Quantifying expressive word learning in the presence of background speech