The articulatory in-out effect describes the preference for words with articulation moving inward compared to words with articulation moving outward. A promising explanation is that inward words are more fluent than outward words, but experimental evidence for such reasoning was offered only recently: By training selectively inward or outward words, fluency and consequentially liking was altered, leading to reversed or attenuated in-out effects when outward words were trained. However, it remains unclear whether such training procedures actually impact fluency of inward/outward movements, or whether they solely change the fluency of trained grammars. In two experiments (one preregistered, total N = 501), we show that training inward/outward words increases fluency and liking for trained grammar, but these effects do not generalize to in-out movements. The results show that training effects on liking reflect a structural mere-exposure effect rather than a change in liking for in-out motor movements. Findings are discussed regarding their implications for the fluency account, and the mental representation of inward and outward words.