The current study compared equivalence‐based instruction (EBI) to self‐study of American Sign Language (ASL) videos to teach eight 3‐member classes of signs with 24 college students. Four of the equivalence classes consisted of verbs (i.e., throw, touch, blow, spin) and four classes were object nouns (i.e., truck, dollar, egg, ball). Each class consisted of (a) a picture corresponding to the targeted sign, (b) a video clip depicting the sign's topography, and (c) a printed corresponding word. Pretests and posttests assessed the degree to which participants (a) showed emergence of classes during match‐to‐sample (MTS) tests, (b) could sign both single words and verb‐noun pairs (i.e., speaker tasks), and (c) could comply with signed requests consisting of ASL verb‐noun pairs (i.e., listener tasks). EBI consisted of training the AB relation (picture to video clip) followed by tests of all possible derived relations. The AC relation (picture to printed word) was not trained because it was assumed to be present in the participants' repertoires. Self‐study consisted of viewing video clips of the eight ASL signs simultaneously presented with the printed word. The MTS posttest showed equivalence classes emerged for all participants from both EBI and self‐study groups. In addition, all EBI participants met passing criterion for emitting all single signs, all verb‐noun sign pairs, and they complied with all listener tasks. In contrast, only 4 of 12 participants from the self‐study group emitted the single signs. After re‐exposure to self‐study, most participants then passed all posttests. These results are important because they demonstrate recombinative generalization of ASL signs, which increases the utility of ASL skills.