Cog Neuro [Language]

What is phonology

the sounds that make up words

what is syntax

how sentences are structured

what is semantics

what words mean

what is pragmatics

what might be implied by how the person chooses to use their words

what are impairments in language broadly referred to as?

aphasias

• aphasias are generally impairments of both written and verbal speech

What is anomia

• a fluent form of aphasia

impaired

• characterized by a difficulty naming and finding words

• know what the word is but cannot retrieve it → dissociates object knowledge from retrieval of linguistic label)

intact:

• affected patients still have fluent language and excellent comprehension/semantics

What is Broca’s Aphasia?

• a non-fluent/expressive aphasia

  • can understand language though if you give them even moderately complex sentence structures sometimes they have more subtle comprehension deficits

    • agrammatic aphasia → trouble understanding grammatically complex sentences

  • rhythm of their speech is very natural

  • but they do not express actual content when they talk

• extreme cases → patients just repeat the same nonsense word over an over but with the intonation and rhythm of normal speech

• less extreme cases → telegraphic speech: speech is focused on key content words

  • e.g. “son…university…smart…boy…good…good”

• oftentimes involves damage beyond the Broca’s area (esp. the insula)

Where is Broca’s Area

near the motor strip at the back of the frontal lobe; at the inferior part of the frontal lobe, close to the temporal lobe

What is Wernicke’s Aphasia

• the inverse of Broca’s apahsia

  • patients have trouble comprehending spoken or written language

  • speech production is fluent + follows a natural rhythm but is nonsensical

  • word choice is not entirely random, sometimes words are semantically related to intended ones (semantic paraphasia)

• Doesn’t require damage to Wernicke’s area!

  • typically involves surrounding posterior temporal lobe or underlying white matter (arcuate fasciculus)

Where is Wernicke’s Area Located

in the superior temporal lobe, near where the auditory stuff is

What is conduction aphasia?

patients have difficulty repeating back words/sentences

they hear their errors but can’t correct them → thought to be a disruption between understanding and production

• motor production is reasonably well intact

• understanding is intact

• have specific impairment in repeating words back

What are some other language production deficits?

speech apraxia → difficulty generating articulator motor plan

dysarthria → trouble controlling articulatory muscles in their mouth and tongue (lower level motor deficit)

What are some other language comprehension deficits`

pure word deafness → inability to recognize speech presented auditorily

alexia → inability to recognize words presented visually

What are the key stages of word processing

General Overview:

• starts with two streams of input

  • spoken word

  • written word

• do some type of acoustic or visual analysis

• Access the word concepts

Three Main Stages:

1. Lexical Access → assumes that you got the output of perceptual analysis and now a word representation is being activated in your mental lexicon (your mind’s dictionary) and the semantic (related words) and syntactic (grammar) attributes go with that word.

2. Lexical Selection → finding the best match in your lexicon

3. Lexical Integration → integrating words into the rest of a sentence or some type of broader context, in order to understand what the overall message of the speech is

leads to speech production → actually getting your motor system to say the word

What is a mental lexicon?

• a collective store of not just the word itself but:

  • its semantics → what it means

  • its syntax → how it fits in with other words structurally (grammar)

  • its orthography → how it’s written

  • its phonology → the sounds that make up the word

• We match percepts and language to our mental

What are phonemes

the smallest unit of speech sound that a listener can distinguished

• e.g. /f/

what are morphemes?

the smallest unit that signifies meaning; consist of multiple phonemes

• e.g. ‘frost’, ‘de-’

What are the organizing principles of the mental lexicon

• semantically related lexical representations are located nearby to one another

• phonemes that have similar sounds also share connections (e.g. meat, beat, seat)

• activation of a given unit spreads to nearby units

  • this means that both phonemic and semantic priming effects can occur

not a theory about the brain → theory about the mind

Describe the early models of semantic networds

nodes → represent concepts based on discrete categories, relate to each other via hierarchical ordering

• nodes are connected logically through propositions like ‘has’ or ‘can’

  • e.g. “ a living thing can grow)

• propositions get inherited down the hierarchy

  • if a living thing can grow, that also mean an animal=[ can grow, a fish can grow, and a salmon can grow

  • new attributes get added along the way

What are the strengths of the early semantic networks models?

• it is efficient to store things this way because of the inheritance property

  • just need to know that a salmon is a fish and a fish is an animal to know a lot about what a salmon can and can’t do

• can generalize to new concepts → can easily add in new pieces of information and have existing information apply to them

What are the problems of the early semantic network's models?

• they don’t account for exceptions

  • e.g. birds that can’t fly

• in practice, people don’t have equal access to all of these features → some properties are easier to retrieve than others

  • e.g. “oak is tal” vs. “oak has bark”

• model does not account for the fact that people don’t remember all items the same → some items are more canonical representations than others

What is the connectionist approach to semantic networks

• there is no strict hierarchy that is organized, instead there is a way to learn everything from scratch

  • whats learned is distributed across the model

• the model will learn based on the natural co-occurrence of words in natural language and semantic features (e.g things that have 4 legs often run)

• learns things from experience such that after it learns it can activate correctly associated words

• relies on the idea that we have these different attributes that are stored seperately from these individual concepts → through learning we learn the link between the attributes and the concepts

  • this allows properties to be associated probabilistically → allows for different strengths of associations (and that there are more typical examples of certain concepts)

• captures generalization through pattern association

What is the classical model of semantic networks in the brain

• things like attributes, properties, and features are going to be represented in the part of the brain that processes those attributes, properties, and features

  • e.g. v4 is going to activate when we think about things like color

What is the evidence for the classic model of the semantic network

• if you show people words of colors → activates ventral, posterior color regions like v4

• if you show people action words like swim → activates more dorsal, posterior temporal regions that code more for actions

• if you read action words involving a body party, you’re going to get activation in the motor strip that corresponds to movement of that body part

people noticed early on from lesion patients that sometimes people present with categorical anomias

• categorical anomia → inability to name objects in specific categories

• deficits naming tools, animals, and people go from posterior to anterior temporal cortex

  • people → most anterior

  • animals → in the middle

  • tools → most posterior

• similar type of topography observed with PET scanning of healthy subjects

this suggests that there is some type of segregation of this information in the ventral temporal cortex

What is the alternate view of semantic networks

• Instead of segregating information by living/non-living distinctions, maybe the anterior temporal love is critical for specific-level naming

  • because

    • Non living things are distinguished by fewer features than living things

    • and specific naming requires more retrieval and integration, moreso for living

study

• give participants a set of the same images but ask them to classify them at different levels of specificity

  • e.g. is it a car or a dog (more specific) versus is it an animal or a vehicle (less specific)

results

• more posterior regions of the temporal cortex are activated by both specific and general naming

• as you move anterior it is only the specific level naming that causes activation

the most anterior part of the temporal lobe may be necessary for representing these very specific exemplars and linking them to the relevant features and attributes

What is semantic dementia

degenerative atrophy that starts in the temporal pole and the progresses posteriorly and medially

• leads to impairments in finding the appropriate word and in comprehension

  • in early stages it looks like anomia but then it progresses to become much more severe where they can’t even access the categories (impairment goes from impairing exemplars to impairing categories)

    • additional support for the alternate view of semantic networks

What are the impairments of someone with another variant of frontal temporal dementia that more aggressively affects the frontal lobe?

• see a lot of personality and mood changes with some control changes

how do we understand spoken language?

• go from acoustic processing to speech → mediated through the auditory processing system

• hierarchy of selectivity to acoustics vs. speech

  • primary auditory cortex → responds to all sounds, non-selective for speech

  • superior temporal sulcus (STS) → selective for speech sounds (words, non-words that sound like words, and reversed speech)

    • impairment to STS leads to pure word deafness → trouble recognizing speech sounds but not other sounds, otherwise intact reading and other types of language abilities

how do we understand written language

• people have not been writing for very long so we likely do not have dedicated brain circuits that have been evolved for reading

• parts of the object recognition streams that are generally good at recognizing objects have been co-opted to specifically recognize letters

  • orthographic input gets process through the visual system (V4) the same as anything else → then lands in an area called the visual word form area (VWFA)

VWFA us sensitive to things that look like words and semantic content

• keyboard mashes do not activate the VWFA

• selectivity emerges as you go further in to the visual processing system

What are the cultural differences in learning to read?

• English: complex mapping of orthography to phonology → so learning to read is hard → increased developmental dyslexia rates

  • 1120 letter/ letter combinations to 40 sounds

• Italian: simple mapping - 33 letters to 25 sounds → all children lear to read in about 6 months → developmental dyslexia rate is half that of English

people who speak different languages have different reliances on orthography vs. phonology

• English > Italian → VWFA: greater reliance on orthography (spelling of word)

• Italian > English → Second auditory: greater reliance on phonology (how the word sounds)