I read “Academic publishing requires linguistically inclusive policies” with great interest. I find the article to be important and timely with good, concrete recommendations for journals. I found the questions scored from websites and sent to editors to be generally quite comprehensive. I believe this article will be an important contribution, and I hope the article leads to reforms in publishing towards greater linguistic inclusivity.

I do, however, have some big pic6ture concerns and some smaller suggestions with the present form of the manuscript, detailed below.

First, for the “website” part of the study, my understanding is that the author team collected data on linguistic inclusivity only from the author guidelines sections of websites. I have personally read the websites of over 500 journals to look at policies, and that experience showed me how different the organisation can be between websites, esp. by publisher. In my experience, the largest, for-profit publishers like Springer and Elsevier are typically the most streamlined and “boilerplate” whereas journals published by small publishers can often have fairly disorganized websites. The large publishers will often have all information in one or two spots, but smaller publishers might have information spread out across multiple pages and under tabs you wouldn’t expect to find language policies. I have found fairly non-inclusive English language policies for some journals in places you wouldn’t necessarily expect like under publishing charges pages or “about the journal” type pages. Similarly, I’d like some clarity in where the authors were looking for policies for reviewers and editors. The way the manuscript is written, I believe the authors only looked at the author guidelines sections for statements for reviewers/editors, but often I would find those under reviewer/editor guidelines or sections labelled something like “review ethics” or “publishing policies and ethics”. All in all, I am not suggesting that the authors go back and look at every page on the journals’ websites, but I do think the data collection protocol could bias some of the results towards the larger publishers appearing less inclusive because their websites are more likely to have a lot of information streamlined into one spot (the author guidelines), which I believe is where the authors looked. I think the authors could add a limitations section around how there could be other policies in spots of the websites they didn’t look.

Second, because a lot of the results focus around differences in editor-reported policies and those on the journal websites, I think it would be really helpful to know how the website-gleaned policies compare to the editor survey dataset for only the journals with paired editor surveys. I think the authors should do a second set of analyses comparing the subset of journals with both website data and editor survey data. That would help show if the differences are due to systematic differences in who is likely to respond to the survey vs. if it has to do with something like differences in perception, website transparency, or website comprehensiveness. For example, in the SI Results, the authors show that editors of society journals were more likely to respond than non-society journals. I wonder if that is because those people already care about linguistic inclusivity, whereas people who don’t care may just ignore the survey. Comparing results for just the responding set of journals should show if you have a biased set of responses. This new comparison could go in the SI only unless the results are quite different.

Third: Manuscript organisation: The manuscript is in a format and length appropriate for PNAS or Nature Ecology & Evolution. I did a brief skim of the PROC B author guidelines and a few recent papers and saw that there is a 10,000-word limit and that articles can be a traditional format of Methods before Results. I think it would be really helpful for the article if the SI text “Extended Materials and Methods” (excluding the listed survey questions) and “Extended Results” (excluding the listed “Boilerplate” recommendations) could be moved into the main text. I had listed a lot of questions in the main text for my review only to find answers in the SI. I believe the authors should have room to do this with the longer PROC B word limit. If some materials can’t fit, I think the authors should at least better reference the SI text to make it clear to look there for certain things. I find the SI Methods to be crucial to understanding the study, and I think the SI Results are also pretty important. For example, the main text Materials and Methods in their present form don’t mention the editor surveys.

Minor comments:

Starting at L68 and throughout: “Linguistic diversity of editorial boards” makes me think you are referring to something like language richness or evenness rather than the percent of editorial boards with affiliations in predominately English-speaking countries. I think it would be clearer to use a term specifically referencing the English-speaking/non-English speaking dichotomy. I realize this could be a clunky way to do it, but using “proportion English speaking editors” or something to that extent would be more straightforward (although if you used the latter you’d still need to caveat it’s based on affiliation).

Starting at L69 and throughout: “Open Access model” seems like you are referring to a journal being fully open access versus not, rather than the percentage of articles that are published open access. I think it would be clearer to use something like “percentage open access articles” or “proportion open access” throughout.

L76-77: I think it might be more to the point of the paper to replace “Journals play an irreplaceable role in addressing this issue” with something mentioning that journals could either help or hinder the issue.

L105-127: This section seems somewhat out of place. It seems like concluding recommendations to me as written. I think the section needs an introduction to say that these were what you were assessing. I also think it might flow best if you moved it to start around L132 such that you’d introduce the journals you surveyed, the specific linguistically inclusive policies you looked at, then the mediators of inclusive policies you looked at.

L115-118 and throughout: I am slightly confused on what the “non-English-language references” item means. In some spots it seems like you mean the actual manuscripts being cited are not in English but in other spots it seems like this item refers to the citation of the study being in English. In my own reading of journal policies, it seems like I’d often find policies about the reference needing to be in English rather than the original study being cited needing to be in English. The spot that sounded contradictory that I remember for sure was in “Extended Materials and Methods Data collection”. Please clarify what this item means throughout. Also on this item, I personally would not have created a category that was “prohibited/no mention” because I think explicitly prohibiting them is different than not mentioning. I’d assume not mentioning means it’s fine and the journal thought they didn’t need to state it. I am not suggesting the authors go back to the journal websites to change it, but I do think separating no mention and prohibition would have been better.

L148-151: This sentence makes me want some more detail on what the authors considered to count. They mention things like mentoring schemes, but I think adding more detail on exactly what did and didn’t count would help contextualize this. In particular, the results in the Extended Results section “English editing services” makes me think that editors interpret what counts pretty differently than the author team, so I think making it clear in methods what cutoffs/criteria you used would be useful.

L151-153: Is this from the editor survey or website search?

L158-208: This section refers to a lot of results but without referencing the tables where the test statistics could be found nor adding the statistics at the end of sentences for support. I think at minimum, the tables should be referenced. Adding the actual statistics at the end of statements could also be good if there’s room.

L164-166: To me, the previously stated reasons both point to higher Impact Factor journals being less linguistically inclusive

L221-235: The authors have nicely laid out recommendations in the SI section, “Boilerplate linguistically inclusive policies”, but I don’t think that section is ever directly referenced in the main text. It seems like this would be a good place to reference it.

L225: You might consider adding “limited or perceived limited English proficiency”

L248-249: How did you decide to pick these disciplines for the biological sciences?

L255-256: Here you say you searched for “whether the journal was owned by a scientific society” and in Table S1 you say you searched for “Whether the journal is published on behalf of a scientific society”. There are some journals that are very clearly owned by a society like the British Ecological Society journals Journal of Applied Ecology and Functional Ecology but then it can get more gray for some like MDPI journals that have “society collaborators” or “affiliated societies” (e.g., Animals: https://www.mdpi.com/journal/animals/societies). Please clarify what the cutoff was and standardize the descriptions between the Methods and SI table.

L255-258: I wondered in particular for the self-published vs. for-profit/non-profit organizations what your sample size was. It might be helpful to add n= for your different categories.

L271-272: I think the survey protocol should be referenced here. If you merge the SI Methods into the main text like I suggested, this current suggestion may not be applicable.

SI Data: The journal ID and data collector are anonymized on the datasets, and the actual publishers are not disclosed. I would guess that the journal ID anonymisation may be part of your Institutional Human Research Ethics Approval for the editor survey, but I think adding journal ID, data collector, and publisher for the website piece would be more transparent/reproducible if it is prohibited by your Institutional Human Research Ethics Approval.

Figure 1: I am impressed by how much information the authors were able to get into one figure! This is no easy feat.

Figure 2: I think this figure is nice in that it shows that more society journals publish content in languages other than English. I think it might be more informative to show the proportion of journals than the total, though. Perhaps this could be shown in a second panel. It’s not clear what the total possible journals are in each category, and additionally, there are more journals over time, so it’s not clear how much is increase in uptake of inclusive policies vs. there just being more journals. I realize it might be hard to figure out the total journals since you didn’t ask the editors what year their journal started being a journal. I know JCR has first electronic JCR year, but I’m not sure if they show first journal year. If the authors can’t feasibly get total journals per year, perhaps a second panel could show the proportion of society vs. non-society journals that currently publish content in languages other than English to contextualise the ratios.

SI Materials:

Extended Materials and Methods section “Data Analysis”: “Least inclusive answers were coded as 0 and most inclusive answers were coded as 1.” I believe this is referring to binomial question, so I think “Less inclusive answers were coded as 0 and more inclusive answers were coded as 1” would be better. Two lines below that, “function clmm” – should this be “glmm”?

“Extended Results” section “Language of guidelines”: “We noticed that the translated text tended to correspond to sections where commercial English editing services were advertised.” In my own experience going through journal policies, I noticed that using my browser to translate a website in e.g., Spanish vs. reading the website’s own translation that the non-English version also often had less inclusive policies. Perhaps add into your protocol description if you only scored the English versions or if you scored all versions.

Extended Materials and Methods section “Identification of predictors”: I think this section would be more appropriate under the methods with the “Data analysis” section.

SI section “Boilerplate linguistically inclusive policies”: First of all, I appreciate how clear the recommendations are for journals. Second, I think the header could be more straightforward for what the section is for. I would suggest going with something like “Recommendations for linguistically inclusive policies”, “Recommended linguistically inclusive policies”, or “Recommended boilerplate linguistically inclusive policies for journals to use”. You could add into the first section below that that you suggest they could take your statements as “boilerplate language” in their policies.

Table S1. I like how JCR is clearly indicated next to some variables. Could you also specify “website” as applicable such that all variables have clear labels on where they are from?

Tables S2-S4: I think it would be easier to interpret the statistics if you didn't use the "e-" formatting because it takes more effort to see what is significant. If the authors have a strong preference for the “e-“ format because they can use precise numbers like "< 2.20e-16" rather than "<0.0001", perhaps they could add a star or bolding for significant differences to make it really obvious what is significant.

These tables also use “Language of country”, but I think it is “English/not English”, so I think being specific on that would be clearer.

All in all, I find the article to be a very important contribution, and I hope it will help push journals towards greater inclusivity.

I re-read the manuscript “Academic publishing requires linguistically inclusive policies” with interest. I appreciate the work the authors put into their revisions and the study more broadly. I have some additional minor comments for clarity. I will reiterate from my initial review that I hope the manuscript is able to help push publishers and journals towards linguistic inclusivity and believe the work is very important.

Regarding line numbers in my review: In the portal, I created a proof and downloaded the .pdf. The .pdf proof it generated didn’t show the track changes. My line numbers correspond to this .pdf proof – I hope the authors are able to match the line numbers easily.

L79: I had to search “hegemonically” to confirm I knew what it meant. It seems like the word has some qualities that aren’t captured by possible synonyms, but something like “predominately” would probably be more widely known by readers.

L163: It took me a second to understand "HAC". I would suggest revising to something like, "One author (HAC) compiled..."

L166: I am not sure what the PROC B guidelines are on how to reference Supplementary Material, but if it is in accord with their policies, it seems like using more informative references would help guide the reader to the right spot. The way things are referenced, it is often not clear what you’re going to the SI to find. I thought the reference to SI at L181 was clearer than at L166. The SI reference at L203-204 was also clearer than at L166, but the reference at L203-204 was perhaps less clear than L181 because at L203-204 it seems like you’re going there for the ordinal regression details only.

L159-178: The first paragraph seems to be about the author guidelines but then the next paragraph starts with an "Author guidelines" subheading. I would suggest moving up the subheading and carefully screening the flow of information in the two sections.

L216-218: The new analyses comparing only the journals for which you had both website and editor survey data are buried currently. I think either mentioning the new analyses at the end of the paragraph at L216 or after the sentence at L217-218 could work to better introduce them. At L216 you could say something like, "We ran our analyses twice to confirm the subset of journals we received Editor-in-Chief surveys from was unbiased. First, we ran analyses using all data, then we reran analyses using only the journal website data for which we had paired Editor-in-Chief survey data. Results were qualitatively similar (Figure S3, Table SX)." (See my note below about adding a table with your results for this).

L233-235: I was wondering here if it was the BioOne website that implemented the machine translation tools or if you meant that the journal's main website implemented the tool AND those were also archived on BioOne. I went to the Supplementary Material to check, and I didn't see mention of BioOne. I’ll reiterate here that I think the references to Supplementary Material should be more specific as to what you're looking for. Since the reference here seems to correspond to checking out the detailed Extended Results, perhaps it would work better to reference the SI at L218. For example, you could say "...barriers in publishing (Figure 1; see Supplementary Materials Extended Results for detailed findings)." (please modify as appropriate to meet the journal's SI referencing guidelines)

L259-268: I like this new text, but I think it might be better off as a new paragraph. The English language editing services were discussed before the citation of non-English language literature, so it seems like this doesn't completely follow the text about non-English language literature citations right above it. I think you could probably just start a new paragraph and remove "Furthermore". Alternatively, you could move it after your discussion of the English editing services from L228-230.

L270-272: Figure S3 seems like an odd reference here, and I'm not sure if it's relevant to this sentence. As I mentioned before, I think Figure S3 needs to be made more obvious earlier in the manuscript as to what it is and why the separate analysis is of interest. Also, when reading "geographic origin", I was thinking you had a map of some sort showing where high impact factor journals were located. I wonder if saying something like, "with different geographic scopes and whether English is the primary language of the country they are based in" would be clearer. (The authors probably could revise my suggestion to be even clearer or clarify it if I am mistaken.)

L270-278: I still am struggling with the conclusion that the factors given could lead to IF being positively or negatively associated with the adoption of linguistically inclusive policies. To me, I'd assume lower impact would be more inclusive based on the factors given. I reread the section several times and am still not following the logic flow. I could envision other reasons you could expect either positive or negative relationships. For example, I could see more inclusivity at the top because they could get more pressure for inclusivity than a smaller, less known/less read journal. I could also see the high IF ones with high page charges having more money for providing free editing, etc. On the other side, I could see less inclusivity at high IF journals because, as the authors mentioned in the response to reviewers, these journals’ publishers typically have their own commercial services they're advertising. Because the text that follows L278 focuses around high IF journals, maybe mentioning some potential drivers at the "high end" would help tie things together. In the response to reviewers where the reviewers mentioned confusion here, you focused on the drivers at the high end, so I think revising this a bit along the lines of the response to reviewers would help. All in all, I'm not tracking the text at L276-278 following from L272-276, and I think this section could still use some revising.

Starting at L269, "Predictors of linguistic inclusivity section": it's not always clear in here what results correspond to the author guidelines vs. editor survey when data were gathered for both. For example, L278-280, you got data from the websites and editors for English-editing services, but only the website value had P <0.05. I think adding some short modifiers to indicate what dataset the results correspond to would help. For example, at L278-280, you could say, "...were more likely to refer authors to commercial English-editing services in their author guidelines than..." I think this might break up the prose a bit and could make the section long, so it might not always be possible to clearly delineate them. I’d suggest delineating them in cases that it doesn’t add a ton of words.

L299-311: I wondered if a driver of this might be a correlation between the editors’ affiliations’ countries’ languages and the languages of the country the journals were based in. I would assume journals would be more likely to have editors also based in the same country they are based in. I see that Table S3 does show a correlation between the two, although, the information in the table doesn't tell me for sure that is the case (see my comment below about possibly adding means into the table). In the main text, it might be good to mention a correlation with other factors like the country they are based in, but all the factors examined are very intertwined, so I suppose there could be many drivers.

L307-309: I had to read this a few times. I think "in economics" is a bit vague as to whether it means the discipline, economics journals, economic drivers, etc. Perhaps reordering the information in the sentence would help clarify.

L357: If compliant with PROC B's guidelines, I think also referencing the specific subsection in the SI these can be found would help guide people to the right spot

Figure 2: I think it would help to have the total number of society vs. nonsociety journals here for A and B. For A, you could break down society vs. non-society in L434, e.g., "(N = 77, society N = XX, nonsociety N = XX)". For B, maybe you could embed N = within the x-axis labels and/or add that after the "N =" at L437 like how I suggested for A.

SI Extended Materials and Methods subsection “Identification of predictors”: In the second paragraph, I think there are results from the second sentence up the last sentence where you mention where data are archived. These sentences might fit better at the end of the next section, “Extended Results”

Figure S1: The old terminology is still used in this figure

Figure S3. I think this figure needs to indicate significance like the main text figure and also have corresponding statistics in a table like Table S4. The second sentence is also a little unclear. You might rephrase as something like, “These plots show only journals for which we had both editor-in-chief surveys and journal website data (N=262).”
Table S3. Because this table only shows the Chi-square values and p-values, the reader knows there are differences, but you’re left guessing which category has higher values. I think it would help to also have the means +/- SEs for each category in the table. For example, for each cell, you could do something like:

X21 = 34.76
P-value < 0.0001
English 5 +/- 1 (SE)
Non-English 2 +/- 1.2 (SE)

Where this example would be for IF and language of country.

I went to this table originally to see if the language of the country a journal is based in and the proportion of editors based in non-English speaking countries were correlated. That’s when I realized I couldn’t tell for sure if English/English and non-English/non-English are most common (which I’d assume). The same comment about adding means and SEs could apply to Table S2, but I think doing it for discipline could get messy since you have a lot of categories.
Table S4. I think it would be helpful to add into the row labels what the reference category is. For the society owned column header, I think it would help to say if society owned yes or no is the reference. I can infer the reference categories based on the results, but I think it would be clearer to add them into the table itself.

I’d like to conclude by iterating again that I think this is a very important paper, and I look forward to its publication.