1. Aphasia and Cognitive Science
Swathi Kiran
Communication Sciences & Disorders
University of Texas at Austin

2. Stages of speech production

3. What is aphasia?
1. Acquired versus developmental
2. Language versus Speech
3. Language versus involvement of other Cognitive Processes
4. Results from brain damage

4. Etiologies of aphasia 1. CVA: Cerebrovascular accident or stroke
1. Ischemic strokes
2. Hemorrhage
2. Aneurysm
3. Tumor
4. Trauma
5. Unknown Etiology

5. Inferior parietal lobule
Language Areas:
Broca’s area (44, 45)
Wernicke’s area
Inferior parietal lobule
Angular gyrus
Supramarginal gyrus
Dorsolateral prefrontal cortex
Insula
Supplementary motor area

7. Cerebral hemorrhage: tissue necrosis
Cerebral infarct
Cerebral hemorrhage

8. Cerebral infarct
Cerebral hemorhage
Multiple infarct

9. Major signs of aphasia: Naming difficulties also seen in dementia.
This incorrect substitution is called a paraphasia
Semantic paraphasia: if the word is related in meaning
Target word Response
Stethoscope Telescope
Phonemic paraphasia substituting some other sounds:
Paper Paker
Neologistic paraphasia: complete novel word which does not exist in that language
Chicken Barnett

10. Nonfluent aphasia
Pt: Yeah.. wendesday…paul and dad.. hospital.. yeah…doctors, two and teeth
Absence of word combinations
No pronouns
Verbs are lacking
Patients more often block on the initial sound of the word
These patients present with articulatory disturbances although the phonological target may be retrieved the word is still unintelligible
Often see a dissociation between oral and written naming in these patients
Partial preservation of written word finding

11. Fluent aphasia
Well all I know is somebody is clipping the knoples and some one someone have of the kenepung aim why I don’t know
I gave him a God! I know it, Why can’t I say it?
He is falling of the t.. t.. t.. anyhow, the mother is sti..sti. she .. the water is falling over the fink… fink…stink…sink…
Havent been around there at all since we got in to this time here, anything about here.. only because we had to do that, and then she got back with it…
Intact grammatical framework
Inability to retrieve nouns and adjectives
Pure morphological words intact (pronouns, prepositions, copulas, auxiliary and modal verbs)
Substitutions of vague, indefinite words such as “thing”
Circumlocutions
Erroneous words or neologisms

12. Comprehension deficit: Can be impaired to variable degrees.
Repetition:
Damage to the perisylvian region
Comprehension deficit:
Can be impaired to variable degrees.
Grammatical processing
Difficulty generating sentences into which the words have specific slots.
Reading and writing:
Alexia with agraphia: presence of both reading and writing impairments, usually with Wernicke’s or transcortical sensory aphasia.
Alexia without agraphia: separate condition by itself in the presence of fluent, spontaneous speech and good comprehension

13. Types of aphasia Broca’s Aphasia Production: Naming impaired
Nonfluent, mute, sparse output, effortful speech
Telegraphic speech
Nouns are easier than verbs
Selective deletion of functor words and abnormal word order
Use of simple sentences more than complex
Use of over learned stereotypical utterances
Flat melodic contour
Naming impaired
Auditory comprehension relatively spared
Non fluent aphasia

16. Wernicke’s aphasia
Production
Fluent, normal prosody
Paraphasic output
Impaired naming: verbal and literal paraphasias
Lacking in content
No self corrections
Naming is a major deficit
Problems in auditory comprehension
Inability to repeat

18. Conduction Aphasia Production Impaired naming Repetition
Fluent spontaneous output
Difficulty in phonological selection
Impaired naming
Repetition
Severely impaired, most common repetition with phonemic paraphasia
Often omit or substitute words
In repetition, multisyllabic words are more difficult

19. Global Aphasia
Production
Can be fluent or nonfluent
Severe deficit crossing all language modalities
Can gesture well
Repetition
Impaired
Auditory Comprehension severely impaired
All aspects of language are impaired

20. Naming

21. Framework for lexical access
Semantic representation
Phonological representation
Phoneme level
Activation of network of features
Influenced by frequency, imageability, animacy
Sequential: 2 stages (Levelt, 1989)
Interactive: Bi-directional links (Dell, 1986)
While a broad area of my research is language processing and recovery in aphasia, today I am going to specifically talk about one aspect of language- that is lexical semantic processing and naming.
I am going to use this general framework that you see here in making three points
1. The factors in involved in normal lexical semantic processes and naming
2. The effects of brain damage on any of these modules, and I will spend some time on semantic representations
3. The potential effect of theoretically motivated treatment on rebuilding or facilitating improvements in these modules.
The framework of lexical semantic processing is barebone view of naming. There are a multitude of models and theories that explain each and every aspect of this process, and my aim is not to go into any of these models/theories in detail, but to provide you with the basic elements of this process, When a picture is presented for naming, the corresponding semantic representation of the picture is activated, and in order to produce to the name of the target- in this case “rose”- the corresponding phonological representation is activated. To produce rose in the right sequence, the phoneme level is activated. Of course, the story is not as simple as this. Within the semantic representation there are numerous factors that influence the process of semantic access, such as frequency, imageability, animacy, number of meanings, category structure, typicality and so forth.
The next important aspect of this framework is the connection between semantic and phonological representations. On a continuum of theories, one extreme the sequential models suggests this semantic processing to precede phonological processing. On the other extreme, the interactive activation models, such this connection to interactive and simultaneous, that is semantic activation and phonological activation are almost simultaneous.
The next aspect of this framework, is the phonological level as called by the sequential models or the word level labeled by the interactive models. This representation essentially codes the phonological form to produced, and again according to the sequential model is just single address/form selected. According to the interactive models, there would be multiple word nodes activated, and following lateral inhibition, the most active node is selected. The process at the phoneme level is essentially the same, information regarding segmental structure, stress etc, is also coded.
Activation of corresponding address or
Activation of potential nodes

22. Interactive activation model

23. A Neural network model of lexical processing
Plaut, 1996

24. Rapp & Goldrick, 2000

25. Rehabilitation of aphasia

26. Treatment for patients with aphasia
Patients with chronic focal lesions (at least 8 months post stroke) and pervasive language impairments are seen for language treatment
Improvements in behavior are observed as a result of treatment

27. What is generalization??
TREATMENT
Apple
Orange

28. Complex to simple generalization Within category Across sentence types
Cross linguistic generalization
Trained language-> untrained language
Cross modal generalization
Reading-> naming, writing
Writing-> reading, naming
Kiran & Thompson, 2003, JSLHR
Kiran, under revision
Thompson, Shapiro, Kiran & Sobecks, 2003, JSLHR
Kiran, Thompson, & Hashimoto, 2001, Aphasiology
Kiran, 2005, Aphasiology
Edmonds & Kiran, 2006, JSLHR
Edmonds & Kiran, 2004, Aphasiology

29. Modeling Recovery of language deficits
Plaut, 1996

30. Semantic Space: Typical and Atypical Items
Features for the category birds:
wings
flies
two legs
lays eggs
feathers
builds nest
sings
beak
nocturnal
eats insects
eats fish
claws
webbed feet
swims
Bird
One approach to imagine a semantic category representation is a multidimensional space with the prototype in the center of the semantic boundary.This is same model that Plaut uses as well.
Since the typical examples are more similar and share more features with the prototype, they are in the center of the boundary/ So for example, Robins and sparrows share features such as small size, live in trees, have wings, fly and so forth. These are features possessed by the prototype as well
Atypical examples on the other hand share less features with the prototype, and are also different from each other, and they are represented in the periphery of the boundary. So for example, penguin has features such lives near water, eats fish, while ostrich have features such as long neck, long legs, which are not shared by each other, but provide variation regarding the features which exist in a category, and still make a penguin or ostrich a bird.
Kiran, S & Thompson, (2003), JSHLR

32. Typicality in inanimate categories
Clothing
Features for the category clothing
Worn in warm weather
Has buttons
Has zippers
Decorative accessory
Optional
Protective covering
Worn by women
Kiran et al., in preparation

33. Kiran et al., in preparation
P1: Furniture 10 20 30 40 50 60 70 80 90 1 3 5 7 9 11 13 15 17 19 21 23 25
Probes
Percent Accuracy
Atypical-Untrained
Typical-Trained
Treatment BL
P1: Clothing
100 27
Atypical-trained
Typical-Untrained
Behavioral Data
Kiran et al., in preparation

34. Recovery of language in bilingual individuals

35. Interpretation of Participant 1 (balanced) results
Spanish treatment
Semantics
“Apio”
“Celery”
“Cabbage”
“Repollo”
Provision of treatment in Spanish for participant 2 who was equally proficient resulted in improvements in the items trained as well as generalization to the semantically related items in the same language. Crosslinguistic generalization was also observed for English translations like celery and for the semantically related items such as cabbage.
These results suggest that, in a balanced bilingual individual, the connections between all components are relatively equal and hence crosslinguistic generalization occurs as a function of treatment. L1 L2

36. Participant 3 and Participant 2 (both English dominant)
Semantics
“Orange”
“Naranja”
“Apple”
“Manzana”
Spanish (non-dominant language) treatment
Participant 2
Semantics
“Shark”
“Tiburón”
“Whale”
“Ballena”
Participant 3
Spanish (non-dominant language) treatment
So to interpret these results: Participant 1 who was English dominant, upon receiving treatment in his dominant language, was improved on access to trained items and to semantically related items in the same language. No crosslinguistic generalization was observed from English-Spanish. Training this patient on his non-dominant language, however, resulted in improved access to the items trained, although no improvements were observed to semantically related items in the same language. Improvements were observed in English language. That is training naranja improved access to orange and well as further improvements to apple.

37. Participant 1 (equally proficient) results
Acquisition
Treatment
Apio
N=10
Repollo
Generalization (C = 0.705, p = 0.001)
N=10
Celery
Generalization (C = 0.363, p = 0.071)
N=10
Participant 2, who was also unable to name any items in baseline, was treated on spanish set 1, which included items like apio. Performance on those items improved to criteion as did the semantically related items in the same language. Crosslinguistic generalization was also observed for the translations in English, so apio improved its English translation “celery” and generalization was observed the semantically related item in English also, that is repollo improved cabbage.
Only if there are questions
The two items that the subject got correct (2/5) in the Spanish control items were sobre (envelope) and bombero (fireman). Initially, she called sobre “carta” (letter), which is semantically related. She was soon able to make the connection. For “bombero,” she could already access the word in English during baseline, AND her ex-husband was a fireman, so she worked very hard to “translate” fireman to bombero during probes. The English unrelated were “less correct” to start with and so have not improved.
Generalization (C = 0.700, p = 0.001)
Cabbage
N=10

38. Participant 3 (English dominant) Results
Acquisition
Tiburón
N=10
(C = 0.428, p = 0.082)
No Generalization
Ballena
N=10
Shark
N=10
(C = 0.625, p = 0.021)
Generalization
(C = 0.642, p = 0.018)
Generalization
Whale
N=10

40. fMRI studies of language recovery in aphasia

41. Current research questions..
Are behavioral changes associated with functional changes in the brain?
What are the neural correlates underlying brain plasticity in language recovery
What are typical regions in the brain involved in processing language?
What are regions in the damaged brain that can sub serve residual language abilities
What are regions that can support behavioral language recovery

42. Patient DG_nonfluent aphasia Pre treatment_picture naming

43. Patient DG_nonfluent aphasia Picture naming_post treatment

44. In normal controls, activation in Broca’s area

45. Patient 1: Good recovery: activation in perilesional areas or undamaged regions in language dominant hemisphere.

46. Patient 2: Good recovery: activation in perilesional areas or undamaged regions in language dominant hemisphere.
Right homologous area activation observed in this patient.

47. Patient 5: Good recovery: activation in perilesional areas or undamaged regions in language dominant hemisphere.