An Acoustic–Phonetic Description of Hidatsa Vowels
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsReview report
I reviewed the manuscript “An acoustic-phonetic description of Hidatsa Vowels” which presents an acoustic investigation of vowels produced by three female speakers of Hidatsa, an endangered language spoken in the Forth Berthold Indian Reservation in North Dakota. The vowels were examined according to temporal (duration) and spectral (F1, F2, F3) properties. The duration measurements suggest that Hidatsa has a 12-vowel system, including 10 monophthongs differing in length (short vs. long) and two diphthongs. In addition, the authors found that the short and long monophthongs exhibit clear quality-differences in terms of formant frequencies.
Overall, I think that the paper makes a solid contribution to phonetically describing an endangered language that has only very few speakers left. I do, however, have some concerns with regard to the very limited number of speakers who participated in the study and accordingly, the few observations the acoustic analyses are based on. While it seems that the study is based on a solid experimental procedure, the authors need to include more detailed information as to the collection of test tokens and the overall procedure (see detailed comments below).
I have the following suggestions/corrections, including some requests for clarification:
General comments
Your introduction is very short. For someone who is not familiar with Hidatsa, it would be interesting to learn a bit more about this language and its speakers (e.g., Are all of them L1 speakers, what are their second languages – do they have any at all, what do we know about the influence of other languages on Hidatsa, what is the history of this linguistic community, is there perhaps research on linguistic features other than vowels available, etc.). Perhaps also briefly state what your incentive was to have a closer look at this particular language. You do not have to discuss these things at length, but addressing some of these aspects would help to more appropriately contextualize your study.
What worries me is the extremely low number of participants included in your study. Even if it is considered a small-scale study presenting preliminary results, as you state, I am really not sure whether three speakers are enough to come up with anything meaningful in terms of an acoustic analysis (including statistics – depending a bit on the number of observations you have, of course). I can imagine that it is particularly difficult to find speakers of a minority language willing to participate in a study, but in order to obtain useful and meaningful data, I doubt that three speakers are enough – particularly considering that you want “to determine the exact phonetic inventory of vowels in Hidatsa” (line 32, my emphasis) which I think may be rather difficult if you have data of only three speakers.
Therefore, my question is whether it would be possible to include more speakers? From what you write in the manuscript, I understand that you have access to recordings of nine speakers in total – which is still not a high number of participants, but it would provide more data to work with and compare to previous studies.
Overall, I am not entirely sure what the ultimate aim of this study is. I understand that there is no real consent on the phonetic qualities of Hidatsa vowels (see studies cited by the authors), but I am wondering whether a study with three speakers only can add any clarification to this issue (also given that your study includes female speakers only…). It would be good to know how many participants previous studies on Hidatsa vowels included, which methodology they used (elicitation of test tokens etc.) and whether your study is really comparable.
Regardless of the low number of participants, what I miss in the paper is a more in-depth comparison with previous studies – perhaps even in a graph (vowel plot) comparing your formant and duration measurements with the measurements obtained in previous studies. I think that this is particularly important given that there are, if I understand this correctly, only very few studies that have examined Hidatsa vowels so far. And – as mentioned above – it would be helpful to provide more details on the number of participants included in the vowel-studies you cite (in addition to some more details on the methodology and findings).
What I also suggest is to discuss the low number of participants as a potential limitation of your study in more detail – this would show that you are aware of this limiting factor and that the results need to be interpreted with caution.
Detailed comments
Introduction
Line 22: “describe” should be “describes”
Lines 28-29: You say that that it is not clear “what exactly he [Bowers, 1996] means by this terminology”. Which “terminology” do you refer to here? The fortis-lenis distinction?
Line 40f. “Most of the acoustic analysis with regards to procedure follows that done by Elvin et al. (2019)”: I would describe this in a bit more detail (perhaps somewhere in the Methodology-section because this is actually where it belongs), that is, which ‘procedure’ did Elvin et al. (2019) exactly apply?
Line 49: “recording” should be “recordings”
Footnote 1: You say that Crow and Hidatsa are not mutually intelligible (which may suggest that they are different languages?), but “closely related”. Could you elaborate a bit more on what you mean by “closely related” here?
Methods: Participants
Lines 46-47: Do you mention the participants’ full names here? I am not so sure whether you should do this considering data protection issues. Also, I don’t think it is necessary to mention their full names, is it? Or do you have a specific reason to do so, that is, is it important for your study?
Line 50: Here, it says “due to the number of tokens we could find from different individual’s recordings”. Perhaps it is just me, but it is not entirely clear (to me) what this means. Could you perhaps rephrase this to make it clearer?
Line 52-53: You say that your participants were “approximately 45 and 65 years old”. Does this mean that you do not have their exact ages? This also makes me wonder what exactly the dictionary database project you mention in line 45 looks like, particularly in terms of participant selection and data collection. Does this, perhaps, explain why you do not refer to the participants’ exact ages? You do briefly refer to this project in lines 56ff., but I think that adding a bit more information here would be helpful (also in terms of how the production data were collected, etc.).
Line 53-54: Here, you state that “there are fewer than 40 fluent people who speak Hidatsa as an L1” – this is, however, not what you say in line 21 (“approximately 30 fluent speakers”). So, how many are there? Also, it would be interesting to know where you got this number from.
Methods: Vowel Tokens
Would it perhaps be possible to include some of the words (or perhaps all of them?) in the paper, including an English translation? As someone who is not familiar with Hidatsa, it is very hard to imagine what these target words looked like – although you describe them at least with regard to the vowels included. You could easily display some examples in a table or in the appendix.
Also, I can’t find sufficient information (in the methods section) as to how many words/vowels you recorded exactly. You say that you “selected at least 20 words for each vowel” – per speaker or in total? This could also be easily displayed in a table.
Methods: Stress Marking in Hidatsa
Line 73, “… have shown Park’s claims to be erroneous”: This sounds a bit odd. Perhaps consider rephrasing.
Results
Your results section actually only consists of a very brief – sometimes rather vague – description of what is shown in the figures. Did you also conduct statistical analyses (e.g., to see if the duration differences are significant or not?) – or was this not possible due to the limited number of observations?
Line 83, “nearby vowels” > “adjacent” ?
Lines 89-90, “we relied on the general observation that nasal segments of speech have a lower amplitude than vowels …” Do you have a reference for this?
Line 91: “this lower overall intensity” > “this overall lower intensity”?
Normalization
In line 51f. you state that you discuss “the normalization constraints” in section 2.5. However, in this section, I can’t find any discussion of normalization constraints. What exactly do you mean by this? The way you phrase this makes it sound as if there were constraints with regard to normalizing your data. Please explain.
Results
Lines 127ff., “We believe this offers conclusive evidence that Park’s (2012) and Bower’s (1996) claims of a reduced vowel inventory are incorrect”: First of all, I think this should be part of the discussion-section. Secondly, how did Park and Bower arrive at the observation that Hidatsa has a “reduced vowel inventory” – did they also conduct acoustic measurements? If so, I would not refer to this as “claims”, but rather as “observations” – a claim is something for which you lack evidence. In addition – as already stated above – it would be really helpful to know how many speakers Parks and Bowers included in their studies and how they collected their data.
Line 137, “indicated by the letter for the vowel”: Which letter? Do you mean IPA symbols?
Line 143, “These appear to move”: Move where?
Line 145, “proto-typical vowel space”: You mention this “prototypical” vowel space at several places in your paper. I think it would make sense to very briefly (perhaps in a footnote) explain what exactly you mean by ‘prototypical’ in this context.
Line 167f., “This is a gap in the vowel inventory”: Meaning what? That this vowel does, in fact, not exist at all or that you simply did not have it represented in your test tokens? You clarify this in your discussion (see line 227), but I would still briefly mention this here as well, that is, that unstressed [e] may, in fact, exist, but that it did not occur in the items you examined.
Lines 169-174: This is just a description of what Figures 4a and 4b display – but it doesn’t really give any information of what exactly happens with regard to vowel movements.
Discussion
Line 203-204, “counter to both Park’s (2012) and Bower’s (1996) claims regarding gaps in the five-vowel system”: Again, I would be careful with using the word “claim”. As I mentioned above, making a claim basically means that you say something without providing evidence or proof. And although your findings are apparently different from the findings provided by Park and Bower, I assume that they still conducted some kind of proper analysis, right? In addition to this, it may be good – just for clarity – to briefly mention again which “gaps” in the vowel system Park and Bower identified, respectively.
Line 208-209, “how they have been described in the previous literature” > this requires a reference (which literature do you refer to?)
Line 211: Perhaps it’s just me, but I don’t quite get what you mean by “when stressed the F1 and F2 distribution switch in the range of their values”.
Lines 207-223: What I miss here is a more precise comparison with previous studies, that is, to what extent is what you describe here reflected in previous studies? Can potential differences in the results be explained by differences in methodology, perhaps? That is, is it possible at all to compare your findings with other studies – considering that there may have been differences with regard to the methodology used etc.?
Line 217: What exactly do you mean when you say that the [u:/ú:] vowel is “intriguing”?
Conclusion
Line 277f., “indicates that it is not like previous descriptions” – references needed.
Line 291, “we can use the findings presented here as targets for vowel production by L2 learners”: So, there are people who learn Hidatsa as an L2? It may be good to give some more information on this, that is, the linguistic situation and speakers of Hidatsa (see my comment above).
Line 296, “this is the largest and most thorough phonetic description of vowels in a Siouan language”: So, your study is really the ‘largest’ study on H. vowels that has been conducted so far? With only three participants? I would, by the way, avoid saying “most thorough description” as this implies that previous studies were rather superficial.
Comments on the Quality of English LanguageSome minor revisions; see comments in my review report.
Author Response
Please see attachment. We downloaded the reviewer's comments and added our comments regarding responses in those comments.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsPlease see attached pdf file.
Comments for author File: Comments.pdf
It's a well-written paper, some minor typos need to be corrected.
Author Response
Please see attachment. We added comments in the pdf as we worked through the reviewer's comments.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsOverall:
This is a small but important piece of documentation work that gives a thorough account of the elicitation protocol and analytical decisions, contextualized in previous work on Hidatsa and related languages. I would recommend a couple of minor changes to the paper, but generally feel that it is ready for publication.
General comments:
1) There are some inconsistencies in the use of phonemic vs. phonetic transcription, specifically in the abstract and beginning sections of the paper. For example, the abstract gives phonetic transcriptions in square brackets [ ] for the vowels, but transcribes the long/short pairs with the same vowel quality, like [a/aː]. However, this is not what the paper concluded: section 4 describes differences in vowel quality between these pairs, so the abstract should either list these as phonological (/a/ /aː/) or phonetic with these updated vowel qualities ([ə] [aː]).
2A) While the paper goes into great detail about methodology, there are very few examples of words in Hidatsa given. Since part of this paper's contribution is that it addresses existing controversy over the number of vowels in Hidatsa, it would help to list an example word containing each of the vowels tested, especially for short [e] [o] which are not universally agreed upon in the existing literature.
2B) Since the paper has reanalyzed the vowel inventory of Hidatsa, it would help to end the discussion section with an updated inventory of vowels with their stressed/unstressed allophones as well as the changes to vowel quality introduced in that section.
Line-by-line comments:
144 - "stress makes all of the vowels occupy a smaller, tighter, more proto-typical vowel space with the exception of the stressed long [úː]" - In Figure 2, it seems like [úː] is the vowel which has shrunk the most in F1/F2 compared to its unstressed counterpart [uː]. What makes it an exception?
226 - date -> data
Author Response
Please see attached comments. We comments on each of the reviewer's comments next to them in the word doc. Thank you.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Authors,
After reading the revised version of the manuscript entitled “An acoustic-phonetic description of Hidatsa vowels”, I can say that you have carefully addressed my comments and issues raised in my first review.
First of all, I think that the introduction has improved a lot and is much clearer now as it has become more informative, particularly for readers who are not familiar with Hidatsa in general and Hidatsa phonology in particular. In addition, you now very clearly state that your study is, in fact, the first to provide a description of Hidatsa vowels based on phonetic/acoustic measurements – which makes your study a valuable contribution to the linguistic description of this language.
The overall quality of your paper has also improved because you now address potential limitations of your study in a much clearer way in a separate section and briefly outline the need for further research on Hidatsa and other Siouan languages.
Overall, you added much valuable information that does not only make understanding the motivation and goal of your study easier, but it also gives the reader a much better impression about Hidatsa and its speakers (e.g., additional information about their linguistic background, explanation why you did not state your participants’ exact ages, etc.).
I only have some minor comments to address:
Minor comments
Line 24, “is a greatly reduces portion of ...” → “is a greatly reduced portion of…”
Line 37, “it’s interaction with syntax” → “its interaction with syntax”
Line 58, “In the following section, we discuss where in our methodology, we followed some of these procedures in detail.” → the second comma (after methodology) needs to be deleted.
Line 81, “we tried to selected” → “we tried to select”
Footnote 5, “It’s considered” → “It is considered”
Footnote 11, “We wanted to restrict to context where the monophthong vowel boundaries could be clearly identified.” → I would slightly rephrase this: We wanted to restrict our selection to contexts where the vowel boundaries of the monophthong could be clearly identified.
In fact, after reading footnote 11, I am wondering if this should perhaps go in the main text instead of having it in a footnote? I mean, it is basically part of how you chose the vowel tokens for your analysis.
Footnote 3 (p. 2): I think it is good that you mention here that your informants gave you full permission to state their real names in the paper and that you would like to give them credit in your article, which I fully support.
I just want to briefly comment on what you say in the last sentence of this footnote, because it gives the impression that you may have misinterpreted my comment on this in my first review: “It is an unfortunate consequence of colonialist and paternalistic racist attitudes that, in the past, many North American researchers in the fields of linguistics and anthropology failed to give credit to the many people who made their work possible.”
Although I fully agree that it is important to give credit to the people who make linguistic (and other!) research possible, very often (most of the time?), we typically do not give this credit by mentioning full names – simply because of data protection issues and because of the fact that participant names do not have any informative values in a study (again, there are exceptions to this). Conducting empirical research myself and collecting pronunciation data from speakers, I would never (unless really necessary) state my respondents’ full names because I want to protect their privacy. I do understand, however, that this is different in a study like yours, where it is particularly difficult to find participants and where there are only very few speakers left. So, to make a long story short: In my first comment on mentioning full names, I was only concerned about data protection issues since handling participant information in an ethical and confidential way is as important as giving credit to people who participate in our studies.
Footnote 4: Interesting! I really like that you provide the link to the online dictionary.
*****
Other than that, I do not have any comments. As mentioned above, I think that the overall quality of the paper has considerably improved. The authors present a very revealing study that has set itself the goal of linguistically describing a language that has very few speakers left.
Comments on the Quality of English LanguageMinor revisions (see my comments).
Author Response
Notes are added in the attached word doc.
Author Response File: Author Response.pdf
