8 (143) Syntactic Bootstrapping and Morphological Development Study Guide
Foundational Concepts in Word Learning (Review)
Radical Indeterminacy: Refers to the difficulty in determining the exact meaning of a novel word (e.g., "Gavagai!") based solely on an observation of the world.
Word Learning Constraints and Abilities: - The Whole Object Constraint: The assumption that a novel word refers to an entire object rather than its parts or attributes. - The Taxonomic Constraint: The assumption that a word refers to a category of like things (e.g., if shown a dog and told "dax," the child assumes other dogs are "dax," not things related to dogs like bones). - The Mutual Exclusivity Constraint: The assumption that each object has only one name; if a child knows the word for one object, they will assume a novel word refers to a different, unnamed object (Example: "Give me the dax!"). - Fast Mapping: The ability to learn the meaning of a new word after only a single exposure.
Domain Specificity: Research by Markson & Bloom (1997) investigated if these constraints are specific to language by testing immediate, one-week, and one-month retention of words vs. facts in 3-year-olds, 4-year-olds, and adults. Results indicated high performance for word learning across these intervals, often significantly above chance.
The Shape Bias: - Hypothesis 1: Children learn associations between linguistic cues and object properties which blindly guide word extension via attention to features like shape. - Hypothesis 2: Children use shape as a clue to category membership similar to adults, relying on it primarily when they lack other information (e.g., "This is a fendle," standard object compared against a shape match vs. a substance match).
Polysemy and Learning: Polysemy (one word having multiple meanings) may facilitate learning by: - Scaffolding guesses about word reference. - Shaping how children generalize words. - Allowing children to spontaneously infer new meanings (e.g., "high" used for physical height vs. "high voice"; or "glass" for the material vs. "a glass" for the object).
The Limitations of Word-World Pairings
Relational and Abstract Words: While concrete words (e.g., "bunny") can be learned through association and social cues, words like "go," "sit," "eat," "think," "know," and "hope" are difficult to acquire by mere observation.
The Problem of Multi-Interpretation: Events can be interpreted in multiple ways. A single scene could be described as "daxing" where it is unclear if the verb refers to the action, the result, or the perspective.
Gleitman (1990) Observations: - Mapping Difficulty: Some verbs apply to the same physical events (e.g., "Buy" vs. "Sell," or "Chase" vs. "Flee"). Repeated pairings with world situations do not resolve which perspective the verb takes. - Mental Verbs: Mental states like "think," "guess," "wonder," or "know" do not have distinct physical markers (no "brow furrowing" or "Rodin statue" always present) to distinguish them from situations where no thinking is occurring.
Syntactic Bootstrapping and Sentence Context
The Process of Acquisition: 1. Initial Stage: Children rely on observed scenes to learn concrete object labels (nouns). 2. Concrete Relational Stage: After learning basic nouns, children use sentence context to learn concrete relational words (e.g., "man… GORP… apple" allows the child to guess GORP means GIVE, EAT, or HOLD). 3. Syntactic Bootstrapping Stage: Children use the grammatical structure of sentences to guess meanings, facilitating the learning of abstract words.
Semantic Projections from Syntactic Structure: - Intransitive Frames: Typically represent Agent's self-caused acts (Example: "John sleeps"). Children infer "sleeping" involves one participant and is self-caused. - Transitive Frames: Typically represent an agent acting on an affected thing (Example: "John hits the ball"). - Ditransitive Frames: Typically represent Transfer (Example: "John puts the ball on the table"). - Sentence Complement Frames: Typically represent Perceptual or Mental states (Example: "John sees/thinks that the ball rolled away"). This structure helps explain how abstract verbs like "think" are learned.
Evidence for Syntactic Bootstrapping
The Human Simulation Paradigm (Gillette, Gleitman & Gleitman): - Task: Adults identify a "mystery word" replaced by a beep in a silent video clip. - Results: Nouns are easily identified from scenes alone (~45% accuracy). Verbs are very difficult to identify from scenes alone (~15% accuracy). - Variables: Information conditions include Scenes (S), Nouns (N), and Syntactic Frames (F). Accuracy for verbs increases significantly with "S+F," "N+F," and "Full Info."
International Adoption Studies (Snedeker et al., 2007): - Subject Group: Chinese preschoolers adopted into English-speaking homes. This serves as a "natural experiment" separating cognitive maturation from language exposure. - Findings: Despite being older and having higher cognitive abilities than infants, adopted children follow the same developmental shifts. Their early vocabularies are dominated by nouns, and the proportion of verbs increases only as vocabulary size grows. This supports the Contingent Acquisition Proposal: it is a logical necessity to learn concrete words first to decode syntax. - Statistical Data: The proportion of nouns versus vocabulary size showed for Chinese Adoptees and for EFL Toddlers.
Experimental Studies (Naigles, 1990): - Task: 2-year-olds were shown a training video where two actions occurred (e.g., one character causing another to squat while both wheeled their arms). - Conditions: Children heard either a Transitive frame ("The duck is gorping the bunny") or an Intransitive frame ("The duck and bunny are gorping"). - Test: When asked to "Find gorping!", children in the Transitive condition looked longer at the causal action, while those in the Intransitive condition looked at the non-causal action. This proves toddlers use syntax to constrain verb meanings.
Morphology: The Structure of Words
Lexicon vs. Grammar: The lexicon contains memorized units, while grammar contains rules for combining them.
Morpheme: The minimal unit of linguistic form and meaning. - Free Morpheme: Can stand alone (e.g., "boy"). - Bound Morpheme: Must be attached to another morpheme (e.g., "-ish," "-s").
Word: A fixed sequence of morphemes (e.g., "boyish").
Devices for Creating Words: - Inflectional Rules: Create grammatical variants (tense, number) without changing the grammatical category (e.g., "talk" to "talked"). - Derivational Rules: Create new words from existing ones, often changing the part of speech (e.g., "slow" to "slowness," "modern" to "modernize"). - Compounding: Combining multiple free morphemes (e.g., "football," "blackboard"). Compounds are distinguished from phrases by stress (e.g., "Juvenile court" vs. "shooting defendant").
Hierarchical Structure of Words: Roots combine with suffixes to form stems, which can then take inflectional affixes (Example: Darwin [Root] + -ian [Suffix] + -ism [Suffix] + -s [Inflection] = "Darwinianisms").
Morphological Development and Challenges
Challenges: - Phonological effects: Allophones in rules (e.g., /s/ in "cats" vs. /z/ in "dogs"). - Exceptions: Irregular forms (e.g., "child" to "children," "man" to "men"). - Predictability: Meaning isn't always predictable (e.g., "neighborhood" is not the state of being a neighbor in the same way "brotherhood" relates to brothers; "fireman" does not bring fire).
Order of Acquisition for English (Brown & Cazden, 1968): 1. Present progressive (-ing) 2. Prepositions (in, on) 3. Plural (-s) 4. Irregular past tense (went) 5. Possessive ('s) 6. Articles (the, a) 7. Regular past tense (-ed) 8. Third person present (-s) 9. Auxiliary uncontractable (was) 10. Auxiliary contractable ('s)
The Wug Test (Berko, 1958): Elicited production task showing preschoolers can apply morphological rules to novel words (e.g., "This is a wug. Now there are two. There are two… wugs"). This proves children do not just memorize forms.
The Case of Past Tense: Rules vs. Memory
Theory 1: All Rules: Both regular and irregular forms are derived by rules (e.g., i-a rule for sing-sang). Problem: Rules should apply universally, but we don't say "thank" for "think."
Theory 2: Rules and Words (Pinker): Regular inflections are rule-based; irregulars are memorized entries in the lexicon. - Mechanism: Lexicon and grammar are accessed in parallel. If an irregular form is found (e.g., "held"), it blocks the general "-ed" rule.
Empirical Support for Pinker's Model: - Frequency: Frequency only affects irregulars (low-frequency irregulars like "slew" are regularized to "slayed" over time because they are hard to retrieve). - Similarity: Generalization of irregular patterns (spling-splang) depends on similarity to stored forms (ring-sang), whereas the regular rule applies regardless of sound similarity. - Neuropsychological Evidence: - Agrammatic Aphasics: Impaired at grammar/rules but better with irregulars. - Anomic Aphasics: Impaired at word retrieval (lexicon) but better with regular rules; often overregularize irregulars. - Parkinson's (PD) vs. Alzheimer's (AD): AD patients have more trouble with irregulars; PD patients have more trouble with regular suffixing.
Overregularization Errors and Creative Use
U-Shaped Development: Children initially produce irregulars correctly (sang), then enter a phase of overregularization (singed), and finally return to correctness.
Memory Limitation Hypothesis: Overregularization (rate is ~4% overall) occurs when a child fails to retrieve the memorized irregular form and falls back on the default rule.
Creative Morphology (Clark, 1982): Children create words to fill communicative gaps: - Compounds: "plate-egg" (fried egg). - New nouns: "lessoner" (teacher). - New verbs: "broom it" (sweep).