This document presents an approach for automatically recommending templates for legal requirements based on natural language statements extracted from legal texts. The approach identifies the statement type and main verb to select a template with no counterpart, correlative counterpart, or implied counterpart. An evaluation on two legal corpora achieved 79.8% accuracy. Key challenges include handling nominalization, phrasal verbs, and legal terminology. Incorporating domain knowledge and human feedback improved recall and precision. The work aims to help requirements analysts elicit legal requirements focused on permissions and obligations.