Case, Concord and the Emergence of Default

Round 1

Reviewer 1 Report

Please see attached.

Comments for author File: Comments.pdf

Author Response

Reviewer 1.

 

Reviewer 1 raises two main issues. The first one is that I introduce the functional head K without some key references. To address this concern, I have introduced a new reference to Bittner and Hale (1992) as well as the general reference work Alexiadou et al (2007). Unfortunately, I have been unable to locate the reference suggested by the reviewer, Labelle (1992), which is not listed in her web page or ResearchGate.

 

The second issue is my lack of integration of recent literature, mostly within the Norwegian American project. The literature that R1 alludes to seeks to provide evidence for the Distributed Morphology proposal that roots and categorization should be conceptualized in a syntactic way, as separate terminals. (This separation is assumed in my submission.) My goal is entirely different: an exploration of concord and default exponents, which no work on code-switching has focused on, to my knowledge. It seems to me that an extensive discussion of the references mentioned by the reviewer would take this overly long paper far afield. Thus, I have been unable to comply with the reviewer’s request. However, I have acknowledged the Norwegian American work in lines 106-108 and I have deleted my claim that code-switching is “understudied in linguistic theorizing”, which I now realize is somewhat controversial.

 

I would like to thank R1 for their care in pointing out some of my typos and grammatical errors in my writing. I have done my best to correct them.

Reviewer 2 Report

See attachment

Comments for author File: Comments.pdf

Author Response

Reviewer 2.

 

I would like to thank R2 for their extremely insightful and detailed review.

 

R2 divides their objections in four categories:

(i) “the author’s proposal regarding the valuation of the various heads internal to the KP-nominal …”

I have redone this part of the paper and made sure that operations take place in a coherent order (see line 973 and ff). This has had a positive effect on other sections of the paper and in the problems pointed out in (iii) and (iv).

(ii) “the author’s take on feature compatibility as a restriction on when Fusion may occur (p.23; see comments below);”

I have defined feature compatibility more precisely (lines 1115-1120).

(iii)  “whether the author is in fact employing the Subset Principle consistently (see p.23, (51), and then p. 38, (78-9), for example) […] I’m not quite sure how this figure fits in: a valued counterpart to an unvalued feature isn’t in a subset relation to the unvalued feature.”

I have rewritten the relevant paragraph (lines 1227-1256) hoping to improve clarity. The idea is not that unvalued features are a subset of valued features or vice versa. The idea is that a terminal may be specified only for the feature type – say, [gender] – which does stand in a subset relationship with any terminal that has [gender], including an unvalued gender.

(iv) “the extent to which the author is able to offer a suitable formal distinction between invariant and variable-gender nouns in Spanish (p.33);”

I have rewritten the relevant paragraphs (lines 1235-1285 and 1312-1343). I believe now it is much clearer.

(v) “how the traditional assumption about the dependency between weak and strong inflection is formally implemented in the present approach (p.38).”

Here I have chosen to leave it as is, since it is a complex matter and a long discussion would be distracting. Weak inflection in monolingual German appears when strong inflection is in another constituent – however, this is not important for this submission. What is important is that weak inflection, [e] is default, in as much as it is a collection of unvalued features. However, I have rewritten some paragraphs for clarity.

I would also like to thank the reviewer for the time and effort spent on spotting many typos and grammatical errors in my writing.

Reviewer 3 Report

The goal of this theoretically oriented paper is to propose an analyses of concord and default morphology within the noun phrase. To that aim evidence from three code-switching varieties is presented: Basque/Spanish, German/Turkish and Russian/Kazakh. The author proposes that concord is driven by the head K, which selects the DP and can spell out as case morphology. According to the author, in the absence of the proper form of K, there is no concord, and consequently the noun phrase appears with default morphology. Additionally, the author states that the data presented argues in favour of the Concord-as-Agree hypothesis.

 

The topic of the article is interesting both for theoretical linguistics and applied linguistics. However, there is a problem with the empirical data presented as evidence for the theoretical claims. Despite reporting experimental data, there is no section dedicated to methodology and there is almost no information on the method for data collection (participants, procedure, materials, etc.). According to the author “the method employed in this paper consists of constructing examples in which the case morphology comes from a language without concord while the rest of the noun phrase comes from a language with concord” (p. 3, lines 81-83). Later in section 2.1. when discussing Spanish/Basque code-switching data, the author says that data reported were “extracted in consultation with two speakers from Gernika” and that the data were not obtained via regular grammaticality judgments; rather “we worked with one of the speakers over several sessions in which we tried several types of sentences until generalizations started to emerge” (page 5, lines 184-191). In the same line, the author indicates that they consulted with another speaker in some detail and with other members of the consultant’s social group more casually. Overall, it is not clear if the data reported corresponds to production or judgment data; what does it mean “extract” “worked”, “tried” and “consulted” regarding data collection? It is not specified nor detailed the exact procedure for data collection. It is unclear how were those “sentences that came up during field work and were judged to be acceptable” (page 5, lines 192-193) tested and even how many consultants did the author have. Note that the author indicates on footnote 4 that this is “the best method to extract this kind of data” (page 5, footnote 4), despite the lack of information on the method.

 

Even if his goal is “to seek evidence for a particular theoretical point” the author should provide details on the method for collecting the empirical data presented. 

More specifically, I would like to add that I have some knowledge on one of the code-switching varieties the author mentions, and I have some doubts regards judgments presented in examples 11-16 (especially 11b). Consequently, I disagree in the conclusion that “using [o] as a default form for nouns that may denote feminine or masculine individuals” (page 8 lines 311-314).

 

As far as evidence from Russian/Kazakh code-switching coming from Auer and Muhamedova (2005) and Muhavedova (2006), there is no quantitative information on the phenomena the author refers to. For example, when the author states that concord “only disappears systematically in code-swithing” (page 9, lines 368-269) there is no detail on the number of examples on the corpus. Similar examples are found on page 12 (lines 456-457) and page 13 (line 520).

Third, in the case of the German/Turkish community, the author does not describe the data gathering method. The author only indicates that “I met with two language consultants over several skype sessions as we worked together to construct sentences that sounded good to them and similar sentences that sounded off” (page 10, lines 387-390). The small number of consultants and the lack of systematicity /methodology for data collection makes it difficult to conclude that “the data are indeed real and the judgments cannot be chance” (page 14, lines 560-561).

Regarding the theoretical proposals, I will not evaluate them considering they are based on unclear empirical evidence. My advice for the author would be first to collect data following the same methodological procedure systematically on the three communities with a group of bilinguals in each to see if the data the author refers to in this paper is confirmed, and second to report on the procedure, materials and participants in detail in the paper.

For methodological decisions on empirical studies on code-switcching, see references below:

Beatty-Martínez, A., Valdés Kroff, J., & Dussias, P. (2018). From the Field to the Lab: A onverging Methods Approach to the Study of Codeswitching. Languages, 3(2), 19.

Gullberg, M., Indefrey, P., &Muysken, P. (2009). Research techniques for the study of code-switching. In B. E. Bullock & A. J. Toribio (eds.), The Cambridge handbook on linguistic code-switching, pp. 21–39. Cambridge: Cambridge University Press.

Munarriz-Ibarrola, A., Parafita Couto, M. C. & Vanden Wyngaerd, E. (eds.) (2018). Methodologies for intra-sentential code-switching research. Special Issue of Linguistic Approaches to Bilingualism, 8 (1)

Stadthagen-González, H., López, L., Parafita Couto, M. C., & Párraga, C. A. (2018). Using two-alternative forced choice tasks and Thurstone’s law of comparative judgments for code-switching research. Linguistic Approaches to Bilingualism, 8(1), 67-97. doi:doi:10.1075/lab.16030.sta.

 

Author Response

Reviewer 3.

 

R3 is concerned that the claims of this submission are based on a small set of native speakers and the methodology of data elicitation is not presented in detail. This is absolutely a fair concern. I have written a new pair of paragraphs detailing the manner in which data collection took place (lines 157-230). I understand that the limited field-work that I was able to carry out with language consultants does not lead to firm wide-reaching conclusions. However, the crucial fact is that four speakers of two very different code-switching varieties rejected concord when the case morphology was drawn from a language without concord. These results converge with what Muhamedova found in her own field-work. I think it is worthwhile to submit these results to the community of linguists, with the understanding that any conclusions are meant to be regarded as hypotheses and not as established fac.

Reviewer 4 Report

General comments

The paper offers an original contribution to the understanding of concord and how it may be driven by the head K. The typological scope is adequate (evidence drawn from three distinct language families), and it convincingly argues that code-switching varieties constitute a relevant and original source of data for theoretical syntax.

That being said, two important flaws need to be addressed before publication. First, the data-gathering methodology should be more systematic, and the number of independent speakers who provide judgements/examples needs to be increased in order to reach the scientific standard of replicability. Second, the methods and empirical data should be described more thoroughly to ensure that readers can engage critically with the text. In this respect, it would be useful to include an appendix with the complete list of the code-switched sentences, possibly combined with an overview table with the accepted/rejected combination of features in each code-switched variety.

Specific comments on content

1. Methodology

lines 184-191 and footnote 4: some aspects of the methodology are described here (and at the beginning of section 2.3. However, more details needed, and probably better to devote a specific (sub)section to data-gathering methods. It is stated: "we tried several types of sentences". What is it implied by the verb 'try'? Grammaticality judgements can be fairly nuanced and there are ways to describe those details (see next comment). However, it is not feasible for the reader to evaluate a statement like "The judgments reported here were, to the best of my understanding, fairly consistent", or "some other speakers hesitated somewhat" (line 217).

page 5, footnote 4: argument about consultants overlooking small differences between sentences not compelling. Appropriate experimental designs can deal with those issues, such as forced 2-alternative choice tasks (see e.g. the application (even if marginal) of Thurstone's law in linguistics, and particularly in code-switching studies).

page 5, footnote 4: it is claimed that it would take 30-40 minutes "for consultants to evaluate the relevant stimuli". It would be very helpful for the reader to have access to the complete list of "relevant stimuli", in order to obtain a general overview of the feature combinations which were tested and those which were not.

Considering the previous three comments, I encourage the author to obtain some form of systematic grammaticality judgement (instead of binary, better Likert scale or 2-alternative choice task). These judgements should be performed by a minimum of 3 speakers (to avoid ties) who have not taken part in the construction of the stimuli. Further, the obtained ratings should then be reported in the paper in a clear way.

page 5, footnote 4: "I am not trying to characterize a community of speakers." This statement is problematic. Throughout the paper, the author makes claims about 'code-switching varieties', giving the impression that the data here presented are indeed representative of a community of speakers. If those sentences described as ungrammatical in the paper were consistently judged grammatical (and vice versa) by other members of the relevant community, the theoretical claims would lose their empirical support.

line 192: more details needed about the fieldwork ("run-the-mill code-switched sentences that came up during field work"). Who took part during the fieldwork sessions? Where/in which context did it take place?

line 530: it should be made explicit that example (34) (taken from Vergara 2018) is a constructed example, not a sentence observed in a corpus of naturalistic data.

2. Basque data

Given my expertise, I can assess the Basque/Spanish data, but not the other two varieties.

line 151: the author reports that the consultants prefer Biscaian Basque morphology (i.e. [ie]), but none of the other examples follow the Biscaian dialect, but rather standard Batua verb/noun morphology. It may be worth clarifying.

line 212-213: the argument about the absence of Basque D does not follow. The author claims that Basque D "would normally be obligatory" in the KP; however, note that this is not true when a quantifier is present (i.e. in (11a) there is 'algún', of which the Basque equivalent (e.g. zenbait) would *require* the absence of the D in the N (i.e. zenbait maisuk vs *zenbait maisuak). Example (11b) instead does not include a quantifier and does include a Basque D within 'maestroek'.

line 221-: from examples 10 and 11 onwards, all the Basque/Spanish code-switched examples include the structure Adj+N (e.g. anciano maestrok). This is striking and requires some explanation, since both Spanish and Basque strongly prefer N+Adj order. Does this reflect a direct calque from English? Would the reverse order elicit the same judgements?

line 1256: related to the previous comment: the word order of example (68) suddenly reverses from all the previous Adj+N cases (espléndida cama) to a N+Adj order (cama espléndido). This is left unexplained.

Examples (11e-f) and (13e-f) include either the N or the Adj with a plural morpheme, but there is no alternative where both N and Adj show plural concord (as would be the case in a Spanish DP). Would this alternative be judged grammatical? This would yield a code-switched form which is comparable to, say, Spanish band names (i.e. proper nouns) used within a Basque KP, such as "Lendakaris Muertos-ek joko dute" (found in online music site www.entzun.eus).

line 628: note that not all the cited code-switched examples show an "absence of the Basque D": e.g. (11b) Anciano maestroek...

line 231, 1226 (...): note that the constructed examples with the words 'maestra-ek' or 'cama-ei' include the vowel sequence /ae/, which violates Basque phonotactics. In such cases, when the suffix -ek is added to a noun ending in -a (e.g. neska 'girl'), the vowels coalesce into 'neskek'.

On a more general note: virtually all the constructed examples shown in the text strike me as ungrammatical (although some may be judged better-formed than others). This may be partially due to the unusual Adj+N order, as mentioned above. The paper would gain robustness if it showed that this kind of stucture (Spanish Adj+N DP embedded within a Basque phrase) is attested in some corpus. Additional supporting evidence would come from grammaticality judgements systematically elicited from a handful of speakers.

3. Other

line 434: two Turkish ACC case affixes are mentioned [i]/[unu]. Please explain the alternation if deemed relevant.

line 482: the claim that German plural is "a notorious jungle" would benefit from a supporting reference.

line 543-544: last sentence may be omitted ("I am aware that...").

line 560: claim (i) is an overstatement. Note that only the Kazakh/Russian data are drawn from conversational corpora. For the other two varieties, judgement data has been (informally) collected, from a very limited number of speakers, "to seek evidence for a particular theoretical point" (end of footnote 4). Taking this into account, I would omit claim (i) about the robustness and non-chance nature of the data.

line 625: although it is true that a form like 'ancian' does not violate Spanish phonotactics, many other adjectives in that form would: e.g. alt-o/a, anch-o/a, bland-o/a, amarg-o/a, correct-o/a.

Specific comments on form

1. Minor typos or form issues

line 60: "and selects Num, consists of"

line 64: "in search of valued version"

line 177-178: the word 'masculine' should be replaced by 'feminine', and vice versa

page 5, footnote 4: uppercase 'Q' in 'qualtrics'

line 203: 'dendadia-ri' should be 'dendaria-ri'

line 221: the auxiliary 'dut' seems a typo, as it is a 1st person singular auxiliary, but the sentence requires a 3rd person (singular 'du', or plural 'dute').

line 249: "The examples in (13) we have..."

line 273-274: b and c should be c and d. Same issue in example (16d) (line 306).

line 287-: most instances of the word espléndido/a are missing the accent.

line 289: the gloss should read "on *some* splendid bed"

line 569: 'quite' instead of 'quapite'

line 600: period after parenthesis.

line 601: 'attributive'

line 602: "the latter includeS"

line 615: check missing special characters (check also lines 910, 947, 1063)

line 617: 'German' instead of 'English'

line 634: section 2, instead of (2).

line 928: missing quotes around the gloss 'victim'

line 999: include sample nouns, as in the previous two lines.

page 26, footnote 13, line 2: 'masculine' instead of 'feminine'

line 1142: remove the top node of the tree

line 1156: the translation should read 'man' (not 'boy')

line 1180: "the absence plural agreement"

line 1280: missing space between 'maestr-' and 'teacher'

2. Glossed examples

There are several inconsistencies in the glossed examples. Please double check: which parts are (not) in italics, the word-by-word alignment of the gloss, that all relevant sentences are fully glossed.

line 99-100: check alignment; 'we' glossed as 'hacen'

line 110: 'Korean' should be in italics

line 127-: make gloss consistent with other examples

line 229 and 233: maestro-ek vs maestr-o-s-ek; check consistency of hyphens in glossed examples throughout.

line 272: example (14) is missing some glosses; check throughout.

line 1110: check consistency of glosses within tree diagram examples. E.g. example (61) includes gloss in parentheses, but (60) does not.

line 1306: suffix -ij should be in italics.

Author Response

Reviewer 4.

 

I would like to thank R4 for their detailed and insightful review. In particular, R4 was extremely helpful in detecting some serious errors in my Basque data. This version corrects these errors, as I detail below.

 

R4 is also concerned that the claims of this submission are based on a small set of native speakers and the methodology of data elicitation is not presented in detail. This is absolutely a fair concern. I have written a new pair of paragraphs detailing the manner in which data collection took place (lines 157-230). I understand that the limited field-work that I was able to carry out with language consultants does not lead to firm wide-reaching conclusions. However, the crucial fact is that four speakers of two very different code-switching varieties rejected concord when the case morphology was drawn from a language without concord. These results converge with what Muhamedova found in her own field-work. I think it is worthwhile to submit these results to the community of linguists, with the understanding that any conclusions are meant to be regarded as hypotheses and not as established fact.

 

Here is how I addressed R4’s comments on the Basque data:

 

(i) “the argument about the absence of Basque D does not follow.”

 

That is true, I hadn’t realized that algún could have the same effect on the structure as zenbait. I eliminated the claim.

 

(ii) “the structure Adj+N”.

 

True, this was a bizarre error of mine. I did try both A+N and N+A orders with my consultants but A+N, not surprisingly, was rejected. I have changed the examples to N+A.

 

(iii) “there is no alternative where both N and Adj show plural concord”.

 

True, I didn’t include examples of both N and A inflection in (11) but I did include them in (14). Since I have the relevant data, I added them to (11).

 

‘"Lendakaris Muertos-ek joko dute (found in online music site www.entzun.eus ).”

 

This is a very interesting piece of data that certainly challenges the idea that there is no concord when the noun phrase is headed by a Basque case morpheme. It is discussed under footnote 5, suggesting that since Lendakaris Muertos is a proper name and the apparent suffixes are in fact part of the root.

 

(v) “note that not all the cited code-switched examples show an absence of the Basque D: e.g. (11b) Anciano maestroek...”

 

I assume R4 takes the [e] in maestroek to be a determiner. I am uncertain how to deal with this. I rely on Hualde and Ortiz de Urbina’s grammar, where [ek] appears glossed as plural ergative and [e] is not glossed separately or discussed as a determiner. That is also how I saw it glossed in the other literature I consulted. Thus, I left this unchanged.

 

(vi) “the vowel sequence /ae/, which violates Basque phonotactics” I have added footnote 6 about this.

 

(vii) “two Turkish ACC case affixes are mentioned [i]/[unu]. Please explain the alternation if deemed relevant.”

 

This was another error of mine. [un] is the possessor while [u] is the accusative case marker (I tested some sentences with possessors and other noun phrases with my consultants that never made it to the paper). The difference between [i] and [u] is due to vowel harmony. I corrected the examples taking out the possessor, which is irrelevant for my argument.

 

(viii) “the claim that German plural is "a notorious jungle" would benefit from a supporting reference.”

 

I do not intend this claim to be scientific, I only mean that there are several ways to express number on nouns in German, as I show in the following example (31).

Round 2

Reviewer 3 Report

In this second version, the author has improved the manuscript, however, the methodological issue is not still addressed as previously suggested. As the author indicates in the reply, (s)he added some new paragraphs (pages 5-7, lines 160-237) in an attempt to detail “the manner in which data collection took place”. The author has added more sociolinguistic information on the 31 participants from the Kazakh/Russian bilingual corpora, and the two consultants in Spanish/Basque and German/Turkish communities, respectively (lines 160-179). However, it still is not clear neither the methodology nor the procedure for data collection. As such, the method is not reliable, and, consequently, there is no way that the study is replicated. There is no more detail on the way (s)he “worked with one of the speakers of each language pair over several sessions in which we tried several types of sentences until generalizations started to emerge” (lines 183-186). What procedure did the author use? What does (s)he mean by “ I needed some time to “train” my consultants to focus on the features that I was interested”? I would advice the author includes the exact procedure (s)he followed. In the same vein, (s)he states that “This “training” was crucial: I wanted to check what morphology the noun  and adjective in Lx exhibit when selected by a case marker of Ly and so I tried several possible combinations of gender and number and tested them in isolation and in pairs” (lines 194-196). If this is a judgment study, then the exact sentences in the order presented should be included as an appendix.

Surprisingly, the author him/herself indicates that prepared a Qualtrics survey on the Basque/Spanish data, where (s)he received 6 answers, but that (s)he did not consider such random results as in the author’s opinion “the respondents to the survey were not able to pay  attention to the crucial distinctions”. Consequently, the author “abandoned this path” in favour of the “labor-intensive approach” (which is not detailed at all in the paper). My impression from these explanations is that the author disregarded the judgment data (s)he obtained and (s)he got the data presented here by “training” (forcing?) the participants in order to target what (s)he needs as a proof for his/her theoretical claims. This uncommon procedure and the explanations on the way the data were collected let us put into question the validity of the data. My advice for the author would be to try with another material (sentences) and/or instruction, but not to ignore the data, more so when dealing with CS data, given that data do not always converge.

For example, when he indicates that “I would be very surprised if they accepted examples such as (11 d-f), (13 d-f), (14 c-d) or (16d-e); if any Basque speakers accept these sentences, these judgments would constitute direct counterexamples to my proposal”. The author should show empirical evidence that those sentences are rejected, whereas others are accepted. Note that the author changed in the present version the order of the adjective and the nouns in the examples from the Spanish/Basque community (11-16, a.o.), what leads us to question once again the way (s)he deals with the data.

 In general, I would insist again in giving the importance it deserves to the method for data collection (find some references on this issue below) in order to obtain valid and reliable data even in theoretically oriented papers such as this. Otherwise, the “evidence” for a particular theoretical point cannot be valid.

 

Beatty-Martínez, A., Valdés Kroff, J., & Dussias, P. (2018). From the Field to the Lab: A converging Methods Approach to the Study of Codeswitching. Languages, 3(2), 19.

Gullberg, M., Indefrey, P., &Muysken, P. (2009). Research techniques for the study of code-switching. In B. E. Bullock & A. J. Toribio (eds.), The Cambridge handbook on linguistic code-switching, pp. 21–39. Cambridge: Cambridge University Press.

Munarriz-Ibarrola, A., Parafita Couto, M. C. & Vanden Wyngaerd, E. (eds.) (2018). Methodologies for intra-sentential code-switching research. Special Issue of Linguistic Approaches to Bilingualism, 8 (1).

Author Response

R3 and R4 complain that this paper does not present the results of a survey that includes 30+ speakers followed by a careful quantitative analysis of the results. This is standard practice in psycholinguistically oriented journals such as Bilingualism, Language and Cognition and I have used it myself in some of my published work. But it is not the only approach to data collection taken in the field. Unlike R1 and R2, R3 and R4 seem to have misunderstood the nature of this paper. This paper is a contribution to theoretical linguistics, an area in which it is common practice to present theoretical, exploratory proposals based on a data base provided by a small group of speakers inspected in depth. RQ’s assertion that “the empirical data provided do not meet the minimum standards for publication” is wrong. Proof that the approach taken in this paper is common practice can be found by perusing any of the theoretical linguistics journals current in the field such as Linguistic Inquiry and Natural Language and Linguistic Theory. One can find ample examples also in the pages of Languages. Contrarty to R3’s assertion that “there is no way that the study is replicated”, replicating this sort of study is very easy if one finds the relevant speakers. R3 also seems to believe that the goal of this paper is to provide a description of a geographically delimited community of speakers; I made clear, in a previous response, that that is not my aim. My aim is theoretical exploration.

The comments that R3 and R4 provide suggest that R3 and R4 are not familiar with ordinary field methods of data collection. R3 seems to believe that training participants is about bullying them to give you the acceptability judgments the researcher is looking for. Instead, training participants is about developing their metalinguistic awareness so they pay attention to the linguistic features that the researcher is focusing on.

Further, R3 advises that “the author includes the exact procedure (s)he followed.” Along the same lines, R4 wants me to list all the sentences that I tested with the participants. This is further evidence that R3 and R4 do not understand field work. If they did, they would know that the spontaneous production of sentences that arises in one-on-one field work looks more like brainstorming and does not always allow for careful planning and recording. As for the sentences that I have in fact written down and I didn’t include in the paper, I don’t see how the reviewers or anyone else could be interested in sentences that misfired for irrelevant reasons (bad lexical choice, bad case marker, bad constituent order) or sentences that are identical to the ones I report in the paper except for one or two lexical items.

In the current review, I have made some changes (highlighted in the submission) that emphasize the exploratory character of this paper. All the changes are in section 1.

Reviewer 4 Report

The author has taken into account most of my comments in the latest version of the manuscript. However, the two main weaknesses included as general comments in the previous review have not been adequately addressed. Specifically, the empirical data provided do not meet the minimum standards for publication.

The author has significantly improved the description of the methodology (pages 5 and 6). This constitutes a valuable addition to the paper, and I would suggest that it is preceded by its own heading (e.g. 'Method' or 'Data collection'). However, it is this particular section which highlights the need for a more extensive gathering of data (i.e. possibly grammaticality judgements).

The author argues (lines 220-225) that the Basque-Spanish code-switching community is heterogeneous and that different speaker communities may behave differently. This does not constitute a blank cheque or justify analysing the data of just one or two speakers arguing that, anyhow, there is so much variation. To the contrary, if the author is dealing with Gernika Basque speakers, then the goal should be to gather a sample of, at least, Gernika code-switchers. This is particularly critical given that the current consultants have conducted a double task, namely, producing examples *and* judging their grammaticality.

Hence, in order to back the paper with reliable empirical data, it is necessary that additional speakers provide judgements over a controlled set of sentences. In the revised version, the author now mentions having tried a Qualtrics survey, but "Unfortunately, I obtained only random results". Disregarding seemingly random results looks questionable research practice (if done ad-hoc, in a non-principled way). If that kind of survey is deemed unworkable for some concrete reason, then an alternative method should be considered.

Finally, I insist that it would be very valuable for the reader to have access to the complete set of accepted/rejected sentences; or, if too many, to a systematic summary thereof. Besides the benefit of providing a more thorough presentation of the argument, it promotes data transparency and replicability, which are cornerstones of open-access research venues like the journal Languages.

 

Author Response

R3 and R4 complain that this paper does not present the results of a survey that includes 30+ speakers followed by a careful quantitative analysis of the results. This is standard practice in psycholinguistically oriented journals such as Bilingualism, Language and Cognition and I have used it myself in some of my published work. But it is not the only approach to data collection taken in the field. Unlike R1 and R2, R3 and R4 seem to have misunderstood the nature of this paper. This paper is a contribution to theoretical linguistics, an area in which it is common practice to present theoretical, exploratory proposals based on a data base provided by a small group of speakers inspected in depth. RQ’s assertion that “the empirical data provided do not meet the minimum standards for publication” is wrong. Proof that the approach taken in this paper is common practice can be found by perusing any of the theoretical linguistics journals current in the field such as Linguistic Inquiry and Natural Language and Linguistic Theory. One can find ample examples also in the pages of Languages. Contrarty to R3’s assertion that “there is no way that the study is replicated”, replicating this sort of study is very easy if one finds the relevant speakers. R3 also seems to believe that the goal of this paper is to provide a description of a geographically delimited community of speakers; I made clear, in a previous response, that that is not my aim. My aim is theoretical exploration.

The comments that R3 and R4 provide suggest that R3 and R4 are not familiar with ordinary field methods of data collection. R3 seems to believe that training participants is about bullying them to give you the acceptability judgments the researcher is looking for. Instead, training participants is about developing their metalinguistic awareness so they pay attention to the linguistic features that the researcher is focusing on.

Further, R3 advises that “the author includes the exact procedure (s)he followed.” Along the same lines, R4 wants me to list all the sentences that I tested with the participants. This is further evidence that R3 and R4 do not understand field work. If they did, they would know that the spontaneous production of sentences that arises in one-on-one field work looks more like brainstorming and does not always allow for careful planning and recording. As for the sentences that I have in fact written down and I didn’t include in the paper, I don’t see how the reviewers or anyone else could be interested in sentences that misfired for irrelevant reasons (bad lexical choice, bad case marker, bad constituent order) or sentences that are identical to the ones I report in the paper except for one or two lexical items.

In the current review, I have made some changes (highlighted in the submission) that emphasize the exploratory character of this paper. All the changes are in section 1.