Dynamic attending at the cocktail party through entrained oscillations in the brain: A review across three languages

Everyday listening situations often require the listener to process speech in the presence of other, concurrent speech streams. The so-called cocktail party problem is frequently analyzed as a two-step process, consisting of the segregation of the acoustic mixture into separate speech streams (perceptual stage) and selecting stream(s) for deeper processing (cognitive stage). The specific linguistic characteristics of a given language have been suggested to affect the perceptual stage. However, this dichotomous view of speech processing in a cocktail party situation has been challenged in recent years. To better understand the interplay of perception and cognition, the present review article takes a cross-linguistic perspective by analyzing cocktail parties in three languages that widely differ in temporal properties such as speech rhythms and stress patterns: English, Mandarin Chinese, and Hungarian. We included studies (n = 68) from Web of Science and other sources focusing on the cognitive processing of intelligible speech in a cocktail party situation, and conducted a qualitative thematic analysis. We found that studies using well-controlled, simplified speech material report the effects of perceptual and cognitive factors separately, while studies using more complex stimuli capture the close interplay of perception and attention. We interpret the results in light of the Dynamic Attending Theory, which integrates perception and attention by proposing that attentional oscillations track events, enhancing perception at predicted time points. We argue that this theory provides a useful theoretical framework, not only for interpreting the cognitive and neural mechanisms behind speech-on-speech masking and their potential language-specific characteristics, but also for studying the remaining open questions around these issues.