Temporary ambiguities can occur when a verb phrase can plausibly attach high or low, e.g.,*The uncle of the girl who will ride the carousel is…*, where *ride the carousel* can attach high (*The uncle*) or low (*the girl*). Kamide (2012) showed that listeners may learn individuals’ attachment biases (high/low) from exposure and use the newly-learned speaker-specific biases to resolve ambiguities. We test if speaker-specific biases generalize to new speakers who share features (e.g., gender/ accent). To preview, we fail to find evidence for generalization. We then attempt to replicate Kamide (2012), but find only weak support for the original claims.