@cite{sedivy-etal-1999} #

Achieving incremental semantic interpretation through contextual representation. Cognition 71(2), 109–147.

Empirical Phenomenon #

Visual-world eye-tracking with prenominal adjective instructions. Listeners hearing "the tall glass" in a display containing both a tall glass and a short glass identify the target faster — and consider an unrelated tall pitcher less — than in a display lacking the same-category contrast pair.

The paper reports three experiments. Experiment 1 (subexperiments 1A and 1B, intersective adjectives, N = 12) replicates @cite{eberhard-etal-1995}'s incremental processing finding and shows contrastive interpretation of color/material/shape adjectives. Experiments 2 (N = 24) and 3 (N = 22) probe the mechanism for vague scalar adjectives such as "tall" that lack a stable denotation, contrasting a definite-NP instruction task with an indefinite-NP verification task.

Paradigm #

Built on Paradigms.VisualWorld (@cite{huettig-rommers-meyer-2011}). The display contains four objects (ObjectRole):

target: the intended referent (e.g. tall glass).
contrastingObject: same category, opposite scale pole (short glass). Present only in the contrast condition.
crossCategoryCompetitor: different category but further along the scale (pitcher, taller than the target glasses). Always present.
distractor: unrelated object (key). Always present.

In the no-contrast condition the contrasting object is replaced by a second distractor. Three within-subjects manipulations are crossed:

Contrast (ContrastCondition): contrasting object present vs. replaced.
Typicality: target is a good vs. poor exemplar of the modified NP.
Task (Task): instruction with definite NP (Exp 2, directAction) vs. verification with indefinite NP (Exp 3, verification).

Architectural Role #

This file is an empirical anchor: it defines the experimental cells and qualitative predicates that downstream theoretical models must satisfy. A theory of incremental adjective processing predicts a numeric response for each Cell; the predicates PredictsContrastSpeedsTarget, PredictsContrastReducesCompetitorLooks, and PredictsContrastAttenuatesTypicality constrain those predictions to match the qualitative empirical patterns. F-statistics from the paper are documented in prose at each predicate; numeric means are reproduced only for the norming study (Table 5), which is itself an empirical baseline that operationally defines the typicality manipulation.

Theoretical Significance #

The paper's central theoretical contrast pits two accounts of the contrast effect:

Referential / presupposition account (@cite{altmann-steedman-1988}): the contrast effect derives from definiteness presuppositions on the modified NP.
Lexical comparison-class account (@cite{bierwisch-1989}): scalar adjectives carry a free comparison-class variable in their lexical entry, bound by the contextually salient set.

Experiment 3's indefinite verification task removes the definiteness presupposition while preserving the contrast effect, supporting the lexical account. The qualitative patterns universally quantify over Task, so any theory satisfying the patterns is committed to a task-invariant mechanism.