Speech Perception and Acoustic Processing

Speech Perception Notes

Speech Perception Process

• Conceptual, Syntactic, Orthographic Codes: Essential for understanding language.
• Lexical Selection & Retrieval: Involves choosing the correct word from memory during speech.
• Prelexical Code: Transition from sound signal to language units.
• Phonetic Decoding: Breaks down acoustic signals into phonemes.
• Acoustic Input: Raw sound that initiates speech perception.

Continuous Speech Signal

• Example: "Mushrooms are an edible fungus."
• Spectrogram: A visual representation that shows sound amplitude across different frequencies over time.
• The spectrogram illustrates how continuous speech doesn't have clear boundaries or gaps between words.

Challenges in Speech Perception

• Segmentation: Speech is continuous; linguistic representation is discrete.
• Variability: Phonemes can vary by speaker, context, and accents, making them challenging to identify accurately.
  • Example: The only silence in “ago” is the /g/ sound.

Acoustic Properties of Speech

• Phoneme: The smallest sound unit in a language; meaningful differences.
  • Examples:
    • /b/ in "bad" vs. /p/ in "pad"
    • /d/ in "bad" vs. /t/ in "bat"
• Phones: Individual sounds; a phoneme consists of all its phones, treated the same.
• Allophones: Phones that are variations of the same phoneme in different contexts.
  • Example: In English, [k] and [q] are allophones of /k/.

Structure of the Vocal Tract

• Anatomical Features:
  • Sphenoidal Sinus, Nasal Meatuses, Pharyngeal Tonsil, Uvula, etc.
  • Components of the Ear: Malleus, Incus, Stapes (ossicles), Tympanum (eardrum), Cochlea (inner ear).

Acoustic Properties of Sounds

• Amplitude: Height of sound waves - relates to loudness.
• Frequency: Cycles per second - relates to pitch. Low frequency = high pitch; high frequency = low pitch.
• Harmonics: Higher frequencies that occur at specific intervals of the fundamental frequency, determining timbre.
  • Example: Middle C (262 Hz) with its first harmonic at 524 Hz, etc.

Source-Filter Theory of Speech Production

• Sources: Vocal folds generate sound.
• Filters: The vocal tract shapes the sounds generated by the source, modifying them to produce different phonemes.
  • Resonance: Vocal tract shape affects which frequencies are amplified. Formants are enhanced frequency bands (usually 3-4 formants in speech).

Cues in Speech Perception

• Formant Transitions: Changes in frequencies that occur before or after stops.
• Categorical Perception: Refers to our ability to discriminate between categories of speech sounds, rather than variations within categories (e.g., voice onset time).
• Top-Down Processing: Contextual cues that help listeners process speech even when parts are missing or mispronounced.
  • Example: Phoneme Restoration Effect shows that we can perceive words even when phonemes are obscured.

Neural Mechanisms and Models

• Dual-Stream Model (Hickok and Poeppel): Different auditory regions process information for speech comprehension and production.
• TRACE Model: An interactive activation model that explains how features, phonemes, and words are processed in a layered manner, allowing top-down and bottom-up influences.

Co-Articulation and Context Effects

• Co-articulation refers to how phonetic segments overlap, impacting speech perception positively and negatively.
• Plasticity and Age of Acquisition: Neuroplastic changes occur based on language exposure and adaptability.

Final Notes

• Speech perception is a complex interplay of acoustic signals, cognitive processes, and contextual factors, supported by a biological basis for integration between perception and production.