Second language acquisition (SLA) becomes extremely difficult for late second language (L2) learners, who are assumed to have passed the sensitive or critical period for L2 learning. As one of the major accounts of the post-critical period L2 learning processes, a fundamental distinction between explicit and implicit learning and knowledge was proposed over three decades ago. The first goal of the current study was to develop fine-grained measures for implicit knowledge to distinguish it from automatized explicit knowledge. The second goal was to use these validated measures to explore the interface issue of explicit and implicit knowledge by correlating these measures with several cognitive aptitudes.

One hundred advanced L2 Japanese speakers whose first language was Chinese were recruited; they were given tests for both automatized explicit knowledge and implicit knowledge, along with three cognitive aptitude measures. The present study developed three psycholinguistic tasks that can reliably assess implicit knowledge (the eye-tracking-while-listening task, the word-monitoring task, and the self-paced reading task) and compared them with the existing tasks that have been claimed to measure implicit knowledge (time-pressured form-focused tasks like grammaticality judgment tasks), but which we hypothesized tap into automatized explicit knowledge. The aptitude test battery consisted of LLAMA F, a measure of explicit learning aptitude, the Serial-Reaction Time (SRT) task, a measure of implicit learning aptitude, and the letter-span task, a measure of phonological short-term memory.

In order to validate the measures for automatized explicit knowledge and implicit knowledge, a series of confirmatory factor analyses (CFA), multi-trait multi-method (MTMM) analyses, and structural equation model (SEM) analyses were conducted. Results confirmed that the existing tasks purported to measure implicit knowledge in fact tap into automatized explicit knowledge, whereas the new psycholinguistic measures tap into implicit knowledge. For the participants as a whole, the convergent validity for implicit knowledge measures was less than ideal. When the results were analyzed separately by length of residence, however, acceptable convergent validity for implicit knowledge was obtained for those with longer length of residence but not for those with shorter length of residence.

In order to address the interface issue, SEM analyses were conducted to investigate the relationship between automatized explicit knowledge and implicit knowledge. Results showed that automatized explicit knowledge significantly predicted the acquisition of implicit knowledge. Furthermore, the aptitude for explicit learning was the only significant predictor of the acquisition of automatized explicit knowledge, not for the acquisition of implicit knowledge. The effects of implicit learning aptitude and phonological short-term memory on the acquisition of both types of linguistic knowledge were limited.

In conclusion, the study demonstrated that the newer measures for implicit knowledge are more sensitive and opens up promising directions for developing additional fine-grained measures for implicit knowledge. The current findings provide the first empirical evidence at the latent construct level that automatized explicit knowledge, which develops through explicit learning mechanisms, impacts the acquisition of implicit knowledge.