The current study investigates how Korean L2 listeners identify English minimal pairs contrasting in voicing and manner and attempts to testify the essential tenets of the Perceptual Assimilation Model. Another purpose of our research is to explore other factors which might affect the identification patterns within the identical contrast categories proposed by PAM. Identification tasks were performed with target vowel and consonant contrasts within real word and nonce word conditions. Identification accuracy and reaction time were measured to evaluate the relative difficulty of L2 sound identification. First, we found that Korean listeners encountered the most difficulty in identifying L2 English minimal pairs in the Single Category contrast, followed by those in the Category Goodness contrast, and those in the TC contrast were most accurately and rapidly identified in accordance with the predictions of the PAM model. Next, it was shown that there were varying levels of difficulty in identification tasks within the TC and SC contrast. This finding suggests that PAM is not sufficient to predict L2 identification accuracy and other factors should be taken into consideration, including the Speech Learning Model, and other phonological factors such as manner of articulation, the vocalicity of the segments, individual contrast type, etc.