Preposition Allomorphy in Calabrian Greek (Greko) and Standard Modern Greek and Its Theoretical Implications

Round 1

Reviewer 1 Report

The descriptive part in sections 1-4 is fine, though it fails to notice that the phenomena described are virtually identical to the de-di alternation in Italian. Also, it is not clear whether example (22) (which is repeated four times in the paper) is productive or a frozen expression: can one say 'stu skilu tin kardia' or 'stu tapinu to aftokinito'? If not, the agrammaticality of (76) becomes irrelevant.

As for the theoretical account in 5, it seems that a lot of formalism is introduced that is in fact irrelevant. For example, expanding P into Path source>(Path goal)>Place does not really add anything to the argumentation, and makes the argument not immediately applicable to the Italian de/di allomorphy.
Also the discussion about the analysis of demonstratives as heads or specifiers does not seem to matter, since all that counts is linear adjacency (I totally agree with 82).

Section 5.5 is very interesting and should be given due prominence. Again, it is paralleled in Italian: 
di+la+tesi (stress on te) > della tesi
di+latino (stress on ti) > di latino

Maybe it would be interesting to explore how these phenomena are DIFFERENT from wanna contraction where linear adjacency also matter but does not ignore empty elements and does not happen with the specifier of the complement as in (22), as shown by this example from Pullum:

I don't want to flagellate oneself in public to become common practice in this monastery

Comments for author File: Comments.pdf

Author Response

Thanks for the helpful comments that I have attempted to consider in my revisions. My itemized responses to specific points are included below under bullet points.

The descriptive part in sections 1-4 is fine, though it fails to notice that the phenomena described are virtually identical to the de-di alternation in Italian. Also, it is not clear whether example (22) (which is repeated four times in the paper) is productive or a frozen expression: can one say 'stu skilu tin kardia' or 'stu tapinu to aftokinito'? If not, the agrammaticality of (76) becomes irrelevant.

    - I have added a note concerning the Italian de-di alternation.
    - The configuration may be stylistically marked as somewhat formal or poetic, but it seems to be productive in principle. I've added additional examples.

As for the theoretical account in 5, it seems that a lot of formalism is introduced that is in fact irrelevant. For example, expanding P into Path source>(Path goal)>Place does not really add anything to the argumentation, and makes the argument not immediately applicable to the Italian de/di allomorphy.

    - I find it important to at least acknowledge that the assumption of an internally complex structure is state of the art in the literature on spatial adpositions and, if taken seriously, impacts the way theoretical models can analyse the alternation under discussion. However, I have compressed the relevant section to account for the fact that it is not the main focus.

Also the discussion about the analysis of demonstratives as heads or specifiers does not seem to matter, since all that counts is linear adjacency (I totally agree with 82).

    - That's true, it turns out to be less crucial as I reject the portmanteau analyses after all. I have reduced the prominence of this discussion and just address the issue in order to highlight at which points the choice would make a difference for present purposes.

Section 5.5 is very interesting and should be given due prominence. Again, it is paralleled in Italian: 
di+la+tesi (stress on te) > della tesi
di+latino (stress on ti) > di latino

    - Thanks, I have added a brief reference to that pattern. 

Maybe it would be interesting to explore how these phenomena are DIFFERENT from wanna contraction where linear adjacency also matter but does not ignore empty elements and does not happen with the specifier of the complement as in (22), as shown by this example from Pullum:

I don't want to flagellate oneself in public to become common practice in this monastery

    - I agree that this is an interesting contrast. I won't be able to address it in any detail, but have added the observation to the conclusion.

Reviewer 2 Report

The paper at first seems/is claimed to be about the distribution of an/asce in Greko, but eventually it turns out that the relevant observations/restrictions are a proper subset of those concerning se/s in SMG and that the choice of an account heavily relies on the latter (more so than the facts about an/asce). So, the reader is left wondering why Greko is presented as being the focus of the study (and sometimes why it is discussed at all, since it has at best an ancillary role as offering an additional example of the phenomenon under discussion).

The phenomenon itself can be described in very simple(r) terms: the allomorph of P is chosen iff P is linearly adjacent to a definite article (in any case). I think this very simple generalisation could/should be the explanandum of any technical account. An account immediately focusing on this would make it unnecessary to go to great lengths to rule out the allomorph before anything other than definite articles. It also looks as if the phenomenon at hand somehow discredits the theories previously developed (both within nanosyntax and DP) for portmanteaux. It would suffice to briefly argue/make it clear that "sto"/"ando" are not portmanteaux and then, obviously, none of these theories apply. Similarly, I find the discussion of the internal structure of Ps rather superfluous, I do not understand if it has any bearing to the analysis itself, e.g. what would be different if there was just one layer of structure.

The most crucial fact for/in favor of the linear account is the obligatoriness of 's' in front of fronted definite genitive DPs, as far I understand. Am I correct in assuming that this is not to be found in Greko, as fronted genitives are not possible? (This is why I think that the distribution/the overall evidence that SMG s/se alternations provide is a superset of the evidence provided by asce/an and more useful for an analysis). I would also like to draw the author's/authors' attention to the following question: to the extent that indefinite DPs with fronted definite genitives are possible (I think they are, to some extent), what is the prediction regarding the grammaticality of 'stu/stis'? (and is it borne out?)

Some specific points:

lines 94-97: There is no need to confuse synchrony and diachrony; I don't think it is really necessary for two (synchronic) allomorphs to be etymologically related.

169: platespi or platepsi?

185 (12): the whole presentation of this (quite easy to grasp, in fact) distribution is somewhat repetitive.

Regarding "se/s", the facts are well known, described and discussed in grammars and previous literature (especially works on indirect objects/ditransitives, which are not cited in the paper)

283-284:  "to which extent...structure": something is missing?

examples/trees (24), (25): it is not clear how Pantcheva's tree would differ from Radkevich's (and which of the trees presented here corresponds to which analysis exactly)

The general question regarding the XP vs X0 status of demonstratives is discussed at length in many of the works on demonstratives cited and also in works not cited: a summary of the relevant arguments would be useful, even though it is quite clear that demonstratives (given all sorts of properties, including the fact that they can be modified by locative or degree adverbs, e.g. 'akrivos') are phrasal, at least in pre-D and post-N positions.

example (30b) is rather unfortunate: contraction of se --> s is in fact perfectly possible here.

602-603: it is mentioned that treating Dem as a head would "avoid the wrong predictions" but this clearly also applies to several of the abovementioned account (and I'm not sure that the author made it equally clear)

in relation to 759-760: see my question above regarding potential non-definite Ds heading the overall DP.

 occasional/few typos: e.g. cases of "this" followed by a plural noun etc. 

Author Response

Thanks for the helpful comments that I have attempted to consider in my revisions. My itemized responses to specific points are included below under bullet points.

The paper at first seems/is claimed to be about the distribution of an/asce in Greko, but eventually it turns out that the relevant observations/restrictions are a proper subset of those concerning se/s in SMG and that the choice of an account heavily relies on the latter (more so than the facts about an/asce). So, the reader is left wondering why Greko is presented as being the focus of the study (and sometimes why it is discussed at all, since it has at best an ancillary role as offering an additional example of the phenomenon under discussion).

    - Fair point, I have rearranged the introduction to manage expectations more transparently.

The phenomenon itself can be described in very simple(r) terms: the allomorph of P is chosen iff P is linearly adjacent to a definite article (in any case). I think this very simple generalisation could/should be the explanandum of any technical account. An account immediately focusing on this would make it unnecessary to go to great lengths to rule out the allomorph before anything other than definite articles. It also looks as if the phenomenon at hand somehow discredits the theories previously developed (both within nanosyntax and DP) for portmanteaux. It would suffice to briefly argue/make it clear that "sto"/"ando" are not portmanteaux and then, obviously, none of these theories apply. Similarly, I find the discussion of the internal structure of Ps rather superfluous, I do not understand if it has any bearing to the analysis itself, e.g. what would be different if there was just one layer of structure.

    - I have streamlined the description of the phenomena in Sections 3 and 4.
    - The internal structure of spatial adpositions is relevant a) for ensuring compatibility with the extant literature, b) as possible motivation for portmanteau-style analyses of the P-D combinations and c) to illustrate that this type of linear adjacency condition requires a mechanism that allows ignoring intervening null elements (i.e. something akin to Embick's pruning), in spite of the potential issues raised for this by Moskal & Smith. It is, admittedly, not a major point though, so I have further compressed that section.

The most crucial fact for/in favor of the linear account is the obligatoriness of 's' in front of fronted definite genitive DPs, as far I understand. Am I correct in assuming that this is not to be found in Greko, as fronted genitives are not possible? (This is why I think that the distribution/the overall evidence that SMG s/se alternations provide is a superset of the evidence provided by asce/an and more useful for an analysis). I would also like to draw the author's/authors' attention to the following question: to the extent that indefinite DPs with fronted definite genitives are possible (I think they are, to some extent), what is the prediction regarding the grammaticality of 'stu/stis'? (and is it borne out?)

    - The linear-adjacency account predicts that the definite-sensitive allomorph shows up. While such constructions do indeed seem to be relatively marked, it seems very clear that the reduced form s- must be used if the preposed possessor begins with a definite article, in line with that prediction. I have added a brief discussion in Section 5.4.
    

Some specific points:

lines 94-97: There is no need to confuse synchrony and diachrony; I don't think it is really necessary for two (synchronic) allomorphs to be etymologically related.

    - Fair enough, I have weakened this point and simply mention the coherence between Rohlfs' etymological proposal and his view on the asce/an relation.

169: platespi or platepsi?

    - It's indeed "platespi" (I'm not sure if Greko phonology allows /ps/ clusters, older /ps/ is typically resolved into /sp/ or /ts:/, e.g. psomi "bread" is "spomi" or "zzomi" depending on dialect). 

185 (12): the whole presentation of this (quite easy to grasp, in fact) distribution is somewhat repetitive.

    - I have rearranged Section 3.2 to make the distribution clearer from the outset (also accommodating readers who may want to skip the supporting examples).

Regarding "se/s", the facts are well known, described and discussed in grammars and previous literature (especially works on indirect objects/ditransitives, which are not cited in the paper)

    - It still seems relevant to me to point out the patterns clearly.
    - I am not aware of/haven't found detailed discussions of the aspects I wanted to focus on, but added references to Holton et al. 2012: 53 and to Michelioudakis 2012:100f., where relevant aspects are described.

283-284:  "to which extent...structure": something is missing?

    - Whole paragraph reformulated

examples/trees (24), (25): it is not clear how Pantcheva's tree would differ from Radkevich's (and which of the trees presented here corresponds to which analysis exactly)

    - Pantcheva argues for an additional Source projection on top of Path as in (25) - also in line with naonsyntax' assumption of one feature-one head. Probably this is less relevant now, since I have reduced the discussion of this aspect and there is no separate tree for this anymore.

The general question regarding the XP vs X0 status of demonstratives is discussed at length in many of the works on demonstratives cited and also in works not cited: a summary of the relevant arguments would be useful, even though it is quite clear that demonstratives (given all sorts of properties, including the fact that they can be modified by locative or degree adverbs, e.g. 'akrivos') are phrasal, at least in pre-D and post-N positions.

    - The first draft probably placed an undue emphasis on the role of the phrase/head debate for demonstratives. The revision now reflect that the article does not aim to develop a dedicated contribution to this debate. The reason the issue is introduced is to essentially point out the problems a phrasal analysis raises for all of the portmanteau analyses. The linear-adjacency-based analysis I argue for works equally under either assumption, so a further comparison of the different arguments seems to be tangential.

example (30b) is rather unfortunate: contraction of se --> s is in fact perfectly possible here.

    - Yes, but this is the optional hiatus avoidance I discuss earlier. If this was the same effect as involved in the proposed allomorph s-, then se would be ruled out here, which it is not. I have clarified this point.

602-603: it is mentioned that treating Dem as a head would "avoid the wrong predictions" but this clearly also applies to several of the abovementioned account (and I'm not sure that the author made it equally clear)

    - I am not sure if I understand this point. In my understanding the problem arising from a phrasal treatment of demonstrative for the portmanteau accounts is due to their structural sensitivity and I'm not aware how a phrasal analysis of demonstratives could avoid this issue. Given that the revised version reduces the prominence of the question of a phrasal/head analysis of demonstratives due to the overall issues facing the portmanteau accounts, this issue may be not so important anymore. Otherwise, I'm also happy to add an acknowledgement of additional ways of avoiding the problem if I can get a pointer to relevant discussions. 

in relation to 759-760: see my question above regarding potential non-definite Ds heading the overall DP.

    - added a brief discussion around new example (58)

 occasional/few typos: e.g. cases of "this" followed by a plural noun etc. 

Reviewer 3 Report

This is a very well written paper that meticulously compares possible analyses of two cases of preposition allomorphy in Greko (between asce and an) and Standard Modern Greek (SMG) (between se and s). As the author explicitly states, the claim concerning preposition allomorphy in Greko is not novel but due to Rohlfs. However, the empirical contribution of the paper is important: the generalization concerning Greko (which is, after all, a severely endangered minority language) is very well explored and illustrated (in section 3). The main theoretical conclusion the author offers is that linear precedence is an important aspect of the analysis of the phenomena at hand.

 

The main shortcoming of the paper in its current form is that it includes an extremely detailed comparison of analyses that are very similar to each other (all of them adhering to realizational, non-lexicalist models of grammar). This in itself would not be so detrimental – although it does mean that the paper will be of interest only to a small set of readers – if it were not for the fact that the conclusions reached after this thorough comparison is made are very weak. As far as I can see, the author does not, in the end, commit to any particular analysis. His/her general conclusion is offered on the last page of the paper (see (82)). In its revised form, more of the paper should be devoted to developing the proposed account. Moreover, the author should take a stand on how s/he evaluates the analysis of demonstratives as heads, an analysis that crucially needs to be assumed if certain approaches to the data discussed are to be even partly successful at accounting for them. Section 5.3 could be substantially shortened.

 

A list of more specific comments follow.

 

 

p.2, line 36-7: this is the first mention of the assumption that demonstratives are heads in the extended nominal projection, rather than phrasal specifiers. Does the author endorse the assumption? It would be helpful for the reader to know where s/he is heading.

p.3, line 87, correct to literature.

4. 5, example (8): is there any (optional) phonological process to avoid hiatus ever taking place within asce ecine? This would make stronger the similarity with se/s discussed in section 4. The question similarly (though less plausibly) arises for example (10).
5. 6, line 166, correct to an.
6. 7, lines 200-201: I think this caveat should be removed. It would, after all, benefit the paper if the empirical coverage is widened.
7. 10, re (22): this is very nice, could the author mention that s/he will come back to such data? She does on p. 25, example (73).
8. 10, lines 292-294: I have a hard time processing this sentence, which seems to be incomplete.
9. 10, section 5 (introduction): it would help the reader if the author gives an indication of what we will end up with: are the combinations portmanteaux? Are demonstrative modifiers heads?

p.11, above (24) and (25): the trees include the lexical items from SMG and Greko, but the literature from which they are drawn does not necessarily address these particular varieties. A clarification to this effect should be included above the trees.

p.12, line 370: correct to Scandinavian.

1. 13, above (29): readers may already wonder about the alternative order, as in (20b). The author should point out that s/he will return to it presently (see discussion around (33)). Incidentally, the reasons to discard an alternative analysis, which would involve adjunction of QP, could be included in the discussion.
2. 13, above (30): I think reference to the optional phonological rule resolving hiatus should be made here, if only because the author spent some time discussing it in the previous section (see ex (15e), for instance).
3. 15, line 420: correct to an.
4. 15, lines 424-426. Again, I wonder: is this an assumption that the author considers well-founded or not? Will s/he adopt it?
5. 16, lines 463-464: is this a sound assumption to make? On what grounds?
6. 21, line 609: correct to falsely.
7. 22, fn10: the sentence is incomplete.
8. 22 end of section 5.3. Is the conclusion that the forms in question are not port-manteau or that we need a better, alternative analysis of them as portmanteau? It seems to me that the argument goes as follows: all of the analyses for portmanteau considered here fail, in some way or other. So the forms in question do not constitute a portmanteau. If the author does not believe the forms in question are portmanteaux, could there be a shorter way to reach this conclusion?

p.23, line 662: correct to linear.

1. 26, line 790: correct to ensure.
2. 26, line 804: correct to model.
3. 27, line 825: correct to ‘neuter singlular to or plural ta’.

Author Response

Thanks for the very helpful comments! I have inserted my replies under bullet points below after the relevant points.

This is a very well written paper that meticulously compares possible analyses of two cases of preposition allomorphy in Greko (between asce and an) and Standard Modern Greek (SMG) (between se and s). As the author explicitly states, the claim concerning preposition allomorphy in Greko is not novel but due to Rohlfs. However, the empirical contribution of the paper is important: the generalization concerning Greko (which is, after all, a severely endangered minority language) is very well explored and illustrated (in section 3). The main theoretical conclusion the author offers is that linear precedence is an important aspect of the analysis of the phenomena at hand.

 

The main shortcoming of the paper in its current form is that it includes an extremely detailed comparison of analyses that are very similar to each other (all of them adhering to realizational, non-lexicalist models of grammar). This in itself would not be so detrimental – although it does mean that the paper will be of interest only to a small set of readers – if it were not for the fact that the conclusions reached after this thorough comparison is made are very weak. As far as I can see, the author does not, in the end, commit to any particular analysis. His/her general conclusion is offered on the last page of the paper (see (82)). In its revised form, more of the paper should be devoted to developing the proposed account. Moreover, the author should take a stand on how s/he evaluates the analysis of demonstratives as heads, an analysis that crucially needs to be assumed if certain approaches to the data discussed are to be even partly successful at accounting for them. Section 5.3 could be substantially shortened.

    - The revision attempts to make the central point clearer from the beginning (and throughout) that I take the data to show the need for reference to linear adjacency. I hope it is now clearer that I do, indeed, aim to commit to an analysis that makes reference to linear adjacency, with Embick's C1-LIN the most straightforward possibility (the 
    I have also streamlined sections 3 and 4 in order to present the data more effectively. As suggested, I have also strongly compressed and focussed the discusion of the portmanteau analyses in Section 5.3.  
 

A list of more specific comments follow.

p.2, line 36-7: this is the first mention of the assumption that demonstratives are heads in the extended nominal projection, rather than phrasal specifiers. Does the author endorse the assumption? It would be helpful for the reader to know where s/he is heading.

    - I have strongly reduced the prominence of this aspect, since it does indeed end up not playing a crucial role for the eventual analysis. Occasional reference is only made later on (mainly in Section 5.3 and preparatorily in 5.2) in order to point out the  

p.3, line 87, correct to literature.

    - Done

    5, example (8): is there any (optional) phonological process to avoid hiatus ever taking place within asce ecine? This would make stronger the similarity with se/s discussed in section 4. The question similarly (though less plausibly) arises for example (10).
    
    - Great point, I've added brief remarks to this effect in this section as well.
    
    6, line 166, correct to an.
    
    - Done
    
    7, lines 200-201: I think this caveat should be removed. It would, after all, benefit the paper if the empirical coverage is widened.
   
        - Good point, adopted. 
   
    10, re (22): this is very nice, could the author mention that s/he will come back to such data? She does on p. 25, example (73).
    
        - I added a note that this issue will be picked up.
        
    10, lines 292-294: I have a hard time processing this sentence, which seems to be incomplete.
    
        - It was indeed. That part has been thoroughly revised.
    
    10, section 5 (introduction): it would help the reader if the author gives an indication of what we will end up with: are the combinations portmanteaux? Are demonstrative modifiers heads?
        
        - I hope the revised version of the intro of Section 5 provides a clearer overview.

p.11, above (24) and (25): the trees include the lexical items from SMG and Greko, but the literature from which they are drawn does not necessarily address these particular varieties. A clarification to this effect should be included above the trees.

    - Agreed and clarified.

p.12, line 370: correct to Scandinavian.
    
    - Done

    13, above (29): readers may already wonder about the alternative order, as in (20b). The author should point out that s/he will return to it presently (see discussion around (33)). Incidentally, the reasons to discard an alternative analysis, which would involve adjunction of QP, could be included in the discussion.
    
        - I added a note that this issue will be picked up. Footnotes 16, 21 and the paragraph after (53) briefly sketch issues an account with adjoined QP would raise in the present context.
        
    13, above (30): I think reference to the optional phonological rule resolving hiatus should be made here, if only because the author spent some time discussing it in the previous section (see ex (15e), for instance).
    
        - Good point, the discussion should be clearer on this aspect now.
    
    15, line 420: correct to an.
        
        - Done
        
    15, lines 424-426. Again, I wonder: is this an assumption that the author considers well-founded or not? Will s/he adopt it?
    
        - I wouldn't dismiss a head-analysis of demonstratives out of hand, but on reflection I realised that the head/phrase debate turns out to be only a minor point in the present context, considering that I eventually reject the analyses that would require this assumption. I have tried to make this clearer.
        
    16, lines 463-464: is this a sound assumption to make? On what grounds?
    
        - Grimshaw's original extended projections proposal assumes that prepositions form part of extended nominal projections and Svenonius has to adopt this assumption, otherwise a span analysis of French "au" would fail from the start. I have clarified this in footnote 14.
        
    21, line 609: correct to falsely.
    
        - Done
        
    22, fn10: the sentence is incomplete.
    
        - Fixed.
        
    22 end of section 5.3. Is the conclusion that the forms in question are not port-manteau or that we need a better, alternative analysis of them as portmanteau? It seems to me that the argument goes as follows: all of the analyses for portmanteau considered here fail, in some way or other. So the forms in question do not constitute a portmanteau. If the author does not believe the forms in question are portmanteaux, could there be a shorter way to reach this conclusion?
    
        - The section has been streamlined, attempting to take into account the reviewer's suggestion. (As I state in the new conclusion of subsection 5.3, I'm not sure I can entirely exclude the possibility of a portmanteau analysis, but I think if one was to be developed it would have to be very similar to the proposed allomorphic analysis.)

p.23, line 662: correct to linear.

        - Done

    26, line 790: correct to ensure.
    
        - Done
        
    26, line 804: correct to model.
    
        - Done
    
    27, line 825: correct to ‘neuter singlular to or plural ta’.
    
        - Done

 

Reviewer 4 Report

The paper makes a case for allomorphy triggered by linear adjacency with the definite article for the prepositional forms azze/an ‘from’ in Greko and se/s- ‘in/to’in Standard Modern Greek and Greko. The phenomenon is discussed within different versions of post-syntactic accounts akin to Distributed Morphology and Nanosyntax in order to discard the possibility of a portmanteau analysis and to support the proposal that these are cases of allomorphy determined by linear adjacency of P+D. An important piece of evidence for the linear adjacency requirement is the behavior with demonstratives, quantifier olo and prenominal genitives.

 

The article is clearly structured and presented, and the argumentation is sound. However, its scope is somewhat limited. It devotes a lot of attention to contrasting very close accounts against the background of the data set that constitutes the focus of attention, but it is advisable to include a more general discussion of the contrasted approaches relying on portmanteau morphology vs. linear adjacency.

 

One empirical point to be addressed with respect to SMG is the following:

(i) se apolitos oles tis polis

    in aboslutely all the cities

 

Under the proposed account, one would expect the surface form of the preposition to be s- and not se, contrary to fact. Given the assumptions in the paper, this case seems not to follow, because the adverbial apolitos (arguably a specifier of QP) should trigger the allomorphic form, but it does not.

 

Minor points:

-l. 73: variation > variation

-l. 87: literatur > literature

-l. 295: adopt > DELETE

-l. 392: Specify in footnote or in the text what dxP stands for.

-l. 609: false > falsely

-l. 804: modell > model

-l. 807: Insertation > Insertion

Author Response

Thanks for your helpful comments! I have inserted my replies under bullet points below after the relevant points.

 

The paper makes a case for allomorphy triggered by linear adjacency with the definite article for the prepositional forms azze/an ‘from’ in Greko and se/s- ‘in/to’in Standard Modern Greek and Greko. The phenomenon is discussed within different versions of post-syntactic accounts akin to Distributed Morphology and Nanosyntax in order to discard the possibility of a portmanteau analysis and to support the proposal that these are cases of allomorphy determined by linear adjacency of P+D. An important piece of evidence for the linear adjacency requirement is the behavior with demonstratives, quantifier olo and prenominal genitives.

 

The article is clearly structured and presented, and the argumentation is sound. However, its scope is somewhat limited. It devotes a lot of attention to contrasting very close accounts against the background of the data set that constitutes the focus of attention, but it is advisable to include a more general discussion of the contrasted approaches relying on portmanteau morphology vs. linear adjacency.

- The rewritten version of Section 5.3 now contain a discussion that should hopefully address aspects of this suggestion.
 

One empirical point to be addressed with respect to SMG is the following:

(i) se apolitos oles tis polis

    in aboslutely all the cities
 

Under the proposed account, one would expect the surface form of the preposition to be s- and not se, contrary to fact. Given the assumptions in the paper, this case seems not to follow, because the adverbial apolitos (arguably a specifier of QP) should trigger the allomorphic form, but it does not.

    - I'm not sure I understand the reasoning here. To clarify the analysis I would like to support in the article: On the assumption that linear adjacency of se to the definite article triggers the allomorph s-, the default form "se" should occur here (as attested) - optional (!) hiatus avoidance providing the additional possibility of deleting the final vowel yielding "s' apolitos oles tis polis". See the revised discussion in Section 4, which should be more concise than the first draft.
 

Minor points:

-l. 73: variation > variation

    - Fixed

-l. 87: literatur > literature

    - Fixed

-l. 295: adopt > DELETE

    - Fixed

-l. 392: Specify in footnote or in the text what dxP stands for.

    - clarified in main text

-l. 609: false > falsely

    - Fixed

-l. 804: modell > model

    - Fixed

-l. 807: Insertation > Insertion

    - Fixed

Round 2

Reviewer 2 Report

The author addressed most of my comments and requests.

I still think the argument mainly relies on se/s and that asce/an is simply a parallel phenomenon (not even ancillary evidence), with the relevant contexts (i.e. the ones an argument can be built upon) being a subset of those concerning se/s. I would suggest that the introduction should put at least equal emphasis on the two, rather than stressing asce/an as an empirical discovery that can shed new light to the theoretical question of the paper (if anything, the distribution of se/s offers the crucial evidence and it would suffice even if it were discussed on its own).

I am also still puzzled by the fact that, even though s-to is clearly decomposable/analysable, the author deems it necessary to exclude a portmanteau analysis (of the type needed for French au which is not analysable). The author does mention that "[...] given the tight integration of the preposition and the determiner an analysis of these combinations as portmanteaux does not seem entirely implausible either." I think that the author should provide at least one example of a similar phenomenon where decomposability is only apparent and in fact a portmanteau analysis can be shown to be more suitable. Otherwise the detailed critique of portmanteau analyses in relation to se/sto is quite superfluous.

Overall, all the other issues raised in my first review were addressed satisfactorily. There are only few occasional typos, e.g. "limtations" (p.1). 

 

 

 

Author Response

Many thanks for the additional feedback, which has forced me to modify several formulations, hopefully making the article more clear/accessible in the process. My specific replies are again interspersed below.

 

The author addressed most of my comments and requests.

I still think the argument mainly relies on se/s and that asce/an is simply a parallel phenomenon (not even ancillary evidence), with the relevant contexts (i.e. the ones an argument can be built upon) being a subset of those concerning se/s. I would suggest that the introduction should put at least equal emphasis on the two, rather than stressing asce/an as an empirical discovery that can shed new light to the theoretical question of the paper (if anything, the distribution of se/s offers the crucial evidence and it would suffice even if it were discussed on its own).

    - The introduction has been modified to more explicitly foreground the issue of linear adjacency. In turn, I also attempt to present the asce/an and se/s- alternations on equal footing. (The new footnote 3 also briefly alludes to the practical significance of explicit argumentation for an allomorphic treatment of the asce/an alternation.)

I am also still puzzled by the fact that, even though s-to is clearly decomposable/analysable, the author deems it necessary to exclude a portmanteau analysis (of the type needed for French au which is not analysable). The author does mention that "[...] given the tight integration of the preposition and the determiner an analysis of these combinations as portmanteaux does not seem entirely implausible either." I think that the author should provide at least one example of a similar phenomenon where decomposability is only apparent and in fact a portmanteau analysis can be shown to be more suitable. Otherwise the detailed critique of portmanteau analyses in relation to se/sto is quite superfluous.

    - I agree that a portmanteau analysis of s-to is not particularly attractive and if I were to write only about the se/s- alternation, I might be content with rejecting a portmanteau analysis out of hand with a brief paragraph. However, it is not as obvious to me that for patterns like "asce"+"to" -> "ando" a portmanteau analysis is inherently implausible. Given the limited range of literature on Greko, I cannot point to specific proposals to that effect, but it seems more useful to make the implications of such an approach explicit and thereby motivate the rejection in favour of the allomorphic analysis I advocate in more detail than just stating that the allomorphic analysis is conceptually more preferable. I have tried to make this more transparent at the outset of Section 5.3 now.

Overall, all the other issues raised in my first review were addressed satisfactorily. There are only few occasional typos, e.g. "limtations" (p.1). 

    - Fixed.

Reviewer 3 Report

The author has adequately addressed the issues raised in my previous review. I provide a small list of truly minor suggestions regarding formulation below.

 

Abstract

• The abstract states at the outset that “The article argues that the prepositions asce ‘from’ and an ‘from’ in the south Italian Greek variety Greko, as well as the preposition se ‘in, to, into’ found in both Greko and Standard Modern Greek are allomorphs…”.

 

I would suggest reformulating the second part (“as well as the preposition se..”) in a way that makes reference to the two allomorphic realizations of the SMG preposition se (there is, I think, an absurd reading of the sentence above whereby allomorphy holds between se and asce and an).

 

Section 4

• I would recommend reformulating the title to something like “allomorphic realizations of the preposition se”, to accurately report the idea that the preposition is argued to have two allomorphic realizations.
• Line 206, omit be.
• Beginning of section (either above or below (14)): throughout the paper, the author takes se to be the default and s- to be the marked allomorph of the SMG preposition, and uses these terms to refer to the forms in question. I think it would be good to make the point explicitly (it is left implicit that s- is marked, although on l.216 se is called default).

 

Section 5.2

• Line 353: add the before demonstrative.
• Footnote 10. I’m not sure the particular notation is helpful, since it could also mark ungrammaticality.

 

Section 5.3

• Line 403: a symbol is used [for linear adjacency] that is explained later in the text (l. 418).

 

Section 5.4

• Line 576, ex (48): is there a right square bracket missing?
• 21, last sentence seems incomplete (or whether should be whereas, but then the sentence is too long); reformulate.
• Line 627, ex (53): is there a right square bracket missing?
• Line 640, ex (55): is there a right square bracket missing?

Author Response

Thanks for the helpful comments and corrections! As previously, my replies are in-line below.

The author has adequately addressed the issues raised in my previous review. I provide a small list of truly minor suggestions regarding formulation below.

Abstract

    The abstract states at the outset that “The article argues that the prepositions asce ‘from’ and an ‘from’ in the south Italian Greek variety Greko, as well as the preposition se ‘in, to, into’ found in both Greko and Standard Modern Greek are allomorphs…”.
I would suggest reformulating the second part (“as well as the preposition se..”) in a way that makes reference to the two allomorphic realizations of the SMG preposition se (there is, I think, an absurd reading of the sentence above whereby allomorphy holds between se and asce and an).

        - Good point, I've reformulated the abstract in an attempt to avoid this issue (and ideally improve clarity/readability overall).
 

Section 4

    I would recommend reformulating the title to something like “allomorphic realizations of the preposition se”, to accurately report the idea that the preposition is argued to have two allomorphic realizations.
    
        - Reasonable suggestion, implemented
    
    Line 206, omit be.
        
        - Fixed.
        
    Beginning of section (either above or below (14)): throughout the paper, the author takes se to be the default and s- to be the marked allomorph of the SMG preposition, and uses these terms to refer to the forms in question. I think it would be good to make the point explicitly (it is left implicit that s- is marked, although on l.216 se is called default).

         - Thanks for the suggestion. I have slightly modified some of the formulations surrounding (14) to clarify that I do indeed take s- to be the marked allomorph.

Section 5.2

    Line 353: add the before demonstrative.
    
        - Fixed.
        
    Footnote 10. I’m not sure the particular notation is helpful, since it could also mark ungrammaticality.

        - I figured this should be clear from the discussion, but I can see how it could cause confusion. I've changed the symbol to # (which may have its own issues, but they may be less relevant here, since I am not discussing matters of felicity).
 

Section 5.3

    Line 403: a symbol is used [for linear adjacency] that is explained later in the text (l. 418).

        - The intention is for the discussion following the VEs to address these notational issues. I feel the textual proximity should be enough, but I've slightly amended the sentence introducing the VEs to point out that I discuss them immediately afterwards.
 

Section 5.4

    Line 576, ex (48): is there a right square bracket missing?
    
        - No, this follows the notation of Moskal & Smith's hypercontextual rules. I've added a footnote near the first occurrence to clarify this.
        
    21, last sentence seems incomplete (or whether should be whereas, but then the sentence is too long); reformulate.
    
        - Indeed, the sentence was garbled. I have reformulated the footnote (now footnote 22).
    
    Line 627, ex (53): is there a right square bracket missing?
    
        - No, see above.
        
    Line 640, ex (55): is there a right square bracket missing?
    
        - No, see above.