We aim to construct an academic writing assistance system for learners of Japanese as a second language who are tasked with writing academic papers and reports during the course of their studies. We propose to provide suggestions for rewriting inappropriate expressions within learner compositions using a dictionary and a method of comparing learner composition text data to authentic corpus data. From among many possible language elements, this paper focuses on defining connective expressions and the construction of a connectives expression dictionary. Beginning with a definition of connectives based on their role in text, we select 568 connective expressions using a combination of manual selection and automated extraction, and demonstrate the importance of the automatic approach as well as its weaknesses. Next, we manually annotate the connective expressions with ratings on their suitability to be used in the academic register. Finally, we show the possibilities of the resulting connectives dictionary in providing guidance to learners in the appropriate selection of connective expressions within the academic register.