The questions of whether transfer of learning across languages reflect true cross-language influence and whether languages coactivated at encoding impact long-term memory performance are considered by reviewing relevant evidence from bilingual memory research.
Several studies indicate that content learned in one language can be retrieved in the other language, whether intentionally or automatically, but these effects appear to be based on non-linguistic representations or processes. In explicit memory, findings that words or sentences studied in one language are easily recalled or recognized at test through other known languages can be explained through encoding and retrieval of shared concepts or intentional translation. In implicit memory, where tests do not involve the intentional retrieval of studied items, transfer across languages can be explained by repetition of non-linguistic processes, such as object identification, semantic associate selection, and response selection. Cross-language effects are also observed in more complex tasks that involve both explicit and implicit memory elements, such as savings in learning, false memory, and analogical transfer, but are similarly explained by shared conceptual representations rather than cross-language influence.
In word production, co-activation of a non-target language might be expected to impact immediate processing. However, neither co-activation nor actual spoken production in one language has a lasting impact on spoken production in another language. Similarly, explicit memory does not appear to be affected by spreading activation across languages. Although these factors do not appear to impact long-term memory, there may be other ways in which knowledge of another specific language, language contexts, or studying mixed-language material impacts encoding and retrieval processes, but these possibilities have not been a focus of investigation. Although it is tempting to assume that cross-language influences in long-term episodic memory would be the same as in language processing, there is ample evidence to the contrary.