Perceptual mapping
Languages use different systems to distinguish words through prosody: some rely on stress (like Dutch and Spanish), others on tone (like Yoruba and Cantonese), lexical pitch accent (like Swedish and Serbian), or have no lexical prosodic contrasts at all (like French and Korean). Because these systems use different acoustic cues, such as pitch, duration, and intensity, speakers of different languages may not perceive unfamiliar prosodic contrasts in the same way. There are two main explanations for why this happens. The first, the Native Language Acoustic Cue Approach, argues that listeners are better at perceiving non-native contrasts when they already use the relevant acoustic cues to distinguish words in their own language. For example, speakers of tonal languages tend to be better at perceiving tone contrasts than speakers of non-tonal languages, while French speakers often struggle with lexical stress because French does not use stress to distinguish words. However, this does not mean that listeners are completely unable to perceive contrasts that are unfamiliar to their native language; rather, their sensitivity can vary depending on how relevant a particular cue is in their language. The second explanation, the Target Language Acoustic Cue Approach, puts more emphasis on the acoustic properties of the language being learned and how they interact with the listener’s existing perceptual system. In other words, what matters is not necessarily whether a listener’s native language uses the same type of prosodic contrast, but whether the acoustic cues of the unfamiliar language are easy to distinguish and how they relate to cues the listener already knows. This helps explain why some studies have found that French speakers can still develop representations of unfamiliar stress contrasts, or that English speakers can successfully identify Serbian pitch accents despite not having lexical pitch accents in their native language. The revised Perceptual Assimilation Model also suggests that some contrasts are naturally harder to distinguish because of their acoustic characteristics, even for speakers of languages that use tone. Another relevant framework is Mennen’s L2 Intonation Learning Theory, which looks at differences between languages in terms of the categories they use, how those categories are acoustically realized, what they mean, and how frequently they occur. Overall, the main point is that non-native prosody perception cannot be explained simply by whether a listener’s first language has the same prosodic system. Both the listener’s native language and the acoustic characteristics of the target language can shape perception, and the present study builds on this by examining how native Serbian and non-native French listeners perceive Serbian lexical pitch accent contrasts.
The study looked at how native Serbian speakers and non-native French speakers perceive and group the four types of Serbian lexical pitch accents. The results showed that Serbian listeners generally distinguish between the four accent types, although they sometimes confuse short-rising and short-falling accents. This is not too surprising, since the difference between these two accents is often quite subtle, especially because the F0 movement in short-rising accents can be very limited. In fact, some Serbian dialects, including Belgrade speech, do not consistently maintain this distinction at all. Interestingly, the results also suggest that vowel length plays a more important role than pitch in how native Serbian listeners perceive lexical pitch accents, which was supported by the finding that length, but not pitch, correlated significantly with one of the perceptual similarity solutions. French listeners, on the other hand, tended to group the accents differently, creating three main categories and having particular difficulty with the long-falling accent, which they often confused with other accent types. One possible explanation is that French does not use pitch to distinguish words at the lexical level, but it does use pitch movements at the phrase level, so listeners may be transferring patterns from their native intonation system when processing Serbian accents. The acoustic properties of the long-falling accent may also make it particularly difficult to identify, since its pitch peak and fall can occur at different points within the vowel, making it sound more similar to either a rising or a falling accent depending on its realization. Despite these difficulties, French listeners were actually quite successful at distinguishing most of the Serbian pitch accents, even though their native language does not have lexical pitch accents or lexical prosodic contrasts. This is important because it challenges the idea that listeners can only perceive unfamiliar prosodic contrasts when their native language uses the same acoustic cues for distinguishing words. Instead, the findings support the Target Language Acoustic Cue Approach, suggesting that listeners can use the acoustic properties of the target language, such as pitch movement and vowel duration, even when these cues serve different functions in their native language. More broadly, the study suggests that listeners can transfer perceptual strategies across different prosodic domains, for example, from phrase-level intonation in French to word-level pitch accent perception in Serbian. So, the main takeaway is that perceiving non-native lexical prosody is not simply a matter of whether your native language has the same prosodic categories; it also depends on how listeners pick up, integrate, and interpret the actual acoustic cues in the unfamiliar language.