Lecture 20 - Week 12 Thursday - Language Processing

Phoneme Perception Problems - What are the various problems with phoneme perception?

Lack of Invariance: There is no 1-to-1 mapping between the acoustic signal and the phoneme. 2. Speaker Variability: Different people say the same phoneme differently. 3. Coarticulation: Sounds overlap in time. 4. Segmentation: There are no physical gaps between words in speech, making it hard to tell where one ends and the next begins.

Coarticulation - What is coarticulation? Give an example?

Coarticulation is the phenomenon where the production of a phoneme is influenced by the phonemes that come before and after it because our mouths prepare for the next sound while finishing the current one. Ex. The "n" in "ten" is different from the "n" in "tenth" because your tongue is already moving toward your teeth for the "th" sound.

Phoneme Restoration Effect - Define the phoneme restoration effect and why it’s a solution?

It is a perceptual phenomenon where listeners "hear" a missing phoneme that has been replaced by a non-speech sound (like a cough). It is a solution because it uses Top-Down information (context and semantics) to "fill in the blanks" of a messy or incomplete acoustic signal.

Phoneme Restoration Example - What is an example of the phoneme restoration effect?

In a study, subjects heard the sentence "The state governors met with their respective legi*slatures," but the "s" in legislatures was replaced by a cough. Subjects reported hearing the full word and could not accurately identify where the cough occurred because the brain restored the missing phoneme.

McGurk Effect - What is the McGurk Effect and why is it a solution?

The McGurk Effect demonstrates that speech perception is multimodal (uses both hearing and vision). It is a solution because when the auditory signal is ambiguous, the brain uses visual "lip-reading" cues to clarify the sound. Ex. Hearing the audio "ba" while seeing a video of someone saying "va" results in the brain perceiving "va."

Mondegreens - What are mondegreens?

Even adults make errors when trying to segment phonemes into words. These errors, usually seen in misheard song lyrics, are called Mondegreens. They result when your perception makes a mistake in grouping phonemes into words.

Homographs - What are homographs?

Homographs are two words with identical spelling but different pronunciations and meanings. Ex. "Who can lead [leed] the effort to remove dangerous lead [led] water pipes?"

Homonyms - What are homonyms?

A homonym is when two words are spelled and pronounced the same way but have different meanings. Ex. "If you want to read this sentence about a basic human right the right way, you should read from left to right."

Top-Down Solution - Why is using top-down information a solution to phoneme perception?

Because the acoustic signal is so variable and messy, we cannot rely on bottom-up data alone. We use our knowledge of words, grammar, and social context to "expect" certain sounds, which helps the brain categorize ambiguous inputs correctly.

Categorical Perception - What is categorical perception and why is it a solution?

Categorical Perception is the phenomenon where we perceive continuous acoustic changes as belonging to discrete categories. Ex. Even as the "Voice Onset Time" (VOT) changes gradually between a "ba" and "pa," your brain doesn't hear a "hybrid" sound; it hears "ba" until a specific threshold, then suddenly hears "pa." This solves the variability problem by forcing messy sounds into clean boxes.

Word Perception Difficulty - Why is word perception difficult?

It is difficult because of Segmentation (speech is a continuous stream of sound) and Ambiguity (homophones and words that sound similar, like "iced cream" vs. "ice cream").

Lexical Matching Process - What is the solution to the problem of word perception?

The solution is a matching process between the INPUT and your Lexicon (mental dictionary). This involves complex top-down models where the brain tries to find the best fit between the speech sound and the representations of meaning and spelling you already know.

Lexical Access Testing - What is lexical access and how do scientists test for it?

Lexical Access is the process of "looking up" a word in your mental dictionary and activating its meaning. Scientists test for it using Cross-modal Priming tasks to see how quickly a word's meaning is activated in the mind.

Cross-Modal Priming - What is cross-modal priming? Give an example?

It is a task where a subject listens to a sentence (auditory) and must simultaneously perform a task on a screen (visual). Ex. While hearing the word "bug" in a sentence about a spy, the words "ant" (insect) and "spy" (electronic) are flashed on a screen. If the subject reacts faster to both, it shows they have accessed multiple meanings of the word.

Priming & Matching - Why is cross-modal priming a good test for lexical access?

It reveals that the matching process is exhaustive at first; when we hear an ambiguous word, the brain briefly activates all known meanings before using context to discard the irrelevant ones. This shows how the brain handles the "input-to-lexicon" match.

The Parser - What is the parser and what job does it do?

The Parser is the mental mechanism responsible for assigning a phrase structure to a sentence as it is being heard. Its job is to take a sequence of words and determine the grammatical relationship between them in real-time.

Parser Cues - What are the 3 cues that the parser uses to decide which phrase structures to pay attention to?

Key Words 2. Word Order 3. Principle of Minimal Attachment.

Key Words - What are key words? Give an example?

Key words are specific words that provide a strong hint about the upcoming phrase structure. Ex. The word "which" or "that" usually signals the start of a relative clause (e.g., "The car that…").

Word Order - What is word order as a cue? Give an example?

The parser relies on the standard Subject-Verb-Object (SVO) order of English to assign roles. Ex. In "The cat chased the mouse," the parser immediately assumes the first noun is the actor and the second is the recipient.

Minimal Attachment - What is the principle of minimal attachment? Give an example?

The parser tries to build the simplest possible phrase structure tree, adding as few "nodes" or levels as possible. Ex. In "The spy saw the cop with the binoculars," the parser initially attaches "with the binoculars" to the verb (the spy used them) because it's a simpler structure than attaching it to the cop.

Dealing with Ambiguity - What is the process that deals with ambiguity?

When the parser encounters a "Garden Path" sentence (where the initial simple structure fails), it must engage in Re-parsing. It stops, goes back, and builds a more complex structure that fits the meaning.

Information Storage - How does the parser store information?

The parser stores information in terms of Propositions (the underlying meaning/relationships), NOT in the literal surface format or verbatim word-for-word string.

Propositions - What are propositions?

A proposition is the smallest unit of knowledge that can stand as a separate assertion. It represents a Relation (the action or state) and its Arguments (the people or things involved).

Properties of Relations - What are the two properties of relations?

Reflexivity: Whether a relation can apply to itself. 2. Symmetry: Whether a relation works both ways (e.g., if A is a sibling of B, B is a sibling of A).

Native vs. Nonnative Speech Perception - Why do native English speakers make more errors perceiving speech from nonnative speakers?

Native speakers make more errors because nonnative speakers often produce phonemes differently than the native speaker expects, creating a mismatch in perception.

Orthographic Processing - What is orthographic processing and how does it relate to processing letters?

Orthographic processing is the visual aspect of letter processing where the mind recognizes the linguistic meaning of a visual symbol.

Mondegreens - What are mondegreens?

Mondegreens are misheard song lyrics that occur when the perception system makes a mistake in grouping phonemes into words.

Homographs vs. Homonyms - What is the difference between homographs and homonyms in written word processing?

Homographs are words with the same spelling but different pronunciations and meanings, while homonyms share both the same spelling and pronunciation but have different meanings.

Infant Word Learning - How do babies use sound patterns to learn where words end?

Babies learn that sounds in the middle of words are followed by a limited set of sounds, whereas sounds at the end of words have a wider range of possible followers; they use this statistical range to identify word boundaries.

Concrete vs. Abstract Word Processing - How does the processing of concrete words differ from abstract words?

Concrete words are easier to remember, but abstract words are easier to recognize as words because they become active in the lexicon more easily.

Surface Dyslexia - Damage to which processing route corresponds with surface dyslexia?

Surface dyslexia is associated with damage to the spelling-lookup route.

Phonological Dyslexia - Damage to which processing route corresponds with phonological dyslexia?

Phonological dyslexia is associated with damage to the translation rules (phonological sounds) route.

Garden Path Sentences - What is a garden path sentence?

A garden path sentence is one where the cognitive system initially builds one phrase structure that later proves to be incorrect, such as "While Anna dressed the baby played in the crib".

Levels of Text Representation - What are the three levels of representation in texts?

The three levels are the surface code, the textbase (basic units of meaning called propositions), and the situation model (the mental model incorporating long-term memory).

Surface Code Definition - What is the definition of the surface code?

The surface code is the most basic level of text representation, focusing on the exact wording and syntax used in the text.