Psycholinguistics investigates the cognitive mechanisms underlying language comprehension, production, and acquisition — how the mind processes language in real time. Sentence processing research reveals that comprehenders build syntactic structure incrementally, word by word, sometimes committing to an analysis that later proves incorrect — the garden-path effect ("The horse raced past the barn fell" momentarily misleads readers into a main-clause parse). Lexical access studies show that word recognition involves rapid parallel activation of candidates: hearing "cap-" activates both "captain" and "capital" before context resolves the competition. Priming experiments demonstrate that processing a word is facilitated by prior exposure to a semantically or phonologically related word, revealing the associative architecture of the mental lexicon.
Experience garden-path sentences firsthand — read "The old man the boats" and notice the moment of reanalysis, then analyze why the parser initially went wrong. Design a simple lexical decision task (is this a real word?) and predict which pairs should show priming effects. Read about eye-tracking studies of reading to understand how fixation patterns reveal real-time processing difficulty at specific words.
Psycholinguistics asks: what is the mind actually doing when you understand a sentence? Linguists describe grammatical structures abstractly, but the human parser must operate in real time — word by word, in milliseconds — under working memory constraints and without the luxury of pausing to reconsider. What that real-time system does, and how it sometimes fails, is what psycholinguistics studies.
The most striking discovery from sentence processing research is that comprehension is incremental and committed. Your parser does not wait politely until the period before assigning syntactic structure. Instead, it builds an analysis word by word, immediately committing to the most probable interpretation given what it has seen so far. Read "The old man the boats" slowly. Your parser almost certainly tried to read "old" as an adjective modifying "man" (building the noun phrase "the old man"). When "the boats" arrives, that structure fails: "the old" must be nominalized (old people), and "man" must be a verb. The discomfort you feel at that point is the cost of reanalysis — the parser's initial commitment being undone. The garden-path effect makes the parser's incremental commitment visible precisely by making it wrong.
Lexical access — retrieving a word from the mental lexicon — is equally rapid and parallel. When you hear the beginning of a word (like "cap-"), your lexicon activates multiple candidates simultaneously (captain, capital, capsule, caption), and context gradually narrows the competition. This parallel activation, not serial alphabetical search, is what enables conversational-speed word recognition. Priming experiments make the architecture visible: prior exposure to a word (like "doctor") measurably speeds recognition of semantically related words (like "nurse"), because semantic associations in the lexical network cause partial activation of connected entries. The effect is automatic and operates below awareness.
Your background in lexical semantics gives you the vocabulary for describing what the mental lexicon contains — sense relations, polysemy, semantic fields. Psycholinguistics adds the processing dimension: those semantic relations are not merely theoretical constructs but active cognitive connections that fire during real-time comprehension. The mental lexicon is not the abstract meaning structure described by semantics; it is the cognitive reality that executes that structure in milliseconds, under the constraints of attention and memory.
Finally, it is important to distinguish competence — what grammatical knowledge a speaker possesses — from performance — how that knowledge is deployed under real-time constraints. The parser is a performance mechanism: it must make rapid decisions with limited working memory, so it uses heuristics (prefer simple structures; prefer the most frequent reading; use context early). When those heuristics fail, we get garden-paths and reanalysis costs. These failures are not evidence of poor grammar knowledge; they are evidence of an efficient, fast system operating at its working-memory limits — and they are the primary data through which psycholinguists infer the system's design.