Lecture 3 12 Oct., 2005 Language and the Brain Helena Gao Lecture 3; Oct. 12, 2005 Required readings: Gazzaniga, M., Ivry, R., & Mangun, G. (2001). Cognitive Neuroscience: The Biology of the Mind. New York: W.W. Norton and Co. Chapter 9: Language and the brain, pp. 351-399. Vygotsky, L. (1996). Thought and Language. Newly revised and edited by Alex Kozulin. The MIT Press. Chapter 4: The Genetic Roots of Thought and Speech. pp. 68-95. Recommended readings: Shapiro, K., & Caramazza, A. (2003). The representation of grammatical categories in the brain. Trends in Cognitive Science, 7(5), 201-206. Brain areas involved in Language Three major types of Aphasia Rosenzweig: Table 19.1, p. 615 Borca’s aphasia Wernicke’s aphasia Nonfluent speech Fluent speech but unintelligible Global aphasia Total loss of language Others: Conduction, Subcortical, Transcortical Motor/Sensory (see also Kandel, Table 59-1) Broca’s Aphasia Brodmann 44, 45 Lesions in the left inferior frontal region (Broca’s area) Nonfluent, labored, and hesitant speech Most also lost the ability to name persons or subjects (anomia) Can utter automatic speech (“hello”) Comprehension relatively intact Most also have partial paralysis of one side of the body (hemiplegia) If extensive, not much recovery over time Wernicke’s Aphasia Brodmann 22, 30 Lesions in posterior of the left superior temporal gyrus, extending to adjacent parietal cortex Fluent speech But contains many paraphasias “girl”-“curl”, “bread”-“cake” Syntactical but empty sentences Cannot repeat words or sentences Unable to understand what they read or hear Usually no partial paralysis Sign Languages Full-fledged languages, created by hearing- impaired people (not by Linguists): Dialects, jokes, poems, etc. Do not resemble the spoken language of the same area (ASL resembles Bantu and Navaho) Pinker: Nicaraguan Sign Language Another evidence of the origins of language (gestures) Most gestures in ASL are with right-hand, or else both hands (left hemisphere dominance) Signers with brain damage to similar regions show aphasia as well Spoken and Sign Languages Neural mechanisms are similar fMRI studies show similar activations for both hearing and deaf But in signers, homologous activation on the right hemisphere is unanswered yet Dyslexia Problem in learning to read Common in boys and left-handed High IQ, so related with language only Postmortem observation revealed anomalies in the arrangement of cortical cells Micropolygyria: excessive cortical folding Ectopias: nests of extra cells in unusual location Might have occurred in mid-gestation, during cell migration period Acquired Dyslexia = Alexia Disorder in adulthood as a result of disease or injury Deep dyslexia (pays attn. to wholes): Surface dyslexia (pays attn. to details): “cow” -> “horse”, cannot read abstract words Fails to see small differences (do not read each letter) Problems with nonsense words Nonsense words are fine Suggests 2 different systems: One focused on the meanings of whole words The other on the sounds of words Electrical Stimulation Penfield and Roberts (1959): During epilepsy surgery under local anesthesia to locate cortical language areas, stimulation of: Large anterior zone: Both anterior and posterior temporoparietal cortex: stops speech misnaming, impaired imitation of words Broca’s area: unable comprehend auditory and visual semantic material, inability to follow oral commands, point to objects, and understand written questions PET by Posner and Raichle (1994) Passive hearing of words activates: Repeating words activates: No activation in Broca’s area But if semantic association: Both motor cortices, the supplemental motor cortex, portion of cerebellum, insular cortex While reading and repeating: Temporal lobes All language areas including Broca’s area Native speaker of Italian and English: Slightly different regions Due to phonetic alphabet of Italian… (“ghotia”) PET by Damasio (1989) Different areas of left hemisphere (other than Broca’s and Wernicke’s regions) are used to name (1) tools, (2) animals, and (3) persons Stroke studies support this claim Three different regions in temporal lobe are used ERP studies support that word meaning are on temporal lobe (may originate from Wernicke’s area): “the man started the car engine and stepped on the pancake” Takes longer to process if grammar is involved Williams Syndrome Caused by the deletion of a dozen genes from one of the two chromosomes numbered 7 Shows dissociation between language and intelligence, patients are: Fluent in language But cannot tie their shoe laces, draw images, etc. Developmental process is altered: Number skills good at infancy, poor at adulthood Language skills poor at infancy, greatly improved in adulthood Guest speaker in the colloquium, Annette KarmiloffSmith, claims the otherwise: Development alters the end result of the syndrome (?) Lateralization of the Brain Human body is asymmetrical: heart, liver, use of limbs, etc. Functions of the brain become lateralized Each hemisphere specialized for particular ways of working Split-brain patients are good examples of lateralization of language functions Lateralization of functions (approximate) Left-hemisphere: Sequential analysis Analytical Problem solving Right-hemisphere: Simultaneous analysis Visual-Spatial skills Language Cognitive maps Personal space Facial recognition Drawing Emotional functions Synthetic Recognizing emotions Expressing emotions Music Other studies Right ear advantage in dicothic listening: Words in left-hemisphere, Music in right Due to interhemispheric crossing Supported by damage and imaging studies But perfect-pitch is still on the left Asymmetry in planum temporale: Musicians with perfect-pitch has 2x larger PT Evident in newborns, thus suggesting innate basis for cerebral specialization for language and speech The Storage of Words and Concepts: The Mental Lexicon The Mental Lexicon: a mental store of information about word that includes semantic information, syntactic information, and the details of word forms. Most psycholinguistic theories agree on the central role for a mental lexicon in language Some theories propose one mental lexison for both language comprehension and production Other models distinguish between input and output lexia The representation of orthographic and phonological forms must be considered in any model. The mental lexicon is thought to be organized as information-specific networks – a model proposed by William Levelt in 1994 W. W. Norton Support for the model Semantic Priming Studies Using a lexical decision task Subjects are faster and more accurate at making the lexicon decisions when the target is proceeded by a related prime (e.g., car truck) than a unrelated prime (e.g., whip truck). Expectancy-induced priming might occur if the time bet. The presentation of primes and targets is long (e.g., > 500 msec) and the proportion of related word pairs are like car-truck, cat-dog, etc. The Nature of Conceptual or Semantic Representation Question: Is the same conceptual representation of a robin activated regardless of whether one hears the word robin or sees one flying? A model proposed by Collins and Loftus (1975) Word meanings are represented in a semantic network in which words, represented by conceptual nodes, are connected with each other. • Although this semantic network model (Collins & Loftus, 1975)has been extremely influential, it is a matter of debate how Word meanings are represented. • No matter how they are represented, it is agreed that a mental store of word meanings is crucial to normal language comprehension and production. W. W. Norton Support from neurological evidence for the semantic network idea Different types of neurological problems create deficits in understanding and producing the appropriate meaning of a word or concept. Patients with Wernicke’s aphasia makes errors in speech production that are known as semantc-paraphasias (e.g., using the word horse to mean cow) Patients with progressive-semantic dementia initially show impairments in the conceptual system, while other mental and language abilities are spared. Evidence from research done by E. Warrington (1970s-1980s) Findings: semantic problem can be localized specifically to certain semantic categories , such as animals versus objects. E.g., patients who had great difficulties pointing to pictures of food or living things when presented with a word, whereas their performance with man-made objects like tools was much better. Evidence from H. Damasio and her colleagues’ studies (1996) Investigated a large population of patients with brain lesions A naming task in three different conditions: (1) naming famous faces, (2) naming animals, and (3) naming tools 30 patients, 29 of whom had a lesion in the left hemisphere, showed impairments in this task. 7 patients demonstrated impairment in naming faces, 5 in naming animals; and 7 in naming tools. The remaining 11had a combination of problems in word retrieval for faces, animals, and tools, faces and animals, or animals and tools, but never for the combination of faces and tools together without also a deficit in animals. Location of the brain lesions that are correlated with selective deficits in naming persons, animals, or, tools. (Damasio et al (1996) W. W. Norton Pet Scanning • Brain damage in the left temporal pole (TP) correlated with problems in retrieving the names of persons • Lesions in the anterior part of the Left inferior temporal (IT) lobe correlated with problems in naming animals • Damage to the posterolaterol part of the left inferior temporal lobe, along with lateral temporo-occipito-parietal junction (IT+), was correlated with problems in retrieving the names of tools W. W. Norton W. W. Norton Three levels of Representation for Word Knowledge – predicated by cognitive models of word Production Based on Damasio et al’s Results (Caramazza, 1996) W. W. Norton Schematic representation of the components that are involved in spoken and written language comprehension. Input can enter via either auditory (spoken word) or visual (written word) modalities. The flow of info is bottom up, from Perceptual identification to “higher-level” word and lemma activation. Interactive models of language Understanding would predict top-down influence to play a role as well. W. W. Norton Spoken Input – lack of segmentation Courtesy of Tamara Swaab. © 1997 by the Massachusetts Institute of Technology. Courtesy of Tamara Swaab. © 1997 by the Massachusetts Institute of Technology. Spectral properties vary according to sounds (Klatt, 1989) W. W. Norton A model for letter recognition, the pandemonium model (Selfridge, 1959) W. W. Norton McClelland, J.L., and Rumelhart, D.E. (1986). Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Vol. 2: Psychological and Biological Models. Cambridge, MA: MIT Press. Fragment of a connectionist network for letter recognition (McCleland & Rumelhart, 1981) A cohort model of British psychologist William Marslen –Wilson (Marslen –Wilson & Tyler, 1980) W. W. Norton ERPs recorded in response to sentences that start with before (dashed lines) and after (solid lines) Munte et al (1998). Task: “After/Before the scientist submitted the paper, the journal changed its policy” Before are More Negative in polarity Summary of lesions in the anterior superior temporal cortext that lead to deficits in Syntactic processing (Mazoyer et al, 1993) W. W. Norton Outline of the theory of speech production developed by William Levelt (1999) Adapted from Levelt, W.J.M., The Architecture of Normal Spoken Language Use, in Blanken, G., Dittman, J., Grimm, H., Marshall, J.C., and Wallesh, C-W. (Eds.), Linguistic Disorders and Pathologies: An International Handbook. Berlin: Walter de Gruyter, 1993 Speech Production Method: Stimuli: words in Dutch -> Equ.: ERP Proc.: Subjects were asked to make a response only when the words representing the pic. Started with a “b’, and to withdraw their response when the word Started with a “s”. Adapted from van Turennout, M. Hagoort, P., and Brown, C.M. (1999). Brain activity during speaking: From syntax to phonology in 40 milliseconds. Science, 280, 572-574. A study to test whether Lemma selection indeed Precedes activation of Appropriate lexeme (as In Levelt’s model) by Van turennout (1999)