Cross-Regional Patterns of Obstruent Voicing and Gemination: The Case of Roman and Veneto Italian
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsI find this article excellent.
Well described, well referenced. Missing, from a very extensive bibliography, are perhaps only a few titles that could be considered secondary (if other secondary titles were not included).
In particular, the critical issue remains that of considering the segments described for the Roman as (pre-)voiced. The question might be more subtle, if one were to consider e.g. that authors such as Luciano Canepari speak of a murmured voice in these circumstances, and were it not that the I myself has worked on similar phenomena in other varieties on the basis of an important article by Trumper & Mioni (1975) in which the phenomenon is described as lenition.
J. TRUMPER, A.M. MIONI, Osservazioni sulla lenizione nei dialetti salentini e pugliesi, in “Lingua e contesto”, 1, 1975, pp. 167-177
Author Response
Comment: “I find this article excellent.
Well described, well referenced. Missing, from a very extensive bibliography, are perhaps only a few titles that could be considered secondary (if other secondary titles were not included).
In particular, the critical issue remains that of considering the segments described for the Roman as (pre-)voiced. The question might be more subtle, if one were to consider e.g. that authors such as Luciano Canepari speak of a murmured voice in these circumstances, and were it not that the I myself has worked on similar phenomena in other varieties on the basis of an important article by Trumper & Mioni (1975) in which the phenomenon is described as lenition.
- TRUMPER, A.M. MIONI, Osservazioni sulla lenizione nei dialetti salentini e pugliesi, in “Lingua e contesto”, 1, 1975, pp. 167-177”
RESPONSE:
Thank you very much for the positive feedback. We are aware of Canepari’s interpretation of lenited voiceless singletons as produced with ‘murmured’ voice. However, as indicated in the introduction and discussion sections, we have chosen to place the article within the more recent frameworks of Laryngeal Realism and Relativism, which specifically look at patterns of low-frequency periodicity as observed in the acoustic signal. The reference you provide is nevertheless relevant, and we thank you for it. We have included it among the cited sources covering the issue of intervocalic lenition in Italian (lines 194-195 of the PDF showing markups).
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper offers a very detailed contribution addressing the length contrast, which is present and productive in the Italian language, by taking into account the role of consonant voicing and the interplay between vowel and consonant. The focus of the paper is on a selection of Italian voiced and voiceless obstruents in their geminate and singleton realization in two specific varieties of Italian, namely Veneto Italian (VI) and Roman Italian (RI), two varieties that present some phonetic differences on the length contrast: Veneto Italian (VI) - as well as other Northern varieties - is described as maintaining the voicing contrast but in some cases not the length contrast; in Roman Italian (RI) - along with other Central and Southern varieties - the opposite may occur by, for example, pre-voicing intervocalic voiceless singleton obstruents whilst also maintaining the length contrast in the production of obstruents.
The adopted methodology relies on acoustic phonetic analysis of read data collected from speakers of the varieties of regional Italian above mentioned.
The authors acoustically analyse the consonant's prevoicing patterns and the effects and interactions of regional variety, gemination, and (phonological and phonetic) voicing on consonant (C) and preceding vowel (V) durations. Moreover, the analysis focuses on the less used C/V ratio between the two (C/V). Based on outcomes of this study, which are also supported by a careful statistical analysis, the authors show that speakers of both varieties produce geminates with high C/V ratios. The authors also highlight the presence of cross-regional differences in the realization of singletons, with RI speakers showing a tendence to pre-voice voiceless singletons and produce overall shorter C durations and lower C/V ratios (both for voiced and voiceless singletons). Conversely, VI speakers are shown to produce longer C durations and higher C/V ratios for all voiceless singletons causing some overlap between voiceless gemination categories that results in partial degemination through the lengthening of singleton voiceless obstruents.
The article is overall clear and well structured, and the introduction provides sound references to previous and also more recent studies on the specific domain of interest: this gives the reader an overall and informed overview about existing studies on the same topic.
Research questions are clearly stated and hypotheses and expected outcomes are nicely outlined.
The research questions are then addressed building on evidences stemming out from the analyzed data and from the statistical analysis. A detailed discussion of the results is provided and the conclusions summarize the outcomes with reference to the research questions which guided the study.
No major issues are found in the paper.
The authors will find more detailed comments and suggestions on potential issues directly in the manuscript's pdf file. For each highlighted issue I propose suggestions and possible improvements or revisions in order to avoid ambiguity with the aim to improve the paper’s readability and clarity.
A few typos are present and have been marked as well in the pdf file.
The authors are kindly invited to check the quality and resolution of the figures included in the manuscript: it seems that figures have a low resolution which becomes evident when zooming into more detail (if this is just so for the reviewing process then do not consider my comments on this).
There is only one possible infelicity which I highlighted in the manuscript at lines 418-422 with a comment and possible ways to address the issue.
I hope the authors will find the comments and proposed suggestions useful to improve their manuscript.
Comments for author File: 
Comments.pdf
Author Response
Comment: “The paper offers a very detailed contribution addressing the length contrast, which is present and productive in the Italian language, by taking into account the role of consonant voicing and the interplay between vowel and consonant. The focus of the paper is on a selection of Italian voiced and voiceless obstruents in their geminate and singleton realization in two specific varieties of Italian, namely Veneto Italian (VI) and Roman Italian (RI), two varieties that present some phonetic differences on the length contrast: Veneto Italian (VI) - as well as other Northern varieties - is described as maintaining the voicing contrast but in some cases not the length contrast; in Roman Italian (RI) - along with other Central and Southern varieties - the opposite may occur by, for example, pre-voicing intervocalic voiceless singleton obstruents whilst also maintaining the length contrast in the production of obstruents.
The adopted methodology relies on acoustic phonetic analysis of read data collected from speakers of the varieties of regional Italian above mentioned.
The authors acoustically analyse the consonant's prevoicing patterns and the effects and interactions of regional variety, gemination, and (phonological and phonetic) voicing on consonant (C) and preceding vowel (V) durations. Moreover, the analysis focuses on the less used C/V ratio between the two (C/V). Based on outcomes of this study, which are also supported by a careful statistical analysis, the authors show that speakers of both varieties produce geminates with high C/V ratios. The authors also highlight the presence of cross-regional differences in the realization of singletons, with RI speakers showing a tendence to pre-voice voiceless singletons and produce overall shorter C durations and lower C/V ratios (both for voiced and voiceless singletons). Conversely, VI speakers are shown to produce longer C durations and higher C/V ratios for all voiceless singletons causing some overlap between voiceless gemination categories that results in partial degemination through the lengthening of singleton voiceless obstruents.
The article is overall clear and well structured, and the introduction provides sound references to previous and also more recent studies on the specific domain of interest: this gives the reader an overall and informed overview about existing studies on the same topic.
Research questions are clearly stated and hypotheses and expected outcomes are nicely outlined.
The research questions are then addressed building on evidences stemming out from the analyzed data and from the statistical analysis. A detailed discussion of the results is provided and the conclusions summarize the outcomes with reference to the research questions which guided the study.
No major issues are found in the paper.
The authors will find more detailed comments and suggestions on potential issues directly in the manuscript's pdf file. For each highlighted issue I propose suggestions and possible improvements or revisions in order to avoid ambiguity with the aim to improve the paper’s readability and clarity.
A few typos are present and have been marked as well in the pdf file.
The authors are kindly invited to check the quality and resolution of the figures included in the manuscript: it seems that figures have a low resolution which becomes evident when zooming into more detail (if this is just so for the reviewing process then do not consider my comments on this).
There is only one possible infelicity which I highlighted in the manuscript at lines 418-422 with a comment and possible ways to address the issue.
I hope the authors will find the comments and proposed suggestions useful to improve their manuscript.”
RESPONSE:
We would like to thank you for your thorough review of the paper, and for spotting and reporting typos. This is invaluable help, which is much appreciated. We have incorporated your feedback thoroughly as shown in the revised version with tracked changes.
Please note that due to issues with converting the Word file with tracked changes to a PDF, line numbers do not match between the final PDF and the markup PDF. The line numbers referenced in our responses correspond to those in the markup document.
We have replaced figures 1 and 2 with Praat-generated figures, as we could not find a way to get rid of the black background in the EMU webApp. We have also checked the quality of figures and replaced all of them to reflect changes in the names of factors and categories (e.g. “variety” instead of “region”). As for the possible infelicity, we have reworded the highlighted paragraph (lines 675-82 of the PDF showing markups) and clarified the reason why the recordings took place in two different locations. We hope that it is clearer now.
Reviewer 3 Report
Comments and Suggestions for AuthorsVery interesting and thought-provoking paper (with some unexpected findings too).
All comments are in the file.
A couple of general points:
- if I have not missed something, I was wondering: i. was there any registered effect of the position of the word in the sentence (sentence internal v. sentence final)? If not this could be stated explicitly. Are the data analysed all together? ii. was there any speaker effect in the data and results presented? For instance: it is said that in no instance the contrast between /b, dʒ/ v. /bː, dːʒ/ in RI is neutralised. Is this true for the realisations of all speakers? In this instance and in general, it would be interesting to know whether the authors found some significant intraspeaker variation (or some speaker's realisations having more weight than others; e.g. one RI speaker prevoicing more than the others some voiceless singletons and so on) or whether the results presented hold for all the speakers.
- in general, I wonder how much RI prevocing of voiceless Cs (especially singletons) has an effect on other observed patterns. For instance, lines 690-693: it is said that RI singletons show only marginal differences. Is this because voiceless singletons are sometimes prevoiced? Similarly, line 724 and following: "Another striking finding concerns voiceless singletons, which are always significantly longer on average in VI than in RI". If this include the prevoiced voiceless singletons data in RI, it could be somehow expected (RI speakers producing prevoiced, therefore somewhat shorter, voiceless singleton? And again lines 810 and following: a significant difference in C/V ration is observed between RI and VI only for voiceless singletons. If this include RI prevoiced singletons this could be a way to make sense of this data (since Figure 7 seems to show a correlation between prevoicing and C/V ratio). These points go along the lines of what it is said in lines 858-866: the prevoicing tendency of RI speakers determines shorter consonantal realisations.
Comments for author File: Comments.pdf
Author Response
Comment: “Very interesting and thought-provoking paper (with some unexpected findings too).
All comments are in the file.
A couple of general points:
- if I have not missed something, I was wondering: i. was there any registered effect of the position of the word in the sentence (sentence internal v. sentence final)? If not this could be stated explicitly. Are the data analysed all together? ii. was there any speaker effect in the data and results presented? For instance: it is said that in no instance the contrast between /b, dʒ/ v. /bː, dːʒ/ in RI is neutralised. Is this true for the realisations of all speakers? In this instance and in general, it would be interesting to know whether the authors found some significant intraspeaker variation (or some speaker's realisations having more weight than others; e.g. one RI speaker prevoicing more than the others some voiceless singletons and so on) or whether the results presented hold for all the speakers.
- in general, I wonder how much RI prevocing of voiceless Cs (especially singletons) has an effect on other observed patterns. For instance, lines 690-693: it is said that RI singletons show only marginal differences. Is this because voiceless singletons are sometimes prevoiced? Similarly, line 724 and following: "Another striking finding concerns voiceless singletons, which are always significantly longer on average in VI than in RI". If this include the prevoiced voiceless singletons data in RI, it could be somehow expected (RI speakers producing prevoiced, therefore somewhat shorter, voiceless singleton? And again lines 810 and following: a significant difference in C/V ration is observed between RI and VI only for voiceless singletons. If this include RI prevoiced singletons this could be a way to make sense of this data (since Figure 7 seems to show a correlation between prevoicing and C/V ratio). These points go along the lines of what it is said in lines 858-866: the prevoicing tendency of RI speakers determines shorter consonantal realisations.”
---
RESPONSE:
Thank you very much for the insightful feedback. We have modified the paper accordingly (see the markup PDF for tracked changes) and believe that your comments helped improve the quality of the paper significantly.
Please note that due to issues with converting the Word file with tracked changes to a PDF, line numbers do not match between the final PDF and the markup PDF. The line numbers referenced in our responses correspond to those in the markup document.
In response to your questions above:
i) We have added the results for position in the phrase showing that the effect has high statistical significance (as it is very consistent), but only a marginal size of a few milliseconds (lines 1132-7 and 1178-81 of the PDF showing markups).
ii) Yes, the data were analysed together. However, thanks to your comments, we have added specific information on inter- and intra-speaker variability in (a) the production of the /b, dʒ/ v. /bː, dːʒ/ contrast, specifying that there are significant differences for all speakers (lines 1146-50 and 1563), and (b) on patterns of duration across lenited and unlenited voiceless singleton tokens in RI and VI, showing that VI voiceless singletons are considerably longer than the same consonants in RI, even when these are canonically produced (i.e., unlenited) in RI (lines 1162-9 and also 1702-15 in the discussion). We believe that this is a crucial point we did not clearly make in the previous version, so thank you for pointing this out.
iii) We have re-structured the results section following the guest editor’s suggestions and discarded bits like “lines 690-693: it is said that RI singletons show only marginal differences”, which were in fact confusing (these lines specifically were referring to the fact that the voiceless-voiced C duration difference for singletons was reduced in RI relative to VI, partly because /bː dːʒ/ raised the overall average for voiced singletons in RI, but also because RI tended to pre-voice voiceless singletons, and thus produce smaller durational differences across voiceless and voiced singletons). In general, answers to your questions on the role of prevoicing in the observed patters are (hopefully) more clearly given in the revised version (cf. lines 1129-31, and again lines 1162-9, 1702-15 covered in point ii)-(b) above).
