German Listeners Research Articles

Timing and Communicative Functions of Pitch Contours

A new research paradigm is applied to F0 synchronization with articulation, inpeak and valley contours, under four principles: (a) Timing of F0 contours enters thedefinitions of the pitch categories. (b) These phonetic exponents are linked to com-municativefunctions. (c) The listener plays a pivotal role. (d) Contextualization oftest stimuli is essential for pitch data collection. Data are presented and interpretedfrom an experimental investigation of the substance-function relationship in theperception of peak and valley shift series by German listeners, using the semanticdifferential technique. The findings of the substance-function relationship areexplained with reference to the frequency codeand to auditory enhancement,i.e.syntagmatic contrast of high-low or low-high pitch trajectories in synchronizationwith articulatory landmarks of accented syllables.

The role of prosody in the interpretation of structural ambiguities: A study of anticipatory eye movements

An eye-tracking experiment examined whether prosodic cues can affect the interpretation of grammatical functions in the absence of clear morphological information. German listeners were presented with scenes depicting three potential referents while hearing temporarily ambiguous SVO and OVS sentences. While case marking on the first noun phrase (NP) was ambiguous, clear case marking on the second NP disambiguated sentences towards SVO or OVS. Listeners interpreted case-ambiguous NP1s more often as Subject, and thus expected an Object as upcoming argument, only when sentence beginnings carried an SVO-type intonation. This was revealed by more anticipatory eye movements to suitable Patients (Objects) than Agents (Subjects) in the visual scenes. No such preference was found when sentence beginnings had an OVS-type intonation. Prosodic cues were integrated rapidly enough to affect listeners' interpretation of grammatical function before disambiguating case information was available. We conclude that in addition to manipulating attachment ambiguities, prosody can influence the interpretation of constituent order ambiguities.

Lexical status effects on compensation for fricative assimilation

In German, word-initial voiced fricatives may be devoiced if they follow /t/. This assimilation could make recognition of words beginning with /v/ harder, since there may be competing words beginning with /f/, but will not affect words beginning with /z/, since initial /s/ in German is illegal. Previous research has shown that speakers indeed produce more assimilation across word boundaries than across phrase boundaries for /z/, but inhibit the assimiliation across word boundaries for /v/. In the present study, German listeners identified the fricative continua /f–v/ and /s–z/, across word-versus-phrase boundaries, in viable-versus-nonviable contexts for assimilation. Less voicing was required for a /v,z/ judgment in viable than in nonviable assimilation contexts. This context effect was larger after a word boundary than after a phrase boundary. Within the viable-context condition, a prosodic effect appeared for /f–v/, with less voicing required for /v/ judgments after a word than a phrase boundary, but no such effect appeared for /s–z/. This asymmetry reverses the difference observed in production. Thus, listeners adjust phoneme category boundaries to compensate for prosodically conditioned variation where such adjustment is functional for word recognition, but show less compensation where adjustment would have no functional consequences.

Acoustic differences between German and Dutch labiodentals

The present article is a follow-up study of the investigation of labiodentals in German and Dutch by Hamann & Sennema (2005), where we looked at the perception of the Dutch labiodental three-way contrast by German listeners without any knowledge of Dutch and German learners of Dutch. The results of this previous study suggested that the German voiced labiodental fricative /v/ is perceptually closer to the Dutch approximant /ʋ/ than to the corresponding Dutch voiced labiodental fricative /v/. These perceptual indications are attested by the acoustic findings in the present study. German /v/ has a similar harmonicity median and a similar centre of gravity to Dutch /ʋ/, but differs from Dutch /v/ in these parameters. With respect to the acoustic parameter of duration, German /v/ lies closer to the Dutch /v/ than to the Dutch /ʋ/.

Language-specific relevance of formant transitions for fricative

Although the consonant inventories of Dutch, German, English, and Spanish are similar in size, the fricative inventories differ: English and Spanish distinguish labio-dental versus dental fricatives, whereas Dutch and German do not. Three phoneme-monitoring experiments investigated whether the relevance of formant transitions varies across languages, and whether it depends on the types of fricatives in a language. Native Dutch, German, English, and Spanish listeners detected a target fricative, /s/ or /f/, in nonwords. Half of the nonwords were cross spliced to produce misleading formant transitions: an /s/ replaced an /f/, or vice versa. Dutch and German listeners were unaffected by the misleading formant transitions, whereas Spanish listeners missed significantly more fricatives surrounded by misleading formant transitions; these results were obtained whether stimuli were originally spoken by a Dutch or Spanish speaker. English listeners showed the same sensitivity to formant transitions as the Spanish. Despite previous reports that formant transition cues are of negligible significance for fricative identification (Klaassen-Don, 1983), the present findings show that formant transitions are indeed relevant for listeners whose native language distinguishes labio-dental versus dental fricatives. Listeners relied on the transitions even though no dental fricatives occurred in these stimuli, and independently of the speaker’s native language.

Assimilation violation and spoken-language processing: a supplementary report.

Previous studies have shown that spoken-language processing is inhibited by violation of obligatory regressive assimilation. Weber (2001) replicated this inhibitory effect in a phoneme-monitoring study examining regressive place assimilation of nasals, but found facilitation for violation of progressive assimilation. German listeners detected the velar fricative [x] more quickly when fricative assimilation was violated (e.g., see text) than when no violation occurred (e.g., [baxt] or [see text]). It was argued that a combination of two factors caused facilitation: (1) progressive assimilation creates different restrictions for the monitoring target than regressive assimilation does, and (2) the sequences violating assimilation (e.g., *[ix]) are novel for German listeners and therefore facilitate fricative detection (novel popout). The present study tested progressive assimilation violation in non-novel sequences using the palatal fricative [ç]. Stimuli either violated fricative assimilation (e.g., *[see text]) or did not (e.g., [see text]). This manipulation does not create novel sequences: sequences like *[see text] can occur across word boundaries. while *[ix] cannot. No facilitation was found. However, violation also did not significantly inhibit processing. The results confirm that facilitation depends on the combination of progressive assimilation with novelty of the sequence.

Help or hindrance: how violation of different assimilation rules affects spoken-language processing.

Four phoneme-detection studies tested the conclusion from recent research that spoken-language processing is inhibited by violation of obligatory assimilation processes in the listeners' native language. In Experiment 1, native listeners of German detected a target fricative in monosyllabic Dutch nonwords, half of which violated progressive German fricative place assimilation. In contrast to the earlier findings, listeners detected the fricative more quickly when assimilation was violated than when no violation occurred. This difference was not due to purely acoustic factors, since in Experiment 2 native Dutch listeners, presented with the same materials, showed no such effect. In Experiment 3, German listeners again detected the fricative more quickly when violation occurred in both monosyllabic and bisyllabic native nonwords, further ruling out explanations based on non-native input or on syllable structure. Finally Experiment 4 tested whether the direction in which the rule operates (progressive or regressive) controls the direction of the effect on phoneme detection responses. When regressive German place assimilation for nasals was violated, German listeners detected stops more slowly, exactly as had been observed in previous studies of regressive assimilation. It is argued that a combination of low expectations in progressive assimilation and novel popout causes facilitation of processing, whereas not fulfilling high expectations in regressive assimilation causes inhibition.

Measuring the perceptual magnet effect in the perception of /i/ by German listeners.

In a study of the internal category structure of the vowel /i/, Kuhl found a "perceptual magnet effect": Discrimination sensitivity was poorer for category instances that were acoustically similar to the category prototype than it was for category instances that were not. The typicality of category exemplars was determined by goodness judgments and was found to correlate with the acoustics of average production. Analysis and interpretation of discrimination performance relied on two important assumptions: that listeners perceived all stimuli presented as exemplars of the same vowel category and that, apart from the influence of phonetic coding, discrimination sensitivity was the same across the investigated part of the vowel space. In the present study, it is shown that production and perception estimates of the category prototype may diverge, possibly because listeners seem to prefer hyperarticulated variants of vowel categories. An approach towards measurement of intra-category discrimination minima is put forward and tested that protects against intercategory confounds and avoids the isosensitivity assumption.

Different influences of the native language of a listener on speaker recognition

In forensic phonetics, lay or expert witnesses might be confronted with voice samples for auditory evaluation from a language they do not understand. In speaker identification experiments, it has been shown that knowledge of the target language affects recognition results. Köster et al. (1995) showed that German listeners and English listeners with a knowledge of German identified a German voice better than English listeners without knowledge of German. Replicating the same experiment with Spanish and Chinese listeners, the results of this study show that (a) Spanish and Chinese listeners with knowledge of German obtain significantly better recognition results than their compatriots with no knowledge of the target language, and that (b) Spanish and Chinese listeners with knowledge of German perform significantly worse than native Germans and English listeners with a knowledge of German. No clear evidence was found that the typological difference between the native language of the listener and the target language influenced recognition performance.

Cross-language tests of the perceptual magnet effect for /inverted are/ and /l/

Recent experiments by Iverson and Kuhl have suggested that the perceptual organization of the American English /inverted are/ and /l/ categories is strongly influenced by category goodness. American adult listeners exhibit a perceptualmagneteffect characterized by low sensitivity and perceptual clustering near the best exemplars of /inverted are/ and /l/ and high sensitivity and stretched perceptual distances near the worst exemplars. The present study compares the responses of native German and English speakers to evaluate whether the representation of these categories is influenced by language experience. In separate experiments, natural phonemes of adult German (/inverted are/ and /l/) and American (/inverted are/ and /l/) speakers were recorded, and synthesized American English /inverted are/ and /l/ tokens were identified, discriminated, rated for category goodness, and rated for similarity by both groups of subjects. The results demonstrate that German listeners prefer, produce, and have perceptual clustering for /l/ phonemes with a higher F2 frequency than do American listeners. German listeners do not perceive American /inverted are/ phonemes to be good examples of the German /inverted are/ category, and German listeners also exhibit somewhat less perceptual clustering for /inverted are/ phonemes than do American listeners. The results confirm that the perceptual magnet effect for /inverted are/ and /l/ is determined by language experience. [Work supported by NIH.]

Neutralization of syllable-final voicing in German

German is well known for its neutralization of the voicing contrast in word-final obstruents. However, acoustic analysis of ten pairs of German words produced by ten native speakers revealed that the distributions of acoustic parameters for underlying voiced and voiceless stops are significantly different, despite considerable overlap between the two categories. Furthermore, in a listening test, German listeners were able to distinguish the voiced and voiceless pairs with about 60% accuracy—significantly better than chance. Several mechanisms for such a “semicontrast” are considered, including the possibility that it is an experimental artifact. It is concluded that this is not an artifact, yet that it probably serves no communicative function whatsoever. Although there remain several possible explanations, the one that is preferred is that, in addition to the usual implementation rules for the production of segmental units of speech (such as [+ voice] and [–voice]), we are observing here the results of an implementation rule that directly implements the German syllable —one which “devoices” syllable—final obstruent portions of a syllable. This accounts in a natural way for both differences between the two kinds of voiceless stops and for the tendency for this property to carry over so strongly into German-accented speech in other languages.