This paper reports on an experiment in mathemagenic behavior ("Student inspection and processing activities that give birth to learning") as related to second-language vocabulary learning. The experiment was designed to determine whether visual mnemonics are more effective than unelaborated rehearsal technique for learning FL vocabulary, and whether a given visual mnemonic is more effective than one invented by the student. Seventy-five Spanish-language students with no prior training in German were randomly assigned to one of five groups, whose task was to learn the meaning of 24 German words as they were pronounced by native speakers. As each German word was pronounced on tape, the English meaning was flashed on the screen plus a mnemonic phrase. Group MG was told to use the mnemonic phrase in learning the word meanings; Group A was told to use a specific algorithm; Group A-8 was told to make up original algorithms; Group R was instructed to repeat the word and its meaning silently as many times as possible; Group C, the control group, was told to use any method. All groups were allotted 12 seconds per word except Group A-8, which had 20. A learning text was administered immediately after the instruction session, followed by a questionnaire and then by a retention test. A major conclusion is that strategies involving verbal or visual mnemonics were considerably more effective than a repetition technique, and that a self-generated visual mnemonic is more effective than a given one. (AM)