This study focused on the grammatical accuracy of Japanese students who were learning English. The database for the errors came from the Japanese University Student Corpus (JUSC) comprising 61 transcripts containing 51,061 words. An inventory, containing 400 errors in context, was taken from this corpus. The first research question related to the most common errors Japanese L2 learners made in spontaneous speech whereas the second question focused on the interpretability and recognition of errors as being intralingual or interlingual. Results showed that the primary errors were articles, verb tense, prepositions, omission, modifier errors, and subject-verb agreement. These results indicate that L1 is a factor in grammatical accuracy. We concluded that this data highlights the need for language teachers to focus on getting students to use the grammatical forms in actual dialogues.