adults. As the N400 is taken to reflect integration difficulties, these data suggest that integration processes take longer in young children compared to adults. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Processing Syntactic Information In adult syntactic processing ERP studies, the ELAN has been interpreted to reflect highly automatic phrase structure building processes, whereas the P600 is taken to reflect processes of syntactic integration (e.g. Hahne & Friederici, 1999). If so, the P600 should be present earlier than the ELAN which was shown in 6- to 13-year-old children for syntactic violations in passive sentences (Hahne, Eckstein & Friederici, 2003). Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Syntactic Violation F7 -5 µV correct: ELAN "Das Hemd wurde gebügelt." "The shirt was ironed." 5 incorrect: 0.5 1 sec 0.5 1 sec Pz "Die Bluse wurde am gebügelt." -5 µV "The blouse was on ironed." correct incorrect 0 F7 5 PZ 0 P600 Source: Hahne & Friederici, Cognitive Brain Research, 2002 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 6: Syntactic Processing Experiment 6 investigates at what age we can detect syntax-related ERP components. In this experiment syntactic violations were realized in simple active sentences. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Stimulus Material correct Der Löwe brüllt. The lion roars. syntactically incorrect Der Löwe im brüllt. The lion in the roars. correct filler item Der Löwe im Zoo brüllt. The lion in the zoo roars. Source: Oberecker, Friedrich & Friederici, Journal of Cognitive Neuroscience, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 6: Syntax adults ELAN P600 Source: Oberecker, Friedrich & Friederici, Journal of Cognitive Neuroscience, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 6: Syntax 32-month-old children ELAN P600 Source: Oberecker, Friedrich & Friederici, Journal of Cognitive Neuroscience, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 6: Syntax 24-month-old children P600 Source: Oberecker & Friederici, submitted Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences 24-month-old children -10 32-month-old children µV -10 µV ELAN 0.8 1.6 P600 correct incorrect Trieste, 07.05.2006 µV ELAN s 0.8 1.6 1.6 5 10 10 -5 s s 0.8 Adults P600 correct incorrect P600 correct incorrect Max Planck Institute for Human Cognitive and Brain Sciences Conclusion Experiment 6 The data presented show for the first time that the infant‘s brain is sensitive to phrase structure violations in auditorily presented sentences already at the age of 2.5 years. Their ERP pattern reflects initial processes of structure building (ELAN) as well as late processes of syntactic integration difficulty (P600). Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Experiment 7: Syntax 32-month-olds at risk Source: Oberecker & Friederici, in preparation Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences No risk 24-month-olds -10 No risk 32-month-olds µV -10 µV ELAN 0.8 1.6 P600 correct incorrect Trieste, 07.05.2006 µV s 0.8 1.6 1.6 15 10 10 -15 s s 0.8 Risk 32-month-olds P600 correct incorrect correct incorrect Max Planck Institute for Human Cognitive and Brain Sciences Conclusion Experiment 7 Children with risk for SLI do not demonstrate an ELANP600 pattern at the age of 32 months. This suggests that children classified at risk for SLI are delayed in their syntactic processing abilities, not only with respect to initial on-line syntactic structure building, but also with respect to late processes of syntactic integration. Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences Processing Phonotactically legal and illegal Pseudowords ERP of 19-month-old infants: normals and at risk (low scores on word & sentence production) Normals At risk Source: Friedrich & Friederici, Psychophysiology, 2005 Trieste, 07.05.2006 Max Planck Institute for Human Cognitive and Brain Sciences