Comprehensive Study Notes on Language, Computation, and Speech Perception
Language, Computation, and Mind
- Language does not equal communication (Language=Communication).
- Language is defined formally as a system of rules and representations.
- Communication is merely one specific problem or application that a system of language can be used to solve.
- The term "language" is used in non-communicative domains:
- Computer languages are systems of rules and representations used for commanding and controlling machines.
- Computer languages provide the basis for computation by specifying how a problem is represented (representation) and what transformations are applied (rules).
- Computation is the process of solving problems in an information-processing framework, where inputs and outputs consist entirely of information.
- History and evolution of computing:
- In the year 1926, a "computer" was not an electronic machine, but a human being employed to perform mathematical calculations and solve equations.
- Alan Turing, recognized as a founding father of modern computing, analyzed human mathematical problem-solving to design mechanical computing architectures.
- Turing mapped human cognitive processes directly to machine components:
- Algorithm: The step-by-step procedure for solving a problem.
- Long-Term Memory: Designated by Turing as machine memory.
- Short-Term Memory: Designated by Turing as the internal state of the machine.
- This historical progression exhibits a circular conceptual loop: human problem-solving inspired computing machines, which subsequently served as the computational metaphor for human cognitive architecture.
- Distinguishing language from animal communication:
- Many non-human species communicate without possessing language.
- A domestic dog barking at a person walking past a yard communicates a clear functional message (e.g., territory defense) without utilizing a system of rules and representations.
- The general communication problem:
- A sender originates an internal mental thought.
- The sender must convert this thought into a physical signal that can be transmitted external to their body (e.g., acoustic sound waves).
- A receiver detects the external signal and attempts to reconstruct the sender's original internal thought.
Speech Perception: The Motor Theory and Sound Stream Processing
- Information processing stream in speech production:
- Thought / Semantics (Meaning): The underlying abstract mental representation.
- Syntax: The computational structure of a sentence that pieces words together.
- Phonology: The sound system of the language that structures words.
- Motor Commands: Signals sent from the brain to control the muscles of the mouth and vocal tract.
- Information processing stream in speech perception:
- Rather than maintaining an entirely separate cognitive module to decode incoming acoustic signals back into thoughts, the brain reuses its production architecture in reverse.
- Processing incoming acoustic signals backwards through the production pipeline (Motor Commands → Phonology → Syntax → Semantics) represents an economical and computationally efficient design.
- Cognitive definitions and components:
- Semantics / Thought: Everyday meaning.
- Syntax: Structural rules organizing words into grammatical sentence frameworks.
- Phonology: Sound representations forming words.
- Motor Commands: Specific muscular instructions governing the vocal apparatus.
- Qualia and conscious experience in language:
- Qualia refers to the subjective, conscious internal experience of mental processes.
- Conscious processes possess explicit qualia, whereas unconscious processes lack qualia.
- In language comprehension, conscious awareness (qualia) is restricted almost entirely to phonology (the overt sounds of speech).
- Syntactic parsing (identifying nouns, verbs, prepositions, determiners) and semantic decoding occur unconsciously without direct qualia of the underlying computational steps.
- Target of speech perception:
- Listeners do not directly experience raw physical sound waves or direct mental thoughts of others.
- Human speech perception explicitly tracks and represents the speaker's vocal tract and mouth movements.
Acoustic Evidence and the McGurk Effect
- Absence of physical boundaries in speech stream:
- In an unfamiliar foreign language, speech is perceived as an unbroken, continuous stream of sound.
- In one's native language, speech is perceived as discrete words separated by distinct physical pauses.
- Physical acoustic analysis (e.g., spectrograms) demonstrates that native continuous speech contains no physical gaps or breaks between words.
- Perceived word boundaries are constructed internally by the brain tracking speaker motor commands rather than reflecting actual physical silent gaps.
- Acoustic variability and contextual invariance:
- Contextual co-articulation alters raw acoustic waveforms.
- Comparing the short vowel sound in the words "can't" and "cat":
- If the isolated physical acoustic waveform of the vowel is extracted and played in isolation, it does not sound like an identical "a", nor does it sound like human speech (sounding instead like a bird chirping).
- In sentence context, both vowels are perceived as identical because the brain recognizes that the motor command sent to the vocal tract muscles was identical, despite acoustic differences caused by adjacent consonants.
- The McGurk Effect:
- A perceptual demonstration proving that speech perception relies on motor tracking rather than isolated acoustic input.
- Experimental setup:
- Visual Video Input: A video recording of a speaker's mouth visually forming the bilabial consonant "ba" (/b/).
- Auditory Audio Input: An audio soundtrack overlaid with the speaker producing the velar consonant "ga" (/g/).
- Perceived Sound: A listener simultaneously viewing the video and listening to the audio perceives the alveolar consonant "da" (/d/).
- Manipulating perception: Closing one's eyes reveals the true auditory input ("ga"); opening one's eyes forces the integrated illusion ("da").
- Articulatory mechanism of the McGurk Effect:
- Bilabial constriction (/b/): Airflow restricted at the absolute front of the vocal tract (visible lips).
- Velar constriction (/g/): Airflow restricted at the back of the vocal tract (velum/tongue back).
- Cognitive integration: The brain integrates conflicting front visual constriction and back acoustic constriction, calculating that the articulators must be moving at an intermediate location (alveolar ridge), yielding the perception of "da".
Proximal vs. Distal Stimuli in Auditory Processing
- Perceptual psychology definitions:
- Distal Stimulus: The real-world physical object or event in the environment (e.g., a solid physical cube; physical movements of a talker's vocal tract).
- Proximal Stimulus: The energy or pattern hitting sensory receptors (e.g., photons striking the retina; sound pressure waves striking the tympanic membrane).
- Vocal tract movements as the distal stimulus:
- In speech perception, vocal tract mouth movements serve as the distal stimulus.
- Mouth movements are the final physical event prior to sound wave generation and connect directly into the cognitive production pipeline.
- Dual auditory processing systems:
- General Auditory System: Processes environmental non-linguistic sounds (e.g., water boiling, automobile exhaust). These events lack underlying mental thoughts or generative rule systems.
- Specialized Speech Processing System: A dedicated cognitive mechanism designed exclusively to interpret acoustic signals resulting from vocal tract motor commands in order to reconstruct a speaker's mental thoughts.
Phonological Features, Articulation, and Evolutionary Psychology of Accents
- Articulatory dimensions of phonology:
- Voicing: Vibration of the vocal cords during sound production.
- Unvoiced sound: Vocal cords do not vibrate (e.g., /s/).
- Voiced sound: Vocal cords vibrate (e.g., /z/).
- Constriction: Obstruction of airflow within the vocal tract.
- Stop Consonants: Airflow is completely blocked momentarily (e.g., /b/, /p/, /d/, /t/, /g/, /k/).
- Non-stop Consonants: Airflow is restricted continuously without complete closure.
- Position (Place of Articulation): The specific physical site of airflow constriction:
- Lips (Bilabial): e.g., /b/, /p/.
- Tongue to roof/front of mouth or teeth (Alveolar/Dental): e.g., /d/, /t/.
- Nasal: Airflow diverted through the nasal cavity (e.g., /m/, /n/).
- Distinctive features representation:
- Brain representations of phonemes consist of combinations of primitive articulatory features:
- Phoneme /b/ = Voiced+Stop+Lips
- Phoneme /s/ = Unvoiced+Continuant+Dental/Alveolar
- Accents and evolutionary psychology:
- Accent vs. Dialect: An accent reflects systematic shifts in phonological feature implementation across languages or regions.
- Critical Period: Phonological habits lock in by puberty. Acquiring a new language or dialect post-puberty results in a persistent foreign accent.
- Evolutionary adaptation hypothesis:
- In ancestral small-scale societies (e.g., Yanomamö communities studied by Napoleon Chagnon), individuals were genetically related to nearly everyone in their village.
- In modern contexts (e.g., a UC Santa Barbara class of 800 students), only 2 to 4 students might share a close relative.
- In ancestral environments, an accent served as a vocal marker indicating geographic origin and kinship status.
- Outbreeding preference: An accent in a potential mate may signal genetic distance (non-kin status), reducing inbreeding risk and rendering opposite-sex foreign accents appealing or attractive.
- Outgroup competition: An accent in a same-sex individual or non-mate may signal outgroup competition, triggering derogation (e.g., perceiving the individual as unintelligent, such as reactions to Arnold Schwarzenegger's accent).
Motor Commands, Universal Grammar, and Phonemic Pruning
- Motor commands and muscular execution:
- Brain motor commands represent algorithms specifying contraction of specific tongue and facial muscles to position vocal articulators.
- Universal Grammar and selectional pruning:
- Language acquisition does not proceed by building rules from zero (blank slate).
- Under Universal Grammar, infants possess the innate capacity for all possible human language structures and phonemic distinctions.
- Learning a local language is a process of entrainment and pruning: unneeded phonological distinctions present in non-local languages are systematically discarded.
- Cross-linguistic phonemic loss examples:
- Native Japanese speakers often experience difficulty distinguishing English /l/ and /r/ phonemes due to early phonological pruning of that specific articulatory contrast.
- South Asian languages frequently utilize aspirated stop consonants (/bh/, /dh/, /gh/, /kh/); native English speakers lack these distinctions and categorize /bh/ simply as /b/.
- English utilizes distinct interdental and labiodental fricatives (/θ/ as in "th", /p/).
- Adult language learning limits:
- Once innate phonological machinery is pruned during childhood development, re-acquiring discarded phonemic distinctions in adulthood becomes extremely difficult.
Student Discussion and Dialogue
- Course and Reading Logistics:
- Jackendoff Reading Assignment: Covers 2 chapters alongside all associated preceding lectures.
- Exam Coverage: Exam content covers all lecture material and assigned chapters.
- Class Discussion on Teaching Style:
- Students commented informally that the instructor frequently improvises lecture material ("winging it").
- Interpersonal Side Dialogue:
- Sandwich Pricing Exchange:
- Student Inquiry: Asks if sandwiches cost 12.
- Response: Sandwiches cost between 10 and 15, except for egg salad which costs approximately 6 something.
- Outcome: Student decides to hold off on purchasing food until returning home.
- Bodybuilding and Off-Season Off-Topic Conversation:
- Discussion regarding running a laboratory blood test ("bar and test") and off-season muscle mass building.
- Mention of performance compounds, including DHT (Dihydrotestosterone) derivative compounds, 19-nor (19-nortestosterone) steroids, and upcoming competitive season prep.
- Informal Compliment: Exchange regarding a cute hairstyle ("Your hair looks super cute" / "Thank you").