Phonetic accommodation depends on processing conditions: evidence for task-dependent frequency effects.

This study investigates how lexical frequency, behavioural task, and gender shape phonetic accommodation, focusing on speakers' adoption of an alternative pronunciation variant of the German adjectival suffix -ig. Partially overlapping groups of 30-31 native speakers of German completed two experimental tasks: a shadowing task involving immediate repetition of the same target adjective, and an interaction task based on question-answer exchanges in which the perceived and produced -ig adjectives were different lexical items but belonged to the same lexical-frequency category. Lexical frequency was manipulated by selecting high- and low-frequency stimuli, and accommodation was operationalised as a categorical shift from each speaker's dominant baseline pronunciation towards the alternative variant presented in the stimulus. Accommodation behaviour was analysed using a generalised linear mixed-effects model. The results show that accommodation increased with exposure, was stronger in shadowing than in interaction, and was greater overall for women than for men. Accommodation also differed depending on the direction of change, whereas stimulus voice gender did not play a role. Crucially, lexical frequency affected accommodation only in the shadowing task: low-frequency items elicited greater accommodation than high-frequency items, while no comparable frequency effect emerged in interaction. This task-dependent pattern suggests that lexical-frequency effects are strongest when perceived and produced forms are temporally adjacent and lexically identical, as in immediate repetition of the same word. When accommodation requires generalisation across different words sharing only the same suffix, as in the interaction task, frequency effects may be attenuated or fail to emerge. These findings support a hybrid account in which accommodation reflects the interaction of the link between perceived and produced forms, task-specific processing demands, lexical frequency, and speaker-related factors.