SLHS 50100 Test 2

Chapter 5: Speech Perception

Study Questions

When you hear a word like raspberry, which parts of your brain underlie the hierarchical processing of that word’s hierarchical sound structure?

Extends from the dorsal surface of the STG down around its lateral surface and into the upper lip of the STS

What evidence suggests that the early cortical stages of speech perception involve not only the left hemisphere but also the right?

• ”Asymmetric sampling in time”

• Left hemisphere is dominant for processing rapid auditory variation in the 20-80 ms range which is ideal for registering and classifying fine-grained distinctions at the phonemic level

• Right hemisphere is dominant for processing longer-duration auditory patterns in the 150-300 ms range, which is ideal for tracking speech input at the syllabic level

Why was the Dual Stream Model originally motivated by classic findings about double dissociations between comprehension and repetition?

Initial evidence for separate streams comes from neuropsychology, since brain-damaged patients exhibit double dissociation between, on the one hand, the ability to comprehend utterances and, on the other hand, the ability to repeat utterances or closely monitor their phonological makeup

In both cognitive and neural terms, how does the ventral stream contribute to your perception of a word like raspberry?

• ”What” pathway

• From sound to meaning

• Allows the listener to understand the conceptual content of utterances

• Lexical interface

• Combinatorial network

In both cognitive and neural terms, how does the dorsal stream contribute to your perception of a word like raspberry?

• ”How” pathway

• From sound to action

• Allows the listener to link speech perception with speech production

• Sensorimotor interface

• Articulatory network

Why has there been a great deal of controversy over whether the articulatory network facilitates speech perception?

• Gregory Hickok believe that while the articulatory network might modulate the perception of speech in various ways, it is probably not a necessary resource for comprehension

• Opposing studies believe that the articulatory network does make a nontrivial functional contribution to receptive speech processing

Summary and Key Points

Hierarchically Organized Processing

• The dorsal STG carries out fairly simple spectrotemporal analyses

• The mid-to-posterior lateral STG represents the subphonemic features and feature combinations

• The mid-to-posterior STS represents individual phonemes and the sequential phonological structures of whole words

• Bottom-up triggered by acoustic stimuli

• Top-down influenced by prior knowledge and expectations

• Segmentation and identification of phonological structures that have different durations is facilitated by the entrainment of electrophysiological oscillations that have correspondingly different timescales

Bilaterally Organized Processing

• Both hemispheres are recruited in different ways

• ”Asymmetric sampling in time"

• Left hemisphere is dominant for processing rapid auditory variation in the 20-80 ms range which is ideal for registering and classifying fine-grained distinctions at the phonemic level

• Right hemisphere is dominant for processing longer-duration auditory patterns in the 150-300 ms range, which is ideal for tracking speech input at the syllabic level

Dual Stream Model

• After the early cortical stages of speech perception have been completed, further processing proceeds along two separate pathways

  • Ventral stream: leads into brain regions that are involved in comprehending utterances

  • Dorsal stream: leads into brain regions that are involved in converting the auditory representations of words into matching articulatory codes

Where does initial evidence for separate streams come from?

Neuropsychology

Ventral Stream

• ”What” pathway

• From sound to meaning

• Allows the listener to understand the conceptual content of utterances

• Functional-anatomical components

  • Lexical interface: relay station that maps the sound structures of words onto the corresponding semantic structures

    • Depends on the mPTG and pITG in both hemispheres (leftward bias)

  • Combinatorial network: a system for integrating the semantic and grammatical aspects of phrases and sentences

    • Depends on the lateral ATL (predominately in the left hemisphere)

Dorsal Stream

• ”How” pathway

• From sound to action

• Allows the listener to link speech perception with speech production

• Supports not only the overt limitation and repetition of heard utterances, but also covert auditory-verbal STM and some aspects fo speech perception

• Functional-anatomical components

  • Sensorimotor interface: relay station that maps the sound structures of words onto the corresponding motor representations

    • Depends on area Spt in the left hemisphere

  • Articulatory network: underlies the production of utterances

    • Depends on a variety of regions in the left posterior frontal lobe

Classic Wernicke-Lichtheim-Geschwind “House” Model

• Two pathways project from the center for the “sound images” of words (A)

  • One pathway leads to the center for word meanings (B)

  • Another pathway leads to the center for speech production (M)

Early Cortical Stages of Speech Perception

• Processing extends from the dorsal STG down around the mid-lateral STG and into the upper lip of the mid-posterior STS

• Functional organization is both hierarchical and bilateral

• Language has left hemisphere dominance but both hemispheres are important

Hierarchical Organization

• Hypothetical neural network for processing the hierarchical organization of the “tonal scream” of the rhesus monkey

• The circuitry for processing human speech may be similar, but scaled up in complexity

Top of Hierarchy

The upper neuron serves as a “tonal scream detector” by integrating the inputs from neurons T1 and T2

Middle of Hierarchy

Neurons T1 and T2 integrate the auditory features of compromising the early (T1) and late (T2) phases of a tonal scream, specifically by firing only if all three of the appropriate lower-level FM neurons fire

Bottom of Hierarchy

The first three FM (“frequency modulated”) neurons detect the FM components of the early phase of a tonal scream, and the other three detect the FM components of the late phase

Putative Hierarchy in the Human Brain

1. The dorsal STG carries out fairly simple spectrotemporal analyses

2. The mid-posterior lateral STG represents subphonemic features and feature combinations

3. The mid-posterior STS represents individual phonemes and the sequential phonological structures of whole words

Mesgarani et al.’s (2014) ECoG Study: Hierarchical Organization

• Found that individual electrodes over the left mid-lateral STG responded selectively to certain categories of speech sounds

  • Some electrodes were sensitive to certain categories of speech sounds more than others

• Presented subjects with hundreds of naturally spoken sentences

• Plosives, fricatives, nasals

• Sensitive to a critical cue for distinguishing between vowels

• Suggests that this cortical region represents subphonemic features and feature combinations in a topographic yet widely distributed fashion

Chang et al.’s (2010) ECoG Study: Hierarchical Organization

• The spatial topography of neural responses in the mid-lateral STG was highly distributed and complex for each sound category but still distinct

• Created a continuum of 14 speech sounds by incrementally increasing F2

• Further evidence that this region represents subphonemic features and feature combinations

  • Not recognizing the consonants as distinct phonemes yet but distinguishing different places of articulation

• These sounds were perceived as belonging to three categories: /ba/, /da/, /ga/

Liebenthal et al.’s (2015) fMRI Study: Hierarchical Organization

• Left mid/post-STS activation in contrast between phonetic vs. nonphonetic discrimination

• Suggests that this region is where individual phonemes are explicitly recognized

• Two sounds of the same category are hard to distinguish

• Performance is good when they are in two separate categories (even if they are only two categories apart)

Okada & Hickok’s et al.’s (2006) fMRI Study: Hierarchical Organization

• Found mid/post-STS activation when high-neighboring-density words (i.e. those with many similar-sounding associations, like cat) were contrasted against low-neighborhood-density words (i.e., those with few similar-sounding associates, like spinach)

• Suggest that this region represents the pool of phonological associates that are automatically and unconsciously activated in a bottom-up manner during the process of auditory word recognition

• Phonological forms of words reside in these areas

How is hierarchical organization not only bottom-up but also top-down?

• Influenced by prior knowledge and expectations

• Phase of certain electrophysiological oscillations becomes entrained to (i.e., is brought into alignment with) the phase of certain speech rhythms

  • Theta oscillations (4 Hz range) correlate roughly with syllables

  • Gamma oscillations (30-70 Hz range) correlate roughly with subphonemic features and whole phonemes, analyze structures

• Increasing entrainment leads to better perception

Wada Procedures

• Performed a word-picture matching task with phonemic, semantic, and unrelated distractions

• Overall performance was quite good regardless of whether the left or right hemispheres was anesthetized

Distractors

Target word: “bear”

• Phonological: sounds like word (“pear”)

• Semantic: word category (“moose”)

• Unrelated: (“grape”)

“Word Deafness”

• Disorder in which speech perception is impaired, despite intact hearing and sometimes even intact recognition of nonspeech sounds

• Usually requires bilateral lesions to the middle and posterior portions of the STG and underlying white matter

  • While often sparing Heschl’s gyrus

• Very rare - almost always requires two strokes (one in each hemisphere)

“Asymmetric Sampling in Time” (AST) Hypothesis

• LH dominance for rapid changes of ~20-80 ms

  • Ideal for processing very brief aspects of speech (e.g., cues for place of articulation)

• RH dominance for slower changes of ~150-300 ms

  • Better for processing longer aspects of speech (e.g., syllabic structure)

• Support

  • At rest, LH dominance for gamma (40 Hz) oscillations, and RH dominance for theta (4 Hz) oscillations

  • During speech perception, gamma entrainment to short phonological features is stronger in the LH, whereas theta entrainment to longer ones is stronger in the RH

Arsenault & Buchsbaum’s (2015) fMRI Study: Bilateral Organization

• Place distinctions with really fast changing cues should rely on left hemisphere

• Manner distinctions with slow changing cues should rely on right hemisphere

• Voice distinctions with changes cues should rely on both hemispheres (more right than left)

What is the direction for the processing pathway for hierarchical organization?

Bidirectional

A Double Dissociation Between Comprehension and Repetition: Initial Evidence for Separate Processing streams

• Impaired comprehension but intact repetition = transcortical sensory aphasia

• Intact comprehension but impaired repetition = conduction aphasia or logopenic variant PPA

• Impaired comprehension but intact phoneme discrimination (monitoring)

• Intact comprehension but impaired phoneme discrimination (monitoring)

• Phonological representations → repetition → motor-articulatory system OR recognition → lexical-semantic system

Ventral “What” Stream: From Sound to Meaning

• Lexical interface (bilateral pMTG & pITS/pITG) maps phonological structures onto semantic structures

• Combinatorial network (left aMTG & aITS): contributes to integration of sentence meaning

What are the 7 locations of electrode pairs where language was tested for lexical interface in left pMTG/pITG?

1. Syllable discrimination

2. Word and sentence repetition

3. Sentence comprehension

4. Spontaneous speech

5. Oral reading of words

6. Oral reading of paragraphs

7. Oral object naming

What did the 29 sites where stimulated induced transcortical sensory aphasia (TSA) impact? (intact vs. impaired)

• Intact syllable discrimination

• Intact word and sentence repetition

• Impaired sentence comprehension

• Fluent but paraphasic production at 19 of 29 critical sites

What is suggested at the 10 sites where TSA was induced but with intact naming?

Impaired mapping of sound to meaning but normal mapping of meaning to sound

Dronkers et al. (2004) Study

• Serves as an intermediary between the phonological and semantic structures of words as maintained by the Dual Stream Model

• Most severe and pervasive deficits were left pMTG lesions

• Stroke patients with LH lesions

Bonilha et al. (2017) Study

• The posterior aspect of the left MTG possibly plays the role of bridging speech perception with subsequent comprehension

• Main task: noun-picture matching

  • Control task: object-picture matching

• Patients with widely distributed LH lesions

• Split brain

ATL Localizer

Passive listening to sentences → passive listening to noun lists

Semantic Task

Detect semantic anomalies (e.g. the infant was spilling some carpet on the milk)

Syntactic Task

Detect syntactic anomalies (e.g., the plumber with the glasses were installing the sink)

Combinatorial Network in Left ATL

• Only 20% of trials had anomalies and the correct sentence in the two tasks were identical

• Only a few voxels in the ATL had a task preference (semantic → syntactic)

Narain et al. (2003) Study: Combinatorial Network in Left ATL

• Subjects had two types of intelligible speech (A & B) and two matching types of unintelligible speech (C & D)

• Conduction analysis using the subtraction paradigm

Davis and Johnsrude’s (2003) fMRI Study: Combinatorial Network in Left ATL

• Normal speech

• Partly distorted speech

  • Three degrees of distortion: low, medium, high

  • Three types of distortion: vocoded, segmented, embedded in background noise

• Completely distorted speech

  • Signal-correlated noise (SCN)

Abrams et al.’s (2013) fMRI Study: Combinatorial Network in Left ATL

• Multivariate pattern analysis

• Two conditions: normal speech and rotated speech

• Two analyses: subtraction and MVPA

The Dorsal “How” Stream: From Sound to Action

• Sensorimotor interface (left Spt): maps phonological structures onto motor representations

• Articulatory network (left posterior frontal lobe): essential for speech production

Sensorimotor Interface in Left Area Spt

• Area Spt resides in the posterior portion of the planum temporale (PT) which straddles at least four different cytoarchitectonic fields

  • Exhibits both auditory and motor-related response properties

Sensorimotor Interface in Left Area Spt Trial

• 3 seconds of auditory stimulation (speech or tune)

• Followed by 15 seconds of covert rehearsal (speech or humming) of the heard stimulus

• Followed by 3 seconds of auditory stimulation (speech or tune)

• Followed by 15 seconds of rest

• Area Spt was engaged during auditory stimulation and covert rehearsal

  • Mid STG was only engaged during auditory stimulation

What is area Spt a sensorimotor network for?

Just vocal sounds/actions

Conduction Aphasia: Sensorimotor Interface in Left Area Spt

• Results from damage to the left supramarginal gyrus and inferiorly adjacent tissue, including area Spt

• Comprehension is mostly intact because the lesion spares the ventral stream

• Phonemic paraphasias are rampant (especially for long, complex, and low-frequency words) because the motor programming of words can no longer be guided, via area Spt, by the sound-based representations that specify the auditory “targets” of production

• Repetition is severely impaired because it depends critically on area Spt, the neural relay station that translates what one hears into how to say it

Logopenic Progressive Aphasia: Sensorimotor Interface in Left Area Spt

• Atrophy generally includes area Spt

• Similar set of symptoms but milder

Auditory-Verbal Short-Term Memory (STM) aka the “Phonological Loop”

• The perception of an utterance activates sound-based representations in the phonological network

• These representations are kept “alive” by means of corresponding subvocal motor processes in the articulatory network

• This reverberatory cycle is mediated by the sensorimotor interface

• Thus, auditory–verbal STM depends on the dorsal stream

Although the articulatory network might modulate speech perception in various ways, why is it probably not a necessary resource for comprehension?

• Large left frontal lesions severely impair production but not comprehension

• Deactivating the entire left hemisphere leads to the same outcome

• The failure to develop speech production does not preclude normal receptive speech development

• Infants as young as 1-month-old exhibit sophisticated speech perception ability, including categorical perception, well before they acquire the ability to speak

Chapter 6: Speech Production

Study Questions

How do the first three stages of the Lemma Model - specifically, lexical concept retrieval, lemma retrieval, and phonological code retrieval - appear to be organized in the left temporal lobe, and what is some relevant evidence?

• Lexical concept retrieval

  • Concrete nouns and action verbs reside in the ATLs bilaterally but with leftward bias

  • Resolution of conflicts between coactivated lexical concepts may depend on the left IFG

• Lemma retrieval

  • Concrete nouns reside in the varied sectors of the left MTG and ITG

  • Action verbs reside in the left ventral prefrontal cortex (including the IFG)

• Phonological code retrieval

  • Left posterior STG

What exactly are lemmas, and why does syllabification occur after phonological code retrieval?

• Lemma: an abstract word node that not only intervenes between semantics and phonology, but also points to morphosyntactic features like grammatical category

• Each retrieved segment in the phonological code spreads activation to all syllabic gestures in which it partakes

How does the syllabary in the Lemma Model relate to the Speech Sound Map in the DIVA Model?

Syllabary in the Lemma Model

• The theory does not take into account evidence that area Spt serves as an auditory-motor interface that relays signals back and forth between the sound-based representations of words in the left posterior STG and the corresponding motor-based representations of the same words in the left frontal lobe

• The theory predicts that cortical areas linked with phonetic encoding, such as left BA44, should be sensitive to syllable frequency, but several studies suggest that this may not be the case

Speech Sound Map in the DIVA Model

• A repository of acquired speech sound representations (phonemes, syllables, or syllable sequences) that serve as the starting point for articulation, and that reside in the left ventral premotor cortex, which for present purposes is assumed to include not only the rostral portion of the ventral precentral gyrus, but also neighboring regions in the posterior IFG and anterior superior insula

What is apraxia of speech, how is it accommodated by both the Lemma Model and the DIVA Model, and why are its lesion correlates controversial?

• Apraxia affects the speech motor planning operations that are called phonetic encoding in the Lemma Model

• Some fMRI studies with healthy subjects, and some lesion studies with AOS patients, have implicated the anterior insula instead of the pIFG/vPMC

  • But these findings have been challenged, and there is ongoing debate about this whole issue

What is the purpose of the Initiation Map in the DIVA Model, and where does it reside in the brain?

• A module that sends a “go” signal to prepared speech motor commands

• Resides in the SMA bilaterally, with modulatory influences from the basal ganglia

In the DIVA Model, what kinds of processing interactions occur between the Auditory Target Map, and the Auditory State Map, and the Auditory Error Map?

• During speech production, the Speech Sound Map not only sends feedforward instructions to the Articulator Map, but also sends an anticipatory message to the Auditory Target Map, indicating how the utterance should ideally sound

• The acoustic signals of the actual utterance are represented in the Auditory State Map, and those signals are matched against the target representation by the Auditory Error Map

• If the utterance was produced correctly, the error map does not generate any output, but if it was produced incorrectly, the error map alerts the Feedback Control Map, which then sends corrective motor commands to the Articulator Map

What are the peripheral mechanisms of speech production, and what disorders are caused by damage to them?

• Vocal tract representations in the primary
  motor cortex project to brainstem nuclei via the corticobulbar pathway

• These brainstem nuclei contain 12 sets of cranial nerves that innervate the head and neck

• The cells in the primary motor cortex are sometimes called upper motor neurons, and those constituting the cranial nerves are sometimes called lower
  motor neurons

• The cranial nerves not only transmit outgoing motor signals to the organs comprising the vocal apparatus, but also carry incoming sensory signals from the very same organs

To what extent do the neural substrates of speech production appear to overlap with the neural substrates of speech perception?

Summary and Key Points

According to the lemma model, word production consists of what two main subsystems?

• Lexical selection: identifying the most appropriate word in the mental lexicon

• Form encoding: preparing the word’s articulatory shape

Lexical Concept Retrieval and Selection (Lexical Selection Processing Stage)

• Converting the thought one wishes to express into the most appropriate lexical concept → semantic structure

• Neural correlates are still poorly understood

• Lexical concepts encoded by concrete nouns and action verbs may reside in the ATLs bilaterally but with leftward bias

• Resolution of conflicts between coactivated lexical concepts may depend on the left IFG

Lemma Retrieval and Selection (Lexical Selection Processing Stage)

• Mapping the selected lexical concept onto the corresponding lemma

• When the target word is a concrete noun, this process may be subserved by varied sectors of the left MTG and ITG

• When the target word is an action verb, the critical brain regions may be the left ventral prefrontal cortex (including the IFG) and the left inferior parietal lobule and mid/posterior MTG

• The resolution of conflicts between coactivated lemmas may depend on the left IFG

Phonological Code Retrieval (Form Encoding Subsystem Processing Stage)

• Incrementally spelling out the segmental phonological representation of the target word

• Left posterior STG

• Frequency effects occur

Syllabification (Form Encoding Subsystem Processing Stage)

• Determining the syllabic structure of the target word

• Incremental and context-sensitive process

• Takes place “on the fly”

• Sometimes transcends morpheme and word boundaries

• Most likely implemented in left BA44

Phonetic Encoding (Form Encoding Subsystem Processing Stage)

• Taking as input the syllabified target word and generating as output an articulatory score that specifies in a goal-oriented manner the speech motor tasks to be accomplished (e.g., lip closure)

• Associated with several neighboring left frontal regions (BA44, anterior part of the ventral precentral gyrus, and the anterior superior insula)

• If the preceding process of syllabification yields units that match precompiled programs in the syllabary, they are retrieved and concatenated

  • Otherwise, the phonetic form of the target word must be computed on the basis of the segmental and suprasegmental information accessed earlier

Lemma Model Challenges

• The ATLs may lack the connectivity necessary to subserve lexical concepts

• Based on data from brain-damaged patients who make semantic errors that are restricted to either oral output or written output, some researchers have questioned the plausibility of representational level for amodal lemmas

• The theory does not take into account evidence that area Spt serves as an auditory-motor interface that replays signals back and forth between the sound-based representations of words in the left posterior STG and the corresponding motor-based representations of the same words in the left frontal lobe

• The theory predicts that cortical areas linked with phonetic encoding, such as left BA44, should be sensitive to syllable frequency, but several studies suggest that this may not be the case

• The theory assumes that processing flows sequentially from stage to stage, but even though there is substantial support for this, there is also mounting evidence that multiple levels of architecture - and, by correspondence, multiple regions of the brain - are often activated either simultaneously or at several different time points during spoken word production

Where does the DIVA Model begin?

• Where the Lemma Model leaves off

• Phonetic encoding and articulation

According to the DIVA Model, the architecture that supports speech motor control consists of what two main systems?

• Feedforward control: activating motor commands for articulatory gestures and transmitting them to the vocal apparatus via subcortical nuclei

• Feedback control: using auditory and somatosensory input from self-produced speech to recognize errors and send corrective instructions to the articulatory component

Feedforward Control System

• Maps

  • Speech Sound

  • Articulator

  • Initiation

• During speech production, activation of a particular unit in the Speech Sound Map engages the corresponding vocal tract motor commands in the Articulator Map, and those commands are released by a “go” signal from the Initiation Map.

Speech Sound Map

A repository of acquired speech sound representations (phonemes, syllables, or syllable sequences) that serve as the starting point for articulation, and that reside in the left ventral premotor cortex, which for present purposes is assumed to include not only the rostral portion of the ventral precentral gyrus, but also neighboring regions in the posterior IFG and anterior superior insula

Articulator Map

A set of nodes that represent the major components of the vocal tract (larynx, lips, jaw, and tongue), that specify the time series of movements necessary to produce a particular utterance, and that have a rough somatotopic organization in the ventral primary motor cortex bilaterally

Initiation Map

A module that sends a “go” signal to prepared speech motor commands, and that resides in the SMA bilaterally, with modulatory influences from the basal ganglia

Auditory Feedback Circuit

• Maps

  • Auditory Target

  • Auditory State

  • Auditory Error

  • Feedback Control

• During speech production, the Speech Sound Map not only sends feedforward instructions to the Articulator Map, but also sends an anticipatory message to the Auditory Target Map, indicating how the utterance should ideally sound

• The acoustic signals of the actual utterance are represented in the Auditory State Map, and those signals are matched against the target representation by the Auditory Error Map

• If the utterance was produced correctly, the error map does not generate any output, but if it was produced incorrectly, the error map alerts the Feedback Control Map, which then sends corrective motor commands to the Articulator Map

Auditory Target Map

A module that subserves auditory target representations (i.e., acoustic expectations) during speech production, and that resides in the posterior STG bilaterally

Auditory State Map

A module that represents speech-related auditory input (including self- generated utterances), and that resides in Heschl’s gyrus as well as anteriorly and posteriorly adjacent superior temporal areas bilaterally

Auditory Error Map

A module that computes discrepancies between the anticipated and the actual sounds of self-generated utterances, and that resides in the posterior STG and planum temporale bilaterally

Feedback Control Map

A module that adjusts or updates articulatory commands in light of sensory feedback, and that resides in the right posterior IFG

Somatosensory Feedback Circuit

• Maps

  • Target

  • State

  • Error

Somatosensory Target Map

A module that subserves somatosensory target representations (i.e., tactile and proprioceptive expectations) during speech production, and occupies posterior parts of the ventral postcentral gyrus bilaterally

Somatosensory State Map

A module that processes tactile and proprioceptive feedback during speech production, and is represented in a somatotopic manner in anterior parts of the ventral postcentral gyrus bilaterally

Somatosensory Error Map

A module that computes discrepancies between the anticipated and the actual tactile and proprioceptive sensations associated with speech production, and that depends on several cortical regions bilaterally: posterior parts of the ventral postcentral gyrus; anterior parts of the adjacent supramarginal gyrus; and posterior parts of the insula

How does the speech sound map send an anticipatory message to the somatosensory target map?

• Indicates how the utterance should ideally feel in the vocal tract

• The tactile and proprioceptive signals of the actual utterance are represented in the Somatosensory State Map, and those signals are matched against the target representation by the Somatosensory Error Map

• If the utterance was produced correctly, the error map does not generate any output, but if it was produced incorrectly, the error map alerts the Feedback Control Map, which then sends corrective motor commands to the Articulator Map

How do vocal tract representations in the primary motor cortex get projected to the brainstem nuclei?

Via the corticobulbar pathway

How many cranial nerves does the brainstem nuclei contain?

12 sets that innervate the head and neck

Upper and lower motor neurons

• Upper: cells in the primary motor cortex

• Lower: cells constituting the cranial nerves

What type of signals does the cranial nerves transmit?

• Outgoing motor signals to the organs comprising the vocal apparatus

• Carry incoming sensory signals from the very same organs

Whole Lemma Model

Lexical selection (conceptual focusing perspective-taking) → lexical concept → (lexical selection) → lemma → form encoding (retrieving morphemic phonological codes) → phonological codes → (prosodification syllabification) → phonological word → (phonetic encoding) → articulatory score

What does the lemma model relay?

Sound structure and phonology

Lexical Concept

The thought you have converted to a meaning of the word of your language

What influences lexical concept retrieval and selection?

• Cross-lingusitic variation: messages must be “tuned” to the target language

• Perspective: subjective construal; considering the adressess’s state of mind

What are lexical concepts connected to/bind together/conjoin?

All the multifarious semantic features that constitute the conceptual content of the words

How are several semantically related lexical concepts activated?

Typically co-activated in parallel with one (the target) ultimately being selected

Lemma

Abstract word node that not only intervenes between semantics and phonology but also points to various morphosyntactic features like grammatical category (e.g., noun), nominal gender/class (e.g., feminine), verbal transivity (e.g, intransitivite)