Internally-headed and doubly-headed relatives in Japanese – How are they related to each other?

Grosu & Hoshi (2019: 20), in their rejoinder to Kitagawa (2019), propose that apparent violations of island constraints in the so-called internally-headed relative clauses are accounted for by considering them as reduced doubly-headed relative clauses. This paper shows that this claim by Grosu and Hoshi is not empirically sustainable, and further that it misses the discourse function of doubly-headed relative clauses. A discussion of gapless light-headed externallyheaded relative clauses is also presented so as to identify how this construction type interrelates with internally-headed relative clauses and doubly-headed relative clauses.


Introduction
their theoretical interests in the typology of relative clauses started their characterization of Japanese IHRCs as "strange relatives of the third kind" first presented in Grosu & Landman (1998), citing how they are neither restrictive nor nonrestrictive. In a series of papers, including Grosu (2010), Grosu & Landman (2012), ), Landman (2016, and Grosu & Hoshi (2018), that thesis has taken a more concrete form.
The basic thesis they hold is that Japanese IHRCs involve a null operator movement cum lambda abstraction over the operator's trace and are, therefore, strictly island-sensitive; if the internal semantic head (IH) is isolated in a syntactic island, such movement operation must necessarily be blocked. This conclusion, Landman (2016: 5-6) asserts, is fully corroborated by , who present extensive discussion eliminating challenges, and showing "beyond doubt" that a Japanese IHRC violating an island constraint is "invariably illicit" (Landman 2016: 6). It is against this claim that Kitagawa (2019) offers some critical counterexamples, which violate the Complex NP Constraint (CNPC) and the Adjunct Island Constraint (AJIC) (e.g., Kitagawa 2019: 17-18, (34) and (39)). They are far from being "invariably illicit" but are subject instead to idiosyncratic variations in the native speakers' acceptability judgments ranging between totally acceptable, somewhat odd, and totally unacceptable.
It is with this backdrop that Grosu & Hoshi (2019) (henceforth G&H), in their rejoinder to Kitagawa (2019), make the following claim: "prima facie IHRCs with referential IHs are in fact a hitherto unrecognized variety" of doubly-headed relative clauses (henceforth "DHRCs") (G&H 2019: 2), and that "the prima facie allowable violation of island constraints in the apparent IHRCs are accounted for by the thesis that they are in fact not IHRCs, but homophonous 'reduced' DHRCs" (G&H 2019: 20).
The objective of this paper is three-fold: (i) to demonstrate that this claim by G&H is not empirically sustainable; (ii) to point out how G&H miss the import of the discourse function of DHRCs (as referred to in Kitagawa 2019: 25-27);and (iii) to clarify in what manner the three construction types -IHRCs, DHRCs, and gapless light-headed externally-headed relative clauses (EHRCs) -do, or do not, interact with each other.
The outline of the paper is as follows. In Section 2, I discuss issues raised by G&H which are subsidiary to their DHRC-masquerading-as-an-IHRC thesis. Section 3 offers evidence to suggest that their arguments to counter Kitagawa's (2019) "pro-head analysis of Japanese IHRCs" lack empirical validity. Section 4 focuses on the discourse function of DHRCs. Section 5 deals with issues relating to gapless light-headed EHRCs, including their interrelation with IHRCs and DHRCs. "Concluding words" are given in Section 6.

Preliminaries
In this section my main concern involves the following two issues: (i) clarification of Kitagawa's account against the claims of Grosu, Hoshi, and Landman referred to in Section 3 on the idiosyncratic variations of the IHRCs; and (ii) the issue of nonrestrictive relatives that Kitagawa (2019) adopts in his analysis of IHRCs.
On the first question, it is important to note that Kitagawa (2019) does not claim that IHRCs are impervious to syntactic islands; IHRCs are sensitive to them. What he suggests (Kitagawa 2019: 19-20) is that, with an IHRC like (1) even though the effect of syntactic islands such as Complex NP Constraint (CNPC) and Adjunct Island Constraint (AJIC) exists, a pragmatic "repair" mechanism may override the island effect for some native speakers of Japanese (recall that, as amply demonstrated in Kitagawa 2019, the acceptability of the IHRC constructions is idiosyncratic  (34)) applicability-at again marvel.at-past 'I again marveled at the wide-range applicability of that (seemingly) preposterous solution, which the engineer, who is now a big shot in an American company, proposed.' no student-nom that important.be hypothesis-acc kokusaikaigi-de happyoosi-te koohyoo-o hakusi-ta] node] international.conference-at present-ing popularity-acc gain-past because Yano-sensei-ga sukkari otikondesimat-ta] no [pro] i ] no syooryaku-nasi no purinto-ga Y-Prof-nom greatly get.depress-past no no omission-free no copy-nom kore des-u. 2 this be-pres (See Kitagawa 2019:18, (39)) 'This is a complete copy of the important hypothesis, concerning which Prof. Yano was so greatly depressed because his rival's student presented it at an international Conference and received favorable responses.' How is the repair done? It is accomplished by means of "coreference" not encoded in syntax (Reinhart 1983), which involves "referential properties" (Dechaine & Wiltschko 2002:420), taking place between the IH and the external head (henceforth "EH") pro of the IHRC. Having the IH with a specific referential property as the antecedent, the external head pro as an anaphoric 1 Kitagawa (2019: 17-18) reports that eighteen linguistically sophisticated native speaker informants were asked to evaluate the felicity level of sentence (1), the scale of acceptability being: (i) "OK" means that a given sentence is "perfectly acceptable"; (ii) "?" means "somewhat odd but more of less acceptable; and (iii) "*" stands for "not acceptable." For sentence (1) one responded with "OK," seven with "?," and ten with "*" (Kitagawa 2019:17, (34)). 2 Acceptability judgments by the same native speaker consultants for this sentence are as follows "OK": 3; "?": 7; "??": 1; "*": 7 (see Kitagawa 2019: 18, fn.23). definite is interpreted to be that unique referent. It is this structural context of an IHRC that permits such a coreference relation to be formed (Kitagawa 2019: 20).
What is predicted here is that the replacement of the referential antecedent by a clearly indefinite expression would render the IHRC to be incoherent. 3 This prediction is borne out, as exemplified by the infelicitous "indefinite" version of (1) (Kitagawa 2019: 19, (40) zituyoosei-ni imasaranagara kantansi-ta. wide.ranged applicability-at again marvel.at-past (Intended: 'I again marveled at the wide-range applicability of some solutions, which the engineer, who is now a big shot in an American company, proposed.') 3 A reviewer calls into question the implicit assumption that indefinite expressions headed by nanraka cannot be referential and hence cannot be coreferential with a definite pro, based on the acceptability of (i): i: Seibisi-ga nanraka-no kaiketuhoo-o teiansi-ta. pro habahiroi zituyousei-ga at-ta.
engineer-NOM some-no solution-ACC propose-past. wide.ranging applicability-NOM be-past Why (i) is more acceptable than (3) remains unclear, though one possibility is that pro can be dynamically bound in (i) but not in (3). 4 A similar acceptability discrepancy can be seen between the felicitous IHRC (i) (though subject to idiosyncratic variations in the acceptability; see Kitagawa 2019: 17, fn. 19), (36)) and its infelicitous indefinite version (ii) (Kitagawa 2019: 20, (44) suffering.be-pres no no seed-nom contain-polite-pres (Intended: 'Contained in this bag are some cotton seeds that do not (prematurely) germinate, for which the farmers must pay a huge amount of money every year to the company that developed them.') The acceptability scale of (i), according to Kitagawa (2019: 17, fn.19), is: one "OK"; 7 "?"; and 10 "*." (ii) *[[[ ajic [kyoosooaite no osiego-ga nanraka no kasetu-o rival no student-nom some of hypothesis-acc kokusaikaigi-de happyoosi-te koohyoo-o hakusi-ta] node] international.conference-at present-ing popularity-acc gain-past because Yano-sensei-ga sukkari otikondesimat-ta] no [pro]] no syooryaku-nasi no Prof. Y-nom greatly get.depress-past no no omission-free 's purinto-ga kore des-u. copy-nom this be-pres (See Kitagawa 2019:18, (39)) (Intended: 'This is a complete copy of some hypothesis, concerning which Prof. Yano was so greatly depressed because his rival's student presented it at an international and received favorable responses.') This sort of explication is altogether unavailable with a G&H type analysis. The pro-head analysis, in fact, is the only model at present that accommodates such a coreference phenomenon.

Relevancy Condition
Of course, successful interpretation of the IHRCs still needs to satisfy Kuroda's "Relevancy Condition" (Kuroda 1975(Kuroda -19761992). 5 For, as Huang (1984) and Hasegawa (1985) respectively comment on pro's versatility in interpretation in Chinese and Japanese, the interpretation of pro may involve any diverse and idiosyncratic "inference, context, and knowledge of the world" on the part of the reader's or the hearer's part (Huang 1984: 531). Nothing structural in Kitagawa's (2019) analysis forces the speech participants to interpret the EH pro having a coreference relation to the antecedent IH in IHRCs. Thus, Kuroda's Relevancy Condition is required to properly interpret any given IHRC in Japanese.
Conversely, it is one of the revealing properties of Japanese IHRCs that such a condition is required of the IHRCs.

Kitagawa's (2019) analysis of the IHRC as a nonrestrictive relative clause
Turning now to the nonrestrictive relative clause issue, let me first note how an IHRC gets interpreted in Kitagawa's (2019) analysis. Consider an IHRC example (4a): 5 Kuroda proposes a "Relevancy Condition" as follows: for an IHRC to be acceptable, "it is necessary that it be interpreted pragmatically in such a way as to be directly relevant to the pragmatic content of its matrix clause" (1975-76: 86, (6)). As Ito (1986: 114) puts it, such factors as "simultaneity, colocationality, intentional and physical connections" need to contribute to the Relevancy Condition. While descriptively adequate, the existence of the Relevancy Condition on IHRCs has not been given an adequate theoretical explanation to date.
In structural terms, the IHRC represented in (4b) is a nonrestrictive relative. To follow Kitagawa D] is an E-type pronoun. The Θ-Assigner is either a verbal predicate or a syncategorematic noun (Kuroda 1999: 421-423; e.g., an inalienably possessed noun such as sippo 'tail', kekkan 'defect'). Example (4a) can be analyzed as follows: (i) the IH kasetu 'hypothesis' is theta-marked by teiansi-ta 'proposed'; (ii) the external head pro is theta-marked by the syncategorematic noun kekkan 'defect'; (iii) the external head pro and the IH are linked via the E-type relation, which I indicate here by the use of co-indexation (not to be understood as representing a rigorous syntactic identity). In this operation, I assume that the presence of comp is transparent.
It is expected that G&H would object to the idea that Japanese IHRCs are nonrestrictive relatives, because, for them, Japanese IHRCs are "strange relatives of the third kind," neither restrictive nor nonrestrictive. G&H, however, let Shimoyama (1999G&H, however, let Shimoyama ( , 2001) take the lead, tracing her analysis that the IHRCs are analytically distinct from the restrictive relative clauses in denoting a proposition rather than a predicate; to quote G&H (2019: 4-5): she observes that such an analysis "makes IHRCs partly similar to nonrestrictive/appositive" externally-headed relative clauses" (EHRCs), "whose relative clause is also arguably analyzable as denoting a proposition." G&H note that, despite this assumed similarity, "Shimoyama does not propose to bring IHRCs and appositive EHRCs under a common analytical umbrella because of another property that IHRCs share with restrictive EHRCs, but not with appositive EHRCs." That property can be "neutrally described as integration" (into the matrix), i.e., as forming a single illocutionary unit with the matrix. Clauses that are not integrated into the matrix, in particular, appositives, can be suppressed without affecting the truth conditions of the matrix. With this summation, G&H (2019: 10) "conclude" that Kitagawa (2019) "has failed to make a convincing case for viewing the relative clause of an IHRC as nonrestrictive." What G&H fails to see is that the critical integration is achieved in Kitagawa's (2019: 7-9) by means of the anaphoric definite pro-head that relates to the IH in the relative. 6 It is assumed in this process that the IH of an IHRC is confined to a position internal to the relative clause.
6 A reviewer points out that an integrated pronoun can take an antecedent inside a non-integrated clause in cases of cross-sentential anaphora like (i) in English: i: John, who owns a large dog, is interested in animal rights. He treats it as he would a child.
One other minor point is noted before we move to the next section. G&H comment on Kitagawa's (2019) rejection of Shimoyama's (2001: chapter 3) "no referential IH" claim, although they state that "while discourse anaphora is not necessarily of the E-type variety…" they then accept Shimoyama's claim that the anaphoric relation in IHRCs "must be of the E-type, referential IHs being excluded" G&H (2019: 9). They state: Kitagawa (2019) notes that examples which violate this "no referential IHs" generalization (without violating islands) were also submitted to his eighteen consultants and turned out to exhibit a roughly comparable cross-idiolectal variation, a state of affairs for which he admits he has no satisfactory explanation" (the emphasis by the present author). But, Kitagawa (2019)

G&H's proposal
One of the key findings of Kitagawa (2019), as articulated earlier, is that such IHRC examples as (1) and (2) are far from being invariably illicit (see fns. 1 and 3) as Grosu, Hoshi, and Landman would have it, since they are subject to idiosyncratic variation judgments ranging between totally acceptable, somewhat odd, and totally unacceptable, thus disproving their assertion that native speakers would invariably consider them to be illicit "beyond doubt" (Landman 2016: 5-6).
The fact that sentences like (1) and (2) exhibit a systematic variation among speakers with respect to their acceptability status clearly indicates that Japanese IHRCs do not invariably involve the null operator movement envisioned by Grosu, Hoshi, and Landman.
Facing the grammatical status of these examples, G&H now add a new category to their typological inventory relating to Japanese IHRCs, namely, that there exists a hitherto unknown construction type called "maximally reduced DHRC," which is homophonous with the corresponding IHRC but impervious to island constraints. It is this "maximally reduced DHRC" homophonous with the IHRC, they claim, that exhibits "a systematic variation among speakers," not the real IHRC.
Here, the pronoun it in the second sentence is anaphoric on a large dog, which itself is contained in a non-restrictive, and hence non-integrated, relative clause. Why this English example behaves differently from the Japanese IHRC example remains unsolved.
c. The point noted in (5b) is possible because the EH of a DHRC is subject to the semantic and pragmatic redundancy requirement (G&H 2019: 14). The EH material in the scope of sono constitutes redundant information. This requirement is satisfied if the material at issue is less informative than the IH, but not if that material is more informative than the IH (G&H 2019: 13). The redundancy property of the EH material of a DHRC "in no way requires that all the explicit information provided by the IH be redundantly expressed in the EH" (G&H 2019: 18-19). d. "In view of this, note that the semantics of a DHRC will be unaffected if the redundant information in the EH is suppressed altogether, and if sono 'that' is replaced with a null definite D, which is independently needed for IHRCs. We suggest that such maximally 'reduced' DHRCs exist" (G&H 2019: 19).
This "reduced DHRC" thesis by G&H, however, is flawed on three distinct fronts, as described in the following section.

Empirical validity
To follow their theory, let me posit the DHRC versions of (1) and (2) as (6) and (7) respectively, marking the portions either suppressed or made phonologically null in accordance with G&H's proposal with a "double strikethrough." that preposterous solution-acc propose-past engineer-nom now moo Amerika no kaisya-de erakunattei-ru] sono kaiketuhoo i ] no habahiroi already U.S. 's company-in big.shot.be-pres that solution no wide.ranged zituyoosei-ni imasaranagara kantansi-ta. applicability-at again marvel.at-past (Intended: 'I again marveled at the wide-range applicability of that (seemingly) preposterous solution, which the engineer, who is now a big shot in an American company, proposed.') 's student-nom that important.be hypothesis-acc kokusaikaigi-de happyoosi-te koohyoo-o hakusi-ta] node] Yano-sensei-ga international.conference-at present-ing popularity-acc gain-past because Y-Prof-nom sukkari otikondesimat-ta] sono kasetu i ] no syooryaku-nasi no purinto-ga kore des-u. greatly get.depress-past that solution no omission-free no copy-nom this be-pres (Intended: 'This is a complete copy of the important hypothesis, concerning which Prof. Yano was so greatly depressed because his rival's student presented it at an international conference and received favorable responses.'

Whence comes no?
Note that one technical problem about G&H's suggestion can be identified immediately: lacking entirely in their "maximally reduced" account is the necessary presence of the particle no in IHRCs, for which G&H provide no recourse. In Kitagawa's (2019) model, the EH of the IHRC contains the pro (i.e., [ dp [ np pro i ] D]) acting as an E-type pronoun, the no in question being accounted for by the prenominal modification marker insertion rule in Kitagawa & Ross (1982), which is independently motivated (see fn. 2).
Grosu, Hoshi, and Landman fundamentally claim that Japanese IHRCs involve a null operator movement. Thus, the only "two parses" allowable for G&H are the following: one is an infelicitous IHRC parse that is strictly sensitive to the island constraints (Landman 2016: 6); the other is a DHRC parse that is completely felicitous because the "DHRCs are insensitive to the CNPC and AJIC" (G&H 2019: 19). The import of Kitagawa (2019) is that it is this prediction that is distinctly demonstrated to be empirically false, because IHRCs do exhibit idiosyncratic variations with respect to island constraints. So, G&H's "two parses" argument (5a, b) adds nothing to their effort to disprove Kitagawa's (2019) pro-head analysis of Japanese IHRCs.

G&H's "reduced DHRC" thesis, re. (5c, d)
G&H provide two sets of empirical evidence in support of their "IHRC as a 'reduced DHRC'" claim. The first set has to do with their claim (5c, d), cited below as (8a, b, c) (see G&H 2019: 14, (21a, b, c)). G&H (2019: 13) state that the EH material in the scope of sono in (8a, b) "constitutes redundant information" and that this is the reason why (8c) is odd, because here "a requirement for the felicity of DHRCs" -that the EH material is "less informative than the IH" in the sense spelled out in (5c) -is violated. And, sure enough, (8c) is "infelicitous when uttered out of the blue." The fact that (8c) is odd is thus critically important to G&H in introducing their (5a, b) claims.
For their claim (5b) to be valid, their redundancy requirement (5c, d) must necessarily predict that (8c) is infelicitous, since in the context of (8c) the EH akai ringo 'red apple' is in fact not redundant but richer than the IH ringo 'apple'. And it is indeed the case that (8c) is infelicitous.
But, note that the source of (8c)'s oddity need not be because the EH akai ringo 'red apple' is not redundant, but could instead be because the (red) apple when peeled is no longer red. That such is indeed a legitimate issue is demonstrated by the fact that example (8d)  J-top M-nom she-acc hate-pres that introvert-be, plain-be, pitiable-be, aikurusi-i} Ayaka]-o aisitei-ru. lovely-be Ayaka-acc love-pres 'Jun'ya loves that {introvert, plain looking, pitiable, lovely} Ayako who Masao hates.' With (9 a, b), G&H identify the source of the oddity of (9b) in the fact that the EH Ayaka is a proper noun, whereas the IH kanozyo 'she/her' is a pronoun; the EH proper noun is more informative than the IH pronoun, thus violating the redundancy requirement (5c). Note, however, that the oddity of (9b) need not be identified in this manner. I suggest that (9b) is odd because it violates the linear order strategy for pronominalization (i.e., the antecedent must precede the pronoun).

The discourse function of the DHRC
Let me first note a peculiar property of this construction type. As commented in Kitagawa (2019: 25-27), the DHRC is a stylistically awkward and esoteric construction type not normally heard in daily conversation. However, this stylistically awkward and esoteric type of the DHRC construction is still available, since cross-sentential anaphora with the indefinite antecedent is fine when a speech participant wishes to avoid ambiguity as to the identity of the EH. I am indebted to one of the reviewers of an earlier version of this paper for pointing this out.
With this premise, compare the IHRC example (12a)  ]-o syasin-ni mother-nom baby-to milk-acc give-pres no -acc photo-to tot-ta. take-past 'I took a picture of a mother giving milk to a baby.' b. DHRC-1: [[Hahaoya i -ga akago-ni miruku-o ataetei-ru] sono hahaoya i ]-o syasin-ni mother-nom baby-to milk-acc give-pres that mother no photo-dat tot-ta. take-past 'I took a picture of the mother giving milk to a baby.' c. DHRC-2: [[Hahaoya-ga akago i -ni miruku-o ataetei-ru] sono akago i ]-o syasin-ni tot-ta mother-nom baby-to milk-acc give-pres that baby-acc photo-da take-past sono.mono-ga kono e-ni-wa yoku torae-rare-tei-u. 'I took a picture of the baby that the mother is giving milk.' Likewise, while IHRC (14) below is difficult to interpret, as Landman (see 2016: 5, (7b); the grammatical judgment is Landman's) points out, but its DHRC version (14)  The oddity with these DHRC examples is due to the fact that here there is only one potential argument that could serve as its IH and EH. In such a case, there is nothing to disambiguate. In (16a, b), the DHRC construction is used "illegally," resulting in their odd status. These empirical facts thus legitimize the claim that the discourse function of DHRCs is disambiguation. And, if disambiguation by the DHRC format is necessary for meaningful discourse, then that format cannot be just redundant as claimed in G&H (2019, (7c, d)). 8 This claim is based on the behavior of DHRCs with overt EHs. If G&H's proposed maximally reduced DHRCs exist, I take the default assumption to be that they too should obligatorily play a disambiguation function. [Taro-ga zibun no musume i no hitori no susi-o kyaku-ni dasi-ta] sono T-nom self 's daughter 's one 's sushi-acc guests-to offer-past] that musume i ]-o kyaku-ga suguni home-ta. daughter-acc guests-nom right.away praise-past 'Taro offered to guests a plate of sushi that one of his daughters made, and the guests praised the (quality of) sushi right away.' All in all, G&H's suggestion cited in (5b) (i.e., the constructions such as (1) and (6) "allow two parses, one as IHRCs and one as reduced DHRCs") is not only empirically unsustainable but also fails to take into account the specific discourse function of DHRCs (i.e., disambiguation).
Accordingly, Kitagawa's (2019) critique of the basic claim advanced by Grosu, Hoshi, and Landman concerning Japanese IHRCs must stand unaffected.

DHRC's discourse function in the context of gapless light-headed EHRCs
In this section I show that the DHRC's disambiguation function is not limited to IHRCs but applies as well to gapless light-headed EHRCs. Grosu & Hoshi (2016: 6, (13)) note that the bracketed constituent in (18a) "is an unambiguous gapless light-headed EHRC," in which "the connection between the EH and the relative clause is always established by contextually or lexically licensed extensions, since in gapless EHRCs in general, there is not only no gap but also no nominal expression within the relative clause that can be construed as restricting the denotation of the complex DP" (here recast in our own format with pronominal pro). (18a) is an instance of a gapless light-headed EHRC, and (18b) is its DHRC version. [[Sally-ga orenzi-o sibottekure-ta] sono (oisisoona) (orenzi) zyuusu]-wa S-nom orange-acc squeeze-past that delicious.looking orange juice-top John-ga itadai-ta. J-nom have-past '(As for) the (delicious-looking) (orange) juice such that Sally squeezed oranges [to obtain it], John drank it.' As G&H note, the incontrovertible gapless light-headed EHRC felicitously allows adjectival modification of no. It is important to note that gapless light-headed EHRCs such as (18a) typically exhibit a unique characteristic about which Grosu, Hoshi, and Landman have so far made no comment, namely, that the requisite property of the EH pro is that it serves as a receptor -and not as an anaphor -for the salient entity to be contextually appropriate. It thus resembles a "part-whole anaphor" in Schwarz's (2009) sense (e.g., I got into a car. The steering wheel was broken. Car → steering wheel), which uses the weak definite article in German. In Schwarz's analysis, this type of bridging is non-anaphoric.
The EH pro of gapless light-headed EHRC (18a), therefore, is a non-anaphoric definite. Now, back to the discourse function of DHRCs,consider (19a) Adding to G&H's (20a), I provide (20b), which is parallel to G&H's (20a) except for the different modifying adjectival phrase preceding no. Note that this (20b) is reasonably felicitous. So, what is going on with (20a)? G&H here subscribe to the commonly held notion that the "pronominal no" projects a derogatory sense when applied to a human being. This is, however, a misconception, as pointed out in Kitagawa (2005: 12-18). The correct generalization should be given as is (21) (Kitagawa 2005: 18, (33)).

(21)
The sense of deference cannot be read into pro if it represents an arbitrary reference.
The reason (20a) is odd, therefore, is not due to the misuse of "pronominal no." For, the acceptability of the sentence would improve if a different modifying adjective that is pragmatically linked to the context is used as in (20b). The simple fact is that there is no such thing as "pronominal no." The no in question is generated in the position directly preceding the non-anaphoric definite pro discussed above by the pre-nominal modification (MOD) rule in Japanese (see fn. 2), providing an appropriate context for which (21) applies.
Predictably, then, the following gapless light-headed EHRC (cf. Kitagawa 2019:11, (19)) is felicitous, where the EH pro that is a non-anaphoric definite can, for instance, be interpreted as itai 'corpse' among other possibilities, the pro still retaining the sense of respect and love by the villagers: The DHRC version of (22) makes what the coffin contains more specific, and the two lexical items, sontyoo 'village chief' and nakigara 'corpse', maintain a "whole" and "part" relation in the sense of Schwarz (2009) referred to above.

Different types of pro's
What the discussions so far reveal is that there are two types of pro's involved, one appearing in the IHRC and the other in the gapless light-headed EHRC, as illustrated in the following: In IHRC (23a), the EH pro is an E-type pronoun, which, as discussed in detail in Kitagawa (2019: 12-14), is an anaphoric definite. In the gapless light-headed EHRC, on the other hand, it is nonanaphoric definite in the sense of Schwarz (2009 no hitugi-o kuruma-de kyookai-e hakon-da. end.up-past that village.chief 's coffin-acc car-by. church-to carry-past 'The village chief, whom the villagers loved and revered, was tortured and murdered by terrorists, and the villagers took the coffin containing him by car to the church.' For those to whom such an interpretation is possible, including the present author, the EH pro in IHRC (24a) is an E-type pronoun, which is an anaphoric definite. And the EH sono sontyoo in its DHRC version (24b) clarifies what the pro in (24a) stands for.
To reiterate, IHRCs, DHRCs, and gapless light-headed EHRCs may be structurally represented as in (25a, b, c) Two types of pros are involved: (26) a. The external head pro of IHRC (26a) is an anaphoric definite; and b. The external head pro of gapless light-headed EHRC (26c) is a non-anaphoric definite. Kitagawa (2019: 19) presents his analysis of IHRC as follows: (i) an IHRC requires a formal link between its IH and the EH pro; (ii) the IH isolated in a syntactic island makes the establishment of the formal link difficult; (iii) given the makeup of a Japanese IHRC as understood in the framework of the pro-head analysis as in Kitagawa (2019), a "repair" is possible based on the notion of "coreference" as defined in Reinhart (1983). The disambiguation of an IHRC by means of a DHRC has the effect of making this "repair" unnecessary precisely because the IHRC's EH pro is replaced by the lexical item in the DHRC, to disambiguate any possibly misinterpretation of the pro. This is the sort of operation that G&H's operator movement analysis cannot be equipped to do, the fact of which leads them to resort to the redundancy requirement route. G&H claims that "the prima facie allowable violation of island constraints involved with the apparent IHRCs is completely accounted for by the thesis that they are in fact not IHRCs but homophonous 'reduced' DHRCs" (G&H 2019: 20), because "DHRCs are insensitive to the CNCP and AJIC" (G&H 2019: 15).

Concluding words
The objective of this paper was three-fold. The first and foremost is to disprove G&H's (2019: 19) claim that apparent IHRC constructions, such as (1) and (6), "allow two parses, one as an IHRC and one as a reduced DHRC." The empirical validity of this claim is questioned in Section 3. The second objective is to point out G&H's misconception concerning the discourse function of DHRCs, critically reviewing in Section 4 their "redundancy" thesis repeated here as (27).
The EH of a DHRC is subject to the semantic and pragmatic redundancy requirement (G&H 2019: 14) and, hence, the semantics of a DHRC will be unaffected if the redundant information in the EH is suppressed altogether, and if sono 'that' is replaced with a null definite D, which is independently needed for IHRCs; "we suggest that such maximally 'reduced' DHRCs exist" (G&H 2019: 19).
Thirdly, this paper explicates the nature of the interrelationship among three construction types: IHRCs, DHRCs, and gapless light-headed EHRCs.
All together, these arguments point to the conclusion that G&H's critique of Kitagawa (2019) is unsustainable.