Previous research has shown that the knowledge structure (KS) complexity of a first language (L1) under certain conditions can strongly influence the KS complexity established in a second language (L2), and then this more complex L2 KS reciprocally influences L2 text comprehension. This present experimental investigation seeks to identify the unique contributions of mapping and writing in L1 (Korean) tasks to support L2 (English) science text comprehension using Pathfinder Network scaling, a graph-theoretic cognitive science approach. Native Korean low proficiency English language learners (n = 245) read a 708-word English (L2) science lesson text, completed one of seven treatment conditions, and then completed a comprehension posttest. The seven conditions consisted of three experimental conditions that required different L1 tasks including: L1 mapping alone, L1 writing alone, or both L1 mapping and writing; and four control conditions that did not receive any L1 treatment: L2 mapping alone, L2 writing alone, both L2 mapping and writing, or reading only. All of the maps and writing artifacts were converted into Pathfinder Networks that were compared to an expert's referent network. Results show that requiring L1 lesson tasks relatively increases L2 KS complexity and concomitant comprehension posttest performance. In order of effectiveness, combined L1 mapping and writing was most effective for posttest comprehension, then L1 writing, and least effective is L1 mapping alone. These findings confirm and extend the earlier findings that the inherent L1 KS complexity can strongly influence L2 KS complexity. Educationally, requiring L1 tasks, especially in text translation, likely engenders richer L2 structure that supports higher-order understanding of the text. Also, these findings further validate this technology-based approach for measuring KS contained in bilingual learners' productions.