Overview of Language Processing Models pt.1

communication process

We can understand and produce language by thinking of the \____ \____ and describing the steps in terms of anatomy and physiology

THE SPEECH CHAIN

communication process can be seen in

cyclical

THE SPEECH CHAIN: communication is depicted as a \___ process

motor

THE SPEECH CHAIN: start with brain, sending \____ signals to vocal muscles

sensory

THE SPEECH CHAIN: produces speech sounds, that the listener's ear receives and transmits to the listener's brain through \___ signals

feedback

THE SPEECH CHAIN: While the aforementioned process is happening, the speech sounds produced by the speaker are also \____ to the speaker's ear which then transmits it into the brain of the speaker through sensory signals.

speech

RECALL: language can be conveyed through \___ & language is separate from __

Brain and vocal muscles

\___ \___ \___ __ covers both language expression and speech production^

THE SPEECH CHAIN

This simplified model makes it hard to say where the delineation is, Also difficult to observe in clinics, More anatomical and physiological in nature

SPEECH-LANGUAGE PROCESSING MODEL (SLPM; Nation & Aram, 1984)

so we then replace the speech chain into a behavioral model so that it is more observable

sensed

SLPM (Nation & Aram): instead sound reaching ear, we say that they \___ and hear the sound

perceived

SLPM (Nation & Aram): brain analyzed the speech sound → person \___ or recognized as speech

comprehended

SLPM (Nation & Aram): brain decoded the meaning of sound waves → person \___ or understood the speech or utterance they heard

Sensation, Perception

SLPM (Nation & Aram): Auditory Reception

Sensation

SLPM (Nation & Aram): Sound waves successfully reached the person's ears; sense or hear that utterance, you don't know what you heard but you detected presence of the sound

Perception

SLPM (Nation & Aram): perceive or recognize it as speech and not environmental sound or speech, high speech was perceived that was produced by a girl

Comprehension, Integration, Formulation

SLPM (Nation & Aram): Central language-thought

Comprehension

SLPM (Nation & Aram): finally understand superficial meaning of utterance, as you have long hair, if we stop you won't understand meaning

Integration (receptive side)

SLPM (Nation & Aram): integrate with all prior experiences and information from the situation or the context to understand it fully, example: receive flowers from 3 different admirers and integrate with superficial meaning and how others use that line: you must be so pretty and attractive

integration (expressive side)

SLPM (Nation & Aram): speech production also starts with integration, or using prior experience or knowledge of the context you formulate an appropriate message, intergate what you will say with the idea you just said, you won't repeat anything you said

Formulation

SLPM (Nation & Aram): with the prior knowledge and experiences, formulate an appropriate message according to integration

Sequencing, Motor Control

SLPM (Nation & Aram): Speech Production

Sequencing

SLPM (Nation & Aram): identify individual sounds you will need to convey that message, first word: kaya , k, a, y, a, you have to arrange in the right sequence

Motor Control

SLPM (Nation & Aram): move articulators (lips, mouth, tongue) to produce all sequence of speech sounds

sensation, perception, comprehension, integration

SLPM (Nation & Aram): receptive processes:

integration, formulation, sequencing, motor control

SLPM (Nation & Aram): expressive processes:

information received from the sensory or auditory reception segment

COGNITIVE-LINGUISTIC SYSTEM AKIN TO A MACHINE: Raw materials:

the output comes to it; the formulated message or speech production

COGNITIVE-LINGUISTIC SYSTEM AKIN TO A MACHINE: Processed materials:

utterance heard (sensation)

COG-LIN SYSTEM AKIN TO A MACHINE: Language Comprehension: raw material \=

meaning of what was heard (comprehension/integration)

COG-LIN SYSTEM AKIN TO A MACHINE: Language Comprehension: processed material \=

meaning speaker wants to convey (integration/formulation)

COG-LIN SYSTEM AKIN TO A MACHINE: Language production: raw material \=

speech

is modality to which language is expressed, hard to separate them clean

utterance or series of articulated speech sounds (motor control)

COG-LIN SYSTEM AKIN TO A MACHINE: Language production: processed material \=

very broad

COG-LIN SYSTEM AKIN TO A MACHINE: limitation: comprehension and formulation:

Info Units, Info Storage and Utilization, Info Processing

THE COG-LING SYSTEM'S MYSTERIES

INFO UNITS

THE COG-LING SYSTEM'S MYSTERIES: Sentence vs. words vs. morphemes vs. ??? Do we process language per word, per morpheme, per sentence or all in one go?; are all the ingredients processed at once passing thru machine (sentence), does it do so in packets with one type of material (words), does it process based on parts of raw materials (morphemes), does it process leave and then stem and then root of morus alba

INFO STORAGE AND UTILIZATION

THE COG-LING SYSTEM'S MYSTERIES: oolong start, while processing the dandelion sa start, they need to save it and store in the encircled yellow, how does machine store these materials for later use (previously processed and learned words, how are these stored in our memory)

INFO PROCESSING

THE COG-LING SYSTEM'S MYSTERIES: How exactly do we manage to understand or produce words?, What steps/sub processes are involved?, How exactly does the machine work?, What mechanical operations or function the machine is built with, allowing to successfully process raw materials

Serial/modular | vs. Parallel (cascading, | interactive)?

(1) ISSUES IN THE STUDY OF LANGUAGE PROCESSING

Serial

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: Language processes happen one after another, does comprehension start only after perception

Parallel: Cascading

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: Succeeding process starts while the previous process is still ongoing, does it overlap, is it one way?

Parallel: Interactive

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: Flow of information is two-way, if it is indeed overlapping, is it two way like that one

Bottom-up | vs. + Top-down?

(2) ISSUES IN THE STUDY OF LANGUAGE PROCESSING

Bottom-up

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: Is the flow of info really one way? toward the bottom, to the up being advance; One-way, going forward

Top-down

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: Certain types of information flow backwards, is there feedback, does info sent from the top to somewhere further down the line, slpm example: does contextual info processed in integration stage help in understanding sentences at a superficial level (comprehension stage)

Local | vs. Distributed?

(3) ISSUES IN THE STUDY OF LANGUAGE PROCESSING

Local

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: assume info processed as words, are words stored in one unit of brain: doll is right there, dog is there, and hair there

Distributed:

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: is it stored in different units of the brain, and stored in different units at the same time, dog, in one basket, phonological info stored in another basket or unit,

Domain-specific | vs. domain-general?

(4) ISSUES IN THE STUDY OF LANGUAGE PROCESSING

Domain-Specific

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: restricted to single unit or department specialized for language function

Domain-General

ISSUES IN THE STUDY OF LANGUAGE PROCESSING: language and cognitive housed in the same unit, sharing resources and mechanisms

sensation, perception, and comprehension

According to the SLPM, we understand words by going through the processes of \___ \___ \___ respectively.

comprehension

broad process, don't know what happens within it

FORSTER'S AUTONOMOUS SERIAL SEARCH MODEL (1976; 1979)

basically words are categorized, and to find them, you have to find them serially, think of minds as library, consider used to have card catalog cabinets and shelves

access files

FORSTER'S AUTONOMOUS SERIAL SEARCH MODEL (1976; 1979): instead of title, subject, and author cards, \___ \___ may contain morphological, phonological, semantic, and information of a word we are trying to recognize

master files

FORSTER'S AUTONOMOUS SERIAL SEARCH MODEL (1976; 1979): shelves, can be labeled as \___ \___ section, where we want to get to, pertain to the words that is stored in our minds (we called this as lexicon or mental lexicon)

serial

FORSTER'S AUTONOMOUS SERIAL SEARCH MODEL (1976; 1979): wrong access file: you thought it is waffle, after you checkout, you move on to next possibility that is why it is called \___, you find words one after the other

FORSTER'S AUTONOMOUS SERIAL SEARCH MODEL (1976; 1979)

It is not the predominant model these days. Researchers believe that we access multiple words all at once instead.

ONE VERSION OF MORTON'S LOGOGEN MODEL (Morton & Patterson, 1980)

access words all at once, this process is efficient for us to recognize what it means; "logogen" in the original model (Morton, 1969, 1970)

logogen

unit theorized in our brain where each word is associated with one logogen

resting activation

ONE VERSION OF MORTON'S LOGOGEN MODEL: all of the logogens have \____ \___, even without any external stimuli, and they are activated to a small degree

Activation

ONE VERSION OF MORTON'S LOGOGEN MODEL: happens with increasing evidence

activated

ONE VERSION OF MORTON'S LOGOGEN MODEL: moment you hear a, all logogens that correspond with the word a, will be sort of \____, logogen of apple will get more activated as more morphemes are said

threshold

ONE VERSION OF MORTON'S LOGOGEN MODEL: each logogen has a \____: if amount of activation corresponds to that logogen, if you receive more and more evidence threshold will be reached and you will recognize word and understand what it means

threshold

ONE VERSION OF MORTON'S LOGOGEN MODEL: \____ of a logogen is reached whenever there is enough activation or evidence that points to a word being the one you heard. This leads to the word being recognized and understood.

lower

ONE VERSION OF MORTON'S LOGOGEN MODEL: high frequency words, words you use a lot or hear a lot (e.g., your name) have \___ threshold, meaning that they need less evidence to get recognized by our brains

hear

(1) PREVAILING VERSION IN THE SLP LITERATURE: First \___ a word

auditory phonological analysis

(2) PREVAILING VERSION IN THE SLP LITERATURE: You analyze phonological components (\___ \___ \___), what speech sounds is composed of, what phonemes those speech sounds correspond to

Phonological Input Lexicon

(3) PREVAILING VERSION IN THE SLP LITERATURE: Look for its match in your mental lexicon/ library, based on what you heard and look in your mental library Similar to Foster's Serial Search Model however the lexicon is only specific to phonological input/how words sound like. (\____ \___ __)

Semantic system

(3) PREVAILING VERSION IN THE SLP LITERATURE: You retrieve that word or meaning of that word to understand what it means (\___ \___)

FULL-LISTING MODEL

all derivations listed in our mental lexicon, are each of these derived words listed as one entry in our mental lexicon like these?

OBLIGATORY DECOMPOSITION

only the root is listed + rules to be applied→ "affix stripping", is it only the root word listed in our mental lexicon?, you just follow form of rules listed in your entry, based on what kain means and what affix to use or when it should be applied, analyze based on roots and affixes (affix stripping)

affix stripping

OBLIGATORY DECOMPOSITION: only the root is listed + rules to be applied→ "\___ \___"

decomposition

OBLIGATORY DECOMPOSITION: It is called \___ because we break down the words into their respective components (roots and affixes) to understand what they mean separately first before figuring out what they mean altogether.

DUAL-PATHWAY HYPOTHESIS

An intermediate take, Obligatory decomposition. except for commonly used derivations (fully-listed), word recognition is primarily of the obligatory decomposition type, (second take), except for commonly used derivations (full-listing), case for words which are roots, and roots that infrequently use derivation, derived words that is commonly used - andito, andoon, andiyan

EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987)

Semantic System, Phonological Output Lexicon

EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): FORMULATION

Semantic System

(1) EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): thinking of an idea or concept from your semantic system

Phonological Output Lexicon

(2) EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): based on it you look for the most appropriate entry in your mental lexicon, entries in mental lexicon are in form of phonological form wherein you have phonemes,

or a sequence of phonemes (words), even if associated with semantics, syntactic, and phonological information,

word production and not comprehension

EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): output:

phonological output buffer

EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): FORMULATION/SEQUENCING

phonological output buffer

(3) EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): assemble that phonological word form by retrieving its components phonemes, get each of these and sequence and assemble them, arrange themselves

buffer point

(3) EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): sounds are waiting in line to be produced by the articulators,

Spoken Word

EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): MOTOR CONTROL

spoken word

(4) EXTENDED LOGOGEN MODEL (Morton, 1985; Patterson & Schewell, 1987): articulation, apple as the output

Conceptual-semantic representation

(1) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986) start with an idea, called different term from semantic system, deciding what concept you wanna convey

Semantic Feature Network

(2) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): run it through this network, theorized to be organized in a network of sorts, visualize that: map of semantic features: each concept you want to try convey, certain parts of map light up, correspond to the bigger idea

semantic features

(2) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): bits of meaning that each word in our mental lexicon is associated with

network

(2) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): \___ because there might also have other concepts that are in the same SF or category

Lexical Network

(3) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): network of words or terms refer to the meaning stored in the mind, Needs increasing evidence to trigger the specific word

tip of the tongue

(3) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): problem in this stage is experiencing \___ \___ __ \___ moments

Phonological Network

(4) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): map of phonemes connected to one another in our brains, phonemes that match the lexical entry or term or word, seem to match lights up/ gets activated, They are overlapping and interact with each other; they help you to pinpoint what lexical entry you are looking for

Phonological - encoding

(5) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Retrieving all phonemes that were activated and assemble and sequence to articulate the correct word

Articulation

(6) DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): using vocal muscles to say the word

two, interactive

DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Stages aren't very discrete, a bit overlapping, interacting with one another, arrows are pointing at __ directions (middle and right illustration), it is an \___ model

Spreading activation

DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Key processes / concepts: nature of activation is very dynamic because when one semantic feature activates, its as if it will spread in rapid manner in surrounding semantic features, like the example of apple & heart, activation will spread throughout semantic, lexical, phonological network

Competition

DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Key processes / concepts: interesting because it hypothesizes that when we try to retrieve best words to convey idea, there might be other candidates considered (ample, waffle) - they won't win because apple gets activated the most

Decay

DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Key processes / concepts: when a semantic feature is activated, (eg., semantic feature of being sweet) power or amount of activation will slowly decay unless semantic feature gets extra activation from other parts of interactive system, These features should sustain activation enough until you produce the world successfully

Feedback

DELL'S INTERACTIVE ACTIVATION MODEL (1997; 1986): Key processes / concepts: idea information from later processes (phonological processing, phonological network), think of word know what sounds feedback to earlier stages (lexical network stage), identify which is the intended idea