While Standard (Tokyo) Japanese has a lexical tonal system known as 'lexical pitch accent', there are some varieties of Japanese, called 'accentless' dialects, which do not have any lexical tonal phenomena. We investigated the differences in the perception of lexical pitch accent between the speakers of the accentless dialect and those of Standard Japanese, and the robustness of two approaches to investigate such dialectal differences. We conducted two experiments: categorical perception and sequence recall experiments. The former is an approach that has been traditionally employed to study the perception of phonological contrasts. The latter is a more recent method employed in studies of 'stress-deafness' in French by Dupoux and his colleagues, in which participants listen to sequences of several nonsense words and answer the order of the words. The results of the categorical perception experiment showed no clear dialectal differences. On the other hand, the results of the sequence recall task showed that the scores of the 'accentless' group were clearly lower than those of control (Standard Japanese) participants in the discrimination of nonsense words whose pitch accent differences corresponded to lexical differences in Standard Japanese phonology. Thus, it is concluded that the latter experimental approach is more robust to study dialectal differences in pitch accent perception than the former.