Breaking the brackets: Semantic and morphosyntactic cues in the interpretation of bracketing paradoxes

Compositionality is fundamental to language, yet some German adjective and noun compound constructions, called bracketing paradoxes, challenge this principle. Bracketing paradoxes are phrases such as “internationaler Bestsellerautor” (international author of bestsellers) in which the adjective attaches to the non-head noun of the compound, against strict grammatical constraints. These constructions create a systematic mismatch between syntactic structure (the adjective needs to agree with noun 2) and semantic interpretation (the adjective modifies the meaning of noun 1), but the factors licensing such interpretations remain poorly understood. Across three experiments, we examined two factors that potentially promote a bracketing paradox interpretation: the semantic compatibility of an adjective with each noun in a compound, and the morphosyntactic congruence between the adjective and the non-head noun. Experiment 1 elicited acceptability judgments for the adjective together with the nouns in a compound individually, and established a corpus of 204 bracketing paradox phrases. Experiment 2 assessed the attachment preferences for the full adjective and compound phrases using a forced-choice task. It revealed that readers routinely attach the adjective to the non-head noun, yielding robust bracketing paradox interpretations. Experiment 3 investigated whether the adjective attachment is a categorical decision or whether the adjective can be simultaneously attached to both head and non-head nouns, showing that many phrases support multiple plausible attachments. The results show that semantic effects ultimately govern bracketing paradox interpretations with morphosyntactic congruence between the adjective and the non-head noun having only a minor supporting effect. These findings suggest that semantic cues can override grammatical constraints and may temporarily suspend compositional processing. We tentatively interpret the results as evidence for competing constraints during interpretation, with pragmatic ambiguity-resolution processes shaping the final reading of bracketing paradoxes. The database of bracketing paradox phrases created through these experiments offers a rich resource for investigating other factors contributing to a non-compositional interpretation.