PYB304 Behavioural Neuroscience - Language and Aphasia Lecture Notes

Language II: Language Disorders and Brain Representation

Aphasia

• Defined as the loss of language processing ability after brain damage.
• Individuals are referred to as "people with aphasia (PWA)" not "aphasics."
• Aphasia is not:
  • An impairment of intellectual functioning.
  • A psychiatric disturbance.
  • A primary motor or sensory deficit.
  • A developmental disorder.

Classic Views of Aphasia

• Three principles:
  1. Localization of language processors.
  2. Damage to a single processor can produce multiple deficits.
  3. Language processors are localized based on their relationship to primary sensory/motor functions.

Broca’s Aphasia

• Characterized as non-fluent, expressive aphasia.
• Symptoms:
  • Major disturbance in speech production.
  • Retention of some nouns and verbs.
  • Loss of pronouns, articles, and conjunctions (telegraphic speech).
  • Intact comprehension.
• Broca's area integrity (left frontal convolution) is necessary for articulation (Broca's 1862 conclusion based on patient Leborgne).
• Lesions extend into deep white matter, including the insular cortex and basal ganglia (Dronkers et al., 2008).

Wernicke’s Aphasia

• Characterized as fluent, receptive aphasia.
• Symptoms:
  • Major disturbance of auditory comprehension.
  • Fluent speech with normal rate, rhythm, and intonation.
  • Disturbances of sound and word structures.
  • Semantic substitutions or paraphasias.
  • Poor repetition and naming abilities.

Conduction Aphasia

• Symptoms:
  • Failure to repeat words.
  • Paraphasias (phonemic errors).
• Disconnection syndrome: disruption of the arcuate fasciculus (dorsal white matter tract between Broca’s and Wernicke’s areas) (Yeh et al., 2018).

Wernicke-Lichtheim Model

• The classic disconnection model of aphasias.
• Lichtheim (1885) systematized Wernicke’s (1874) model using a 'house' diagram to show functional areas and input/output pathways.
• Able to predict some aphasias.
• Revived by Norman Geschwind in the mid-1960s.
• Ludwig Lichtheim: "Our task is to determine the connections and localisation of the paths of innervation subservient to language and its correlated functions".

Lichtheim's House Model

• Components:
  • Concept Center (C)
  • 'Articulate speech' Motor area
  • 'Auditory word forms' Auditory area
• Broca's aphasia affects the motor area causing impaired articulate speech.
• Wernicke's aphasia involves the auditory area impacting auditory word forms.
• Conduction aphasia occurs due to damage to the arcuate fasciculus.

Transcortical Sensory Aphasia

• Disconnection syndrome: disconnection of auditory and concept centers; damage to tracts in the temporo-parietal-occipital junction.
• Extremely rare!
• Symptoms more often observed via transient interruption of these pathways via direct electrical stimulation in pre-surgical mapping (Boatman et al., 2000).
• Symptoms:
  • Disturbance of auditory comprehension.
  • Semantic paraphasias.
  • Fluent, grammatical speech.
  • Good repetition.

Transcortical Motor Aphasia

• Disconnection syndrome: disconnection of concept centre from motor and auditory language centres.
• Lesion to tracts superior and/or anterior to Broca’s area.
• Symptoms:
  • Intact auditory comprehension.
  • Good repetition.
  • Severe disturbance in initiating responses (adynamic).

Wernicke-Lichtheim-Geschwind Model

• The Wernicke-Geschwind model of the 1960s added a role for the angular gyrus in silent reading (with input to Wernicke's area) and Heschl's gyrus (primary auditory cortex) in silent listening.

Psycholinguistics and Aphasia

• Limitations of classic aphasia syndromes:
  • Poor classification of patients.
  • Lesion overlap/variability.
  • Little assistance for treatment planning.
• Psycholinguistics emphasizes language processing operations instead of just "production" and "comprehension."

Key Language Processing Operations:

• Phonology: sounds that compose language and the rules that govern their combination.
• Semantics: words and their meaning.
• Syntax: methods for combining individual words to convey propositional meaning.

Psycholinguistics: Phonology

• Two ways to represent sound in speech:
  1. Phonemes: smallest unit of sound that can distinguish one word from another; (e.g., /b/ in /bat/ and /p/ in /pat/).
     • Allophones are different representations of the same phoneme (e.g., /p/ in /pill/ vs. /spill/).
     • In English: lips, slip, spill, pills, and lisp comprise the “same sounds” in different orders.
  2. Phonetic: how phonemes are produced in different contexts (i.e., select correct allophone).
     • International Phonetic Alphabet (IPA).
     • e.g., the difference between English and French pronunciation of ‘cave’.
• People with Broca’s aphasia:
  • Mispronounce phonemes.
  • Poor phonetic ability and phonemic selection/discrimination.
• People with Wernicke’s aphasia:
  • Substitute phonemes (e.g., /p/ for /b/).
  • Poor phonemic selection/discrimination BUT preserved phonetic ability.

Psycholinguistics: Syntax

• People with Broca’s aphasia have difficulty with syntactic production:
  • Few function words (e.g., verbs) are produced.
  • Content words preserved (telegraphic speech).
  • Evident in spontaneous speech, repetition, and writing.
  • Anterior lesions are also associated with comprehension deficits.
• Agrammatic aphasia (a graded impairment):
  • Production and comprehension are impaired, although deficits are dissociable.
  • “agrammatism” first coined by Adolf Kussmaul (1877).
  • Example: Sentence: “The girl the boy is chasing is tall”. Question: “Who is being chased?” - Broca’s patients have trouble answering.
  • Wernicke’s aphasia patients usually do not have difficulties with syntax.

Psycholinguistics: Semantics

• Word meaning considered separate to lexical (surface) form.
• Nb. Simplistic distinction
• Lesions resulting in “double dissociation” of lexical and semantic representations of words:
  • Intact semantic knowledge, impaired naming = ‘Tip of-the-tongue’ & anomic deficits.
  • Intact naming, impaired semantics = knowledge loss; semantic dementia/primary progressive aphasia.
• Semantic processing is relatively spared in anterior lesioned (Broca’s) patients.
• Problems arise for anterior lesion patients when syntax is important in sentence comprehension (e.g., “Place the blue circle on top of the big red square”).
• Posterior lesions (Wernickes) are more often characterized by semantic impairments.
• Lesions to the anterior temporal lobes (semantic dementia) are also characterized by semantic impairments, but affect spontaneous (fluent) speech and sentence comprehension less.

Psycholinguistic vs Classical

• Classical characterisation of aphasia deficits:
  • Broca’s = poor speech production
  • Wernicke’s = poor speech comprehension
  • Dissociation: comprehension vs. production
• Psycholinguistic characterisation:
  • Anterior = syntactic processing
  • Posterior = semantic processing
  • Dissociation: syntax vs. semantics

Dysarthria and Apraxia of Speech

• Both are articulatory-motor disorders.
• They are not disturbances of language, i.e., not aphasias.

Dysarthria

• Disturbance of speech musculature in terms of speed, strength, steadiness, coordination, precision, tone, and range of motion.
• Corticospinal tract is commonly affected

Apraxia of Speech (AoS)

• Articulatory-motor disorder (rare).
• Characterized by articulatory errors increasing with increasing word and phrase length.
• Impairment in the ability to program speech movements.
• Debate about the critical lesion (but most strongly associated with motor cortex lesions).

Alexia and Agraphia

Alexia

• Impairment in the ability to read.
• Subtypes:
  • Surface
  • Phonological
  • Deep

Agraphia

• Impairment in the ability to write.
• Dissociable
• Alexia without agraphia
• Joseph Jules Déjerine (1891/1892)
  • Two patients with left parietal lesions
  • One with poor reading and writing: Alexia with agraphia
    • Disturbance to the ‘optical images for words’
  • Second patient with poor reading and preserved writing: Alexia without agraphia (i.e. pure alexia)

Alexia and Dual Routes to Reading

1. Phonological route to reading – convert letter strings to sounds  understand the meaning (grapheme-to-phoneme)
2. Direct route – printed words are directly linked to meaning in a visual form system
   • When reading irregular words, such as “yacht” or “colonel” or “pint”

Alexia Subtypes

• Surface Alexia
  • Read by sound, using grapheme-to-phoneme conversion
  • No difficulty with regular words and non-words
  • Difficulty with irregular words (e.g. yacht)
  • Associated with anterior temporal lobe atrophy
  • Damage to direct route
• Phonological Alexia
  • Able to read previously learned words (regular or irregular) via the direct route
  • Extract the meaning directly from the visual form of the word
  • Difficulty reading new words and nonwords (pseudowords) both regular or irregular
  • Associated with lesions to superior temporal lobe, parietal lobe (supramarginal gyrus, angular gyrus)
• Deep Alexia
  • Semantic substitutions (related to target word, e.g., “hot” for “cold”)
  • Influence of “imageability” of word: more difficulty with abstract words
  • Better with nouns than verbs (imageability?)
  • Visual errors (e.g. skate for scale)
  • Damage to both direct and phonological routes

Brain Regions Involved in Reading

• The “Visual word form area” (VWFA)
  • Left fusiform gyrus (lesions produce alexia)
  • EEG shows N170 (100-200ms) response to letter vs symbol strings
• Writing is only ~5000 years old, so how did our brain evolve a specialised region for orthographic processing so quickly?
• Reading requires coordinated activation in areas responsible for visual word form recognition, phonological and semantic processing (reading correct vs scrambled sentences; Yarkoni et al., 2008)

Agraphia

• Inability to write or spell while writing
• Subtypes
  • Central dysgraphia refers to problems accessing orthographic information from lexical stores or from applying sound-to-spelling phonological rules (dual route)
  • Peripheral dysgraphia reflects distortions in written letter formation, oral spelling or typing (motor programs)

Language and Attention

• Attention and language are two of the most researched topics in experimental psychology today
• In the late 19th century, aphasiologists such as Broca, Wernicke and Lichtheim all considered processes of speech perception and production to be automatic
• Experimental psychologists such as Wilhelm Wundt instead proposed language required an additional attentional control system
• The first experimental psychology laboratory was devoted to psycholinguistics!

Wernicke-Lichtheim-Wundt model

• According to Wundt (1880, 1902), a central attention “apperception” system located in the frontal cortex controls a word production and perception network centered around perisylvian cortex.
• Production begins with a complete thought that becomes sequentially organized and articulated. The comprehension process is the same sequence in reverse, proceeding from sound segments to the complete thought.
• Wundt’s (1880, 1902) model. reproduced by Roelofs (2021)

Language and Attention

• Speech production is often assumed to be an automatic or effortless process. Errors are rare, on average 1-2 per 1000.
• Yet we need to ensure what we say is contextually appropriate.
• Ensuring the production of a contextually appropriate response is considered the cardinal feature of ‘cognitive control’.
• Multitasking
• Performing some tasks while speaking requires quite a bit of control, e.g., talking on a mobile OR hands-free impairs driving ability equally (Strayer & Drews, 2007)

Speech production

• Computational models implemented in the 1990s did not include attention control systems for spoken word production… Indefrey & Levelt (2004).
• Articulation Picture presentation 0 ms
  Conceptual preparation
  Lemma selection
  Phonological code retrieval
  Syllabification
  Self monitoring
  Self monitoring

Attention and Speech Production

• Comprises 2 linked systems (Botvinick et al., 2001)
  • a system that monitors conflict in information processing, and
  • a compensatory control system that is engaged when conflict is detected.

Attention and Speech Production

• Botvinick et al. (2004).

• Neuroimaging studies of Vocal, Manual and Oculomotor tasks involving conflict (e.g., error inducing paradigms, response bottlenecking) reliably activate the anterior cingulate cortex (ACC).

• Control then implemented by the frontal cortex to resolve conflict

• A ‘domain general’ system.

• Speech production requires some involvement of domain general mechanisms of attention under some circumstances – processes up to and including phonological encoding in word planning delay, or are delayed by, the performance of concurrent unrelated non-linguistic tasks

• Yet, conflicts in word planning may be resolved while concurrently performing an unrelated non-linguistic task, making a task decision, or making a go/no-go decision

• So it doesn’t require full involvement of domain general attention capacity to be achieved de Zubicaray (2023)

Attention and Speech Perception

• Language comprehension (i.e., the process of extracting meaning from the linguistic signal) tends to be a relatively automatic process
  1. Are there circumstances when a domain general control system is necessary?
     • e.g. bilingual switching
     • but what about simultaneous interpreters?
     • e.g., syntax - “the ball that the boy kicked”
     • but these are rare forms
  2. Are there circumstances when language comprehension proceeds without conscious awareness, i.e., without directed attention?
     • There are numerous studies purportedly demonstrating the ability of unconscious processing to semantically analyse single to multiple word strings
     • but just as many have not.

Domain Generality of Language?

• Domain generality vs. domain specificity, i.e., language might not be unique
• Language function has typically been associated with a primarily left-hemisphere network of frontotemporal brain regions (Wernicke-Lichtheim-Geschwind model)
• Yet many neuroimaging studies report activity in dorsal frontal, parietal and medial temporal lobe regions during language processing tasks that are known to be engaged by attention and memory
• Attention and memory are considered 'domain-general' systems
• So, either:
  • Domain-general systems are required for language processing, or
  • Domain-general systems are not required for language processing and are instead an artefact of the tasks typically used to study language

Domain Specific Language

• “Broca’s area contains two sets of subregions lying side by side, one quite specifically engaged in language processing, surrounded by another that is broadly engaged across a wide variety of tasks and content domains.” (Fedorenko et al., 2012).
• A domain general attentional mechanism
• Consistent with Wundt’s (1880, 1902) model!

Domain Specific Language

• Multiple demand (MD) network:
  • a domain-general bilateral frontoparietal network implicated in executive functions
• Does the MD network contribute to language comprehension?
• Diachek et al. (2020):
  • large-scale fMRI investigation using diverse word and sentence comprehension experiments
  • passive comprehension tasks failed to elicit a response above the fixation baseline in the MD network

Domain Specific Language

• What about music and language perception?
• Two human-unique capacities
• Chen et al. (2023):
  • 3 fMRI experiments in English and Mandarin (a tonal language) using sentence comprehension and nonword perception
  • orchestral music, single-instrument music, synthetic drum music, and synthetic melodies, etc
  • language regions’ responses to music were generally low, often below the fixation baseline, and never exceeded responses elicited by nonmusic auditory conditions.

Language Without Awareness

• Can language comprehension occur nonconsciously?
  • Equivocal evidence from masked single word primes
    • Due to limited processing time?
• Subliminal presentation of sentences with semantic violations
  • Continuous flash suppression consists of a presentation of a target stimulus to one eye and a simultaneous presentation of rapidly changing masks to the other eye. The rapidly changing masks dominate awareness until the target “breaks into” consciousness. (Sklar et al., 2012; PNAS)

Language Without Awareness

• Subliminally presented sentences with semantic violations broke suppression faster than conventional sentences
  • (e.g., “I ironed coffee”; “the bench ate a zebra”)
• Masked expressions with affective meanings as primes
  • The more negative a verbal expression was, the faster it became conscious
• Evidence from masked priming has been used to draw conclusions about the automaticity of syntactic processing
  • But low statistical power and low reliability of visibility measures may also contribute to the results (Hernández-Gutiérrez, D et al., 2025).

Language and attention - summary

• Language production sometimes requires domain general attentional control, but does not rely on it
• Language comprehension seldom requires domain general attentional control
• And might not even require conscious awareness
• Language influences perception
• Directs attention to gain conscious awareness

Cortical Representation of Word Meaning

• Models of word meaning beginning with Lichtheim (1885) proposed abstract or amodal representations distributed throughout the cortex.
• “…in the diagram B is represented as a sort of centre for the elaboration of concepts, this has been done for simplicity's sake; with most writers, I do not consider the function to be localised in one spot of the brain, but rather to result from the combined action of the whole sensorial sphere. (p. 477; italics added).”

Cortical Representation of Meaning

• Distributed models do not localize meaning representation to any particular cortical structure
• Distributed-plus-hub models propose a key role for the anterior temporal lobe (ATL) in mediating amodal conceptual representations
• Lesion evidence from semantic PPA/semantic dementia, a progressive neurodegenerative disorder characterized by loss of semantic memory
• Patients lose the ability to match words or images to their meanings

ATL Hub: healthy older adults

• ATL peak atrophy: -44, 14, -34

• Two ventral white matter pathways:

  • Inferior-fronto occipital fasciculus
  • Uncinate fasciculus
  • Replication of the region of atrophy in semantic variant primary progressive aphasia (PPA; aka semantic dementia) in two independent patient samples

• Replication of relationship between semantic composite scores and inferior fronto-occipital and uncinate fasciculi in patients with brain lesions

• Connectivity between ATL hub and anterior and posterior cortical regions

Ventral and Dorsal Language Pathways

• “Dual Stream” model of speech/language processing (e.g., Hickok & Poeppel, 2007).
• Dorsal pathway
  • interfaces sensory/phonological networks with motor-articulatory systems
  • arcuate fasciculus
• Ventral pathway(s)
  • interfaces sensory/phonological networks with conceptual-semantic systems
  • inferior fronto-occipital fasciculus (IFOF)
  • uncinate fasciculus

Grounded Models

• Language and action are associated
• Grounded language models propose word meaning is represented in the same sensory or motor structures responsible for mediating perception and action
• A prominent example is “semantic somatotopy” in primary and premotor cortex, e.g., Pulvermüller (2005).

Hebbian Association

• “any two cells or systems of cells that are repeatedly active at the same time will tend to become ‘associated’, so that activity in one facilitates activity in the other” (Hebb, 1949, p. 70).
• The frequent co-presentation of an action word with the execution of the action the word refers to might result in a representation being associated with those motor areas (Pulvermüller, 1996, 2005).
• But… “Pair a tone with a puff of air to the eye and pretty soon the tone alone will elicit a blink response. This motor response to a sensory event does not indicate that the blink response is coding the meaning of the tone.” (Hickok, 2009)

Mirror Neurons

• Macaque premotor area F5 (a homologue of human BA44 or Broca’s area) contains a class of visuomotor neurons that respond congruently when goal-directed mouth or hand actions are both observed and executed.
• This mechanism might transform visual information into knowledge coded at an abstract level, i.e., it might be involved in understanding action meaning (Gallese & Lakoff, 2005)

Grounded Action Representations?

• Consider the following phrases:
  • “grasp the cup” = hand motor area
  • “kick the ball” = foot motor area
  • “I see what you mean” = visual area
  • “flew past me” = visual motion areas
  • “hear the music” = auditory areas
• Communicating the action meaning in each phrase involves an indirect reference to the specific effectors (body parts) that perform the action
• Thus, meaning is “grounded” in simulations or interactions between perception, action, the body, the environment, and other agents, typically during goal achievement.

Or Abstract Representations?

• Consider the following sentences:
  • “The fish swam across the lake”
  • “The duck swam across the lake”
  • “The snake swam across the lake”
  • “The man swam across the lake”
• Different effectors are used across species to accomplish the action
  • The fish uses motions of the body with dorsal and caudal (tail) fins under water
  • The duck uses movement of webbed feet beneath the surface of the water
  • The snake uses undulating, lateral body movements on the water surface
  • The human uses coordinated arm and leg movements above or below the surface
• Thus, you must have a representation of what it means “to swim” that is abstracted from the body performing the action in order to comprehend all the sentences.

Mirror Neurons and Action Prediction

• Action understanding in a dog
• “Here is an example of a dog predicting the outcome of a throwing movement in a game of fetch. He is not simply reacting to the trajectory of the ball as the ball is not released on the first throwing action of each trial; the ball is thrown only after the dog turns to run in the predicted direction of the throw. Critically, dog's can't grasp or throw balls, so the "mirror system" is not the basis of this understanding. Conclusion: you don't need mirror neurons to understand and predict actions. (Hickok, 2011)”

Cortical Representation of Meaning

• A large-scale cortical semantic network (Binder et al., 2009)
• Meta-analysis of 1135 activation foci from published neuroimaging studies
• “Given the presence of mirror neurons in premotor cortex …it is somewhat surprising that frontal cortex did not show consistent activation in studies of action or artifact words.”

Cortical Representation of Action Meaning

• Review of neuroimaging studies of action word comprehension (de Zubicaray et al., 2013)
• Approximately chance probability of reported peak actually being in cytoarchitectonically-defined premotor (BA6; orange) or primary motor (BA4; yellow) cortex

Cortical Representation of Action Meaning

• Lesion evidence indicates action execution and action understanding in humans are dissociable
• People with congenital limb dysplasia show similar dissociations
• Patients with limb apraxia are able to comprehend the meanings of pantomimed actions (Negri et al., 2007)
• Patients with lesions to mirror system areas are able to comprehend the meanings of action words (Arévalo et al., 2010, 2011)
• Patients with lesions to mirror system areas are able to generate meaningful words to describe actions (Reilly et al., 2014)

Grounded Perceptual Representations

• Language and multisensory perception are associated
  • McGurk and phoneme restoration effects
• According to grounded accounts, the meanings of words describing motion are partly represented in the perceptual areas that process the actual visual stimuli the words describe (e.g., Barsalou, 2008)
• “flew past me” = visual motion areas
• Neuroimaging studies have reported motion sentences activate visual cortical area MT+/V5 compared to non-motion sentences
• “The wild horse crossed/stood in the barren field.” (e.g., Saygin et al., 2010)
• Might simply reflect Pavlovian association (e.g., Mahon & Caramazza, 2008)
• Tasks requiring matching of verb motion types (hopping & skipping vs running) do not result in activation of MT+/V5 – a stronger test of the hypothesis (Kable et al., 2005)

Grounded Perceptual Representations

• Behavioural studies have demonstrated language influences motion perception
• Responses to motion stimuli preceded by congruent motion-related (e.g., “rise”) words are faster and more accurate
• Stronger effect when words presented in right visual field (RVF)
• Corresponding to the language dominant left hemisphere (e.g., Gilbert et al., 2008)
• If the grounded account is correct, this effect should activate visual cortical area MT+/V5

Cortical Representation of Meaning

• fMRI study of language-mediated motion perception using LVF and RVF presentation
• BUT no modulation of cortical motion perception area MT+/V5
• Contrary to the grounded account
• Modulation of left mid-temporal cortex (lexical-semantic representations) (Francken et al., 2015)

Meaning and the Brain: Summary

• There is considerable evidence for the proposal that word meaning is abstracted away from modality-specific perceptual and motor areas of the cortex
• ATL hub-and-spoke models are currently supported by the most empirical evidence
• Connectivity between sensory/phonological networks and conceptual-semantic systems is likely mediated by ventral white matter pathway(s)
• “Dual stream” model
  • Inferior fronto-occipital and uncinate fasciculi
• The evidence for grounded models of meaning representation is weak at best
• Neither Hebbian association nor Mirror System engagement are sufficient explanations