speech production

Overview of Speech Production

• Speech production is a process initiated by the speaker formulating a message in their mind and subsequently transmitting it to the listener via speech.
• Key aspects of speech:
  • Conveys meaning
  • Expresses emotion
  • Provides information about the speaker (e.g., identity, health, and status)

Anatomy of Speech

• Speech production involves several steps:
  • Selection of spoken words
  • Formulation of phonetics (sounds)
  • Articulation by the motor system in the vocal apparatus
• Source-Filter Theory:
  • The lungs provide the energy (airflow).
  • The vocal folds in the larynx act as the source of sound through vibration.
  • The oral and nasal cavities function as a filter, resonator, and articulator to shape the sound into recognizable speech.
• Important distinction:
  • Speech production differs from language production, as language can be produced manually through signs as well as spoken.

Speech Motor Control

• Illustrative studies (e.g., Real-time MRI of human beatboxing) demonstrate the complex coordination of the tongue, lips, and soft palate.
• Neural control involves specific brain regions:
  • Broca’s area: Traditionally associated with speech planning.
  • Primary Motor Cortex: Responsible for executing the physical movements of articulation.

Challenges in Speech Production

Tongue-Twisters

• Tongue-twisters present unique challenges because:
  • Sounds with similar vocal tract movements (e.g., 'Sss' and 'Shh') complicate processing due to overlapping brain representations.
  • Require rapid sequences of overlapping neural patterns that can overwhelm cognitive processing.

Speaking Complexity

• Speaking is inherently complex; the average person produces roughly $2$ to $3$ words per second.
• Errors can arise from cognitive load:
  • Dell (1986) highlights that understanding the breakdown of speech can reveal the complexities within the system.

Speech Errors

Freudian Slips

• Sigmund Freud argued that many speech errors stem from unconscious thoughts, termed as "Freudian slips."
• Limitations of this view include:
  • Some slips do not align with repressed thoughts.
  • More recent research (Garrett, 1988) suggests errors may reflect the structural workings of the language system more than intrusion of thoughts.

Types of Speech Errors

• Dell (1986) identified various types of speech errors:
  • Semantic substitution: Replacement of a correct word with a similar meaning (e.g., "Where's my tennis bat?" instead of "Where's my tennis racquet?").
  • Word-exchange error: Transposing two words (e.g., "I must let the house out of the cat").
  • Morpheme-exchange errors: Incorrect attachment of inflections/suffixes (e.g., "I randomed some samply").

Spoonerisms

• A form of phoneme exchange error where the initial letters of two words are switched, named after Rev. William Archibald Spooner ($1844-1930$).
• Historical anecdotes illustrate his tendency for these errors (e.g., "You have hissed all my mystery lectures").

Classification of Speech Errors

• Errors can be classified by:
  • Type (substitutions, exchanges, additions)
  • Specific linguistic units involved (e.g., words, morphemes, phonemes).

Examples of Speech Errors

• Illustrates different types of errors and their classifications based on mechanisms:
  • Feature Perseveration: "Turn the knop" → Target: "knob" (the voicing feature of the /n/ persists).
  • Phoneme Anticipation: "The mirst of May" → Target: "first".
  • Word Exchange: "Guess whose mind came to name?"
  • Morpheme Exchange: "I randomed some samply" → Correct: "I sampled some randomly".

Causes of Speech Errors

• Misapplication of rules may cause speech errors such as:
  • One noun exchanged for another noun, leading to phrases like "writing a mother to my letter."
• Demonstrates categorical constraints: Errors usually occur within the same linguistic category (nouns replace nouns, verbs replace verbs).
• Speech errors highlight that utterances are pre-planned with multiple units competing for production.

Speech Disfluencies

• Defined as irregularities or breaks that occur within fluent speech.
• Types include:
  • False starts: Incomplete phrases or sentences.
  • Repaired utterances: Correction of verbal slips or mispronunciations.
  • Fillers: Non-lexical utterances ('uh', 'erm', 'well') which may signal that the speaker is planning their next move.

Stuttering (Stammering)

• Characterized by:
  • Involuntary repetitions and prolongations of sounds, syllables, or phrases.
  • Silent pauses or “blocks” where the person cannot produce sounds.
• Emotional and Functional Impact:
  • Severe social anxiety or fear of specific sounds.
  • May lead to self-isolation and stress.
  • Treatment strategies, such as speech therapy and relaxation techniques, exist, but exact neurobiological causes remain unclear.

Theoretical Models of Speech Production

Model Overview

• Levelt’s model (1989) identifies three primary processes:
  1. Conceptualization: Message level representation (determining what to say).
  2. Formulation: Translating conceptual information into linguistic forms (lexicalization and syntactic planning).
  3. Execution: Detailed phonetic planning and motor articulation.

WEAVER Model

• Proposed by Levelt, Roelofs, and Meyer (1999).
• Sequential stages of activation in speech production:
  • Semantic level: Nodes representing lexical concepts.
  • Lemma level: Abstract word forms in the mental lexicon containing grammatical information (e.g., gender, tense).
  • Lexeme level: Word forms regarding morphemes and phonemic segments.
• Self-Monitoring: The model includes a “loop” where speakers monitor their own pre-speech (inner voice) to catch errors before they are voiced.
• Evidence for model through the “tip-of-the-tongue” (TOT) phenomenon identifying failure in lexeme access after lemma activation.

Spreading-Activation Theory

• Highlights that speech planning happens simultaneously across four processing levels:
  • Semantic: The intended meaning.
  • Syntactic: Grammatical structure of planned utterance.
  • Morphological: Morphemes in the sentence.
  • Phonological: Phonemes within the planned utterance.
• Activation flows down from the semantic level, but can also spread “backwards,” explaining why we make mistakes like saying a word that sounds similar to the target.

Communication Beyond Words

• Speech involves more than simple verbal content:
  • Discourse Markers: Phrases that clarify speaker intentions (e.g., "well," "anyway," "actually").
  • Prosodic Cues: The "music" of speech, which includes:
  • Pitch: High or low frequency of voice.
  • Loudness: Volume used for emphasis.
  • Duration: The length of time a sound is held.

Listening in Communication

Importance of Common Ground

• Shared knowledge, beliefs, and assumptions enhance effective communication.
• Listeners' interpretation is often influenced by their own knowledge instead of the speaker's perspective, which may lead to egocentric misunderstandings.

Research Findings on Common Ground

• Horton & Keysar (1996) demonstrated that interactive communication reduces error rates in shared tasks ($39\%$ errors without interaction vs. $5\%$ with interaction).

Models of Speech Planning (Horton & Keysar, 1996)

• Two approaches to planning:
  1. Initial Design Model: Plans are constructed specifically based on shared common ground with the listener from the start.
  2. Monitoring and Adjustment Model: Initial plans focus on the speaker's own perspective (egocentric), followed by corrections based on feedback and knowledge shared with the listener.
• Research suggests speakers most commonly use the Monitoring and Adjustment Model because it is less cognitively taxing initially.

Establishing Common Ground (Bard et al., 2007)

• Participants demonstrate reliance on:
  • Shared Responsibility Strategy: Expectation that the listener will signal if they don't understand.
  • Cognitive Overload: The difficulty of tracking personal knowledge while simultaneously tracking the listener's knowledge.