Is the learning of artificial phonotactic rules interfered with by the concurrent experience of English?

Adults can rapidly learn new first-order phonotactic constraints like/f/only occurs at the beginning of syllables, by producing strings of nonsense syllables such as "hes feng neg kem". The learning is measured by observing their speech errors, e.g., whether/f/'s then always slip to syllable onset position. Context-dependent (second-order) constraints such as/f/occurs at the beginning of syllables if the vowel is/æ/, but occurs at the end of syllables if the vowel is/ı/can be learned as well, but errors only follow these constraints after a period of sleep. It has been suggested that the knowledge of newly-learned second-order constraints is isolated from English knowledge in a separate "mini-grammar" and that the creation of the mini-grammar requires a period of sleep. The present study investigates the mini-grammar notion in the learning of first-order constraints, which are learned quickly in a single session. We interleaved trials in which participants produced strings of nonsense syllables with trials in which they repeated English sentences. The English sentences and nonsense sequences either showed the same consonant-position constraints or the opposite constraints. Speech error data showed that the English sentences interfered with the learning of the first-order constraints within the nonsense sequences, suggesting that the constraints in the nonsense context were not separated from ordinary English in a mini-grammar. We hypothesize that the formation of mini-grammars may require consolidation and that no mini-grammar is created for first-order constraint learning.