Acoustical and perceptual analysis of the voicing distinction in Dutch initial plosives: The role of prevoicing

TLDR: The paradox raised is that although prevoicing is the most reliable cue to the voicing distinction for listeners, it is not reliably produced by speakers.

About: This article is published in Journal of Phonetics.The article was published on 2004-10-01 and is currently open access. It has received 116 citations till now. The article focuses on the topics: Voice & Voice-onset time.

Figures

Table 1 Full design of Experiment 1

Fig. 10. Experiment 2: histograms of the F0 difference as produced by 10 different speakers, plotted separately for place of articulation (left versus right) and voicing category (top versus bottom). The numbers on the x-axis represent the upper limits of each bin.

Table 5 Experiments 2 and 3: mean percentage of voiced decisions and correct (that is, in agreement with the intention of the speaker) decisions made by the listeners and by the CART analysis (Experiment 3), if it were forced to label tokens as voiced or voiceless on the basis of the probabilities at the end nodes. The numbers in the last row represent the mean percentage of such forced CART decisions in agreement with the decisions made by the majority of the listeners

Fig. 14. Experiment 3: CART analysis of the proportion of voiced responses for the alveolar plosives. Tokens satisfying the rule printed at the top of each split follow the left branch. For each of the final nodes the mean of the proportion voiced responses for the tokens falling in that final node is printed underneath. The number within parentheses indicate the total number of tokens falling in that final node. The histograms show the distributions of the mean proportions voiced responses for the tokens falling in that final node.

Table 3 Experiment 2: mean values for each of the six measures separately for the two places of articulation and two voicing categories and the four combinations of these two factors

Table 6 Materials Experiment 1

Frequently asked questions

Q1. What are the contributions in "Acoustical and perceptual analysis of the voicing distinction in dutch initial plosives: the role of prevoicing" ?

The authors discuss the paradox raised by these findings: although prevoicing is the most reliable cue to the voicing distinction for listeners, it is not reliably produced by speakers. 

Q2. What are the future works in "Acoustical and perceptual analysis of the voicing distinction in dutch initial plosives: the role of prevoicing" ?

Future research on other prevoicing languages and the influence of other language in which prevoicing is not important, should give more insight into this paradox between production and perception. 

Q3. What are the obvious analyses to examine the relative strengths of thevarious acoustic?

The obvious analyses to examine the relative strengths of thevarious acoustic properties for recognition are linear discriminant analysis or logistic regression analysis. 

Q4. What did the presence of prevoicing bring the probability of a voiced response?

The presence of prevoicing alone brought the probability of a voiced response so close to unity that variation in the other cues had no discernible effect. 

Q5. What was the strongest cue for listeners to identify a plosive as voice?

The CART analyses showed that the presence or absence of prevoicing was by far the strongest cue for listeners to identify Dutch initial plosives as voiced or voiceless. 

Q6. What is the effect of a raised intraoral pressure on the oral cavity?

As described earlier, one phenomenon that helps to maintain sufficient transglottal pressure is passive enlargement of the oral cavity due to raised intraoral pressure. 

Q7. What was the mean proportion of voiced responses for the plosives produced without prevoicing?

The plosives produced with prevoicing were consistently perceived as being voiced (mean proportion of voiced responses was 0.99 for labials and alveolars), which is also shown in the histograms below the final nodes.