[(※ In this Preface, Chapters 1, 2, 3, etc. of this volume will be referred to as Papers 1, 2, 3, etc., respectively.)
Paper 1: Hoji, Hajime (1995) "Demonstrative Binding and Principle B," NELS 25, pp.255-271.
Paper 2: Hoji, Hajime (1998) "Null Object and Sloppy Identity in Japanese," Linguistic Inquiry 29-1, pp.127-152.
Paper 3: Hoji, Hajime (1997) "Sloppy Identity and Formal Dependency," WCCFL 15, pp.209-223.
Paper 4: Hoji, Hajime (1997) "Sloppy Identity and Principle B," in H. Bennis, P. Pica, & J. Rooryck, eds., Atomism and Binding, Foris Publications, pp.205-235.
Paper 5: Hoji, Hajime (1998) "Formal Dependency, Organization of Grammar, and Japanese Demonstratives," Japanese/Korean Linguistics, vol.7, pp.649-677, CSLI Publications.
Paper 6: Hoji, Hajime (2003) "Surface and Deep Anaphora, Sloppy Identity, and Experiments in Syntax," in A. Barss, ed., Anaphora: A Reference Guide. Blackwell, Cambridge, pp.172-236.
Paper 7: Hoji, Hajime (2003) "Falsifiability and Repeatability in Generative Grammar: A Case Study of Anaphora and Scope Dependency in Japanese," Lingua, vol.113, No.4-6, pp.377-446.]

3. Papers 1-6
　　Papers 1-6 were written in the context briefly described above. The work in Japanese generative syntax around 1985 was concerned with how one might be able to express some "phenomena" in Japanese in the terms of the theory or theories being pursued at the time and what one might be able to say about the theory or theories on the basis of one's "findings" in Japanese. In the 1980s and the early 1990s, when a large portion of the generative research was concerned with "binding," researchers in Japanese generative syntax tried to see if there might be some "phenomena" in Japanese that could be expressed in the terms of the theory or theories being pursued at the time regarding "binding" and could suggest something interesting about the theory or theories.
　　The Binding Theory (as discussed in Chomsky 1981, among other places) was formulated so as to regulate the hypothesized co-indexation relation between two nominal expressions, on the basis of their hypothesized structural relation, by making crucial reference to the hypothesized features of [+/- anaphor] and [+/- pronominal]. One of the three binding conditions/principles applies to anaphors ([+anaphor, -pronominal]) and another to pronouns ([-anaphor, +pronominal]). When efforts were expended to explore how the Binding Theory might apply to Japanese and what theoretical contributions we might be able to make based on Japanese, the correctness was assumed of the conception of the Binding Theory in terms of [+/- anaphor] and [+/- pronominal] features and in terms of co-indexation. It was assumed, more in particular, that Japanese has an anaphor and a pronoun so that the Binding Theory is indeed applicable to the language under its standard conception.
　　By the time I started preparing Hoji 1990, I had been convinced that there are no expressions in Japanese whose distribution is subject to the binding condition regulating the expressions with the [+anaphor] feature, even if we allowed the subcategories of anaphors, distinguishing so-called local anaphors and non-local anaphors, and assumed that the binding condition in question applies only to the former. As to pronouns, I addressed in Hoji 1990: Ch. 6 the issue of how the child can learn that a given expression has the [+ pronominal] feature (and for that matter, the [+ anaphor] feature, as well). In part based on the considerations discussed there and in part based on the absence of clear effects in Japanese of the binding condition regulating the expressions with the [+ pronominal] feature, I came to pursue the hypothesis that Japanese does not have expressions that have the [+ pronominal] feature.
　　In Paper 1, I observed that BVA can arise at least in two distinct ways, and introduced the notions of Arg-binding and Dem-binding, arguing that the former is subject to the local disjointness condition of Binding Principle B, but the latter is not. I extended the idea to English, looking into cases where we seem to detect local disjointness effects even when the intended bindee is not a pronominal. On the basis of such observations, it is proposed in Paper 1 that so-called Binding Principle B regulates any category (as the dependent term) that does not have the formal [+anaphor] feature rather than regulating categories that have the formal [+pronominal] feature.
　　Turning to the sloppy-identity reading, I came to realize, while preparing Hoji 1990, that it was not an easy matter to demonstrate that the availability of the sloppy-identity reading in Japanese is subject to the c-command condition. More specifically, I came to realize that many "constructions" in Japanese allow a sloppy-identity reading, irrespective of the satisfaction of the c-command condition. In such "constructions," the sloppy-identity reading obtains even when what corresponds to the "sloppy pronoun" is an inherently "referential" one, such as a proper noun or an a-NP. Such sloppy-identity readings are thus most likely not based on an LF c-command-based dependency relation. This line of reasoning is supported by the independent observation that such "ellipsis constructions" also do not exhibit island effects, in sharp contrast to the other type of "ellipsis constructions" where the sloppy-identity reading seemed subject to (i) the c-command condition and (ii) the lexical condition that the "sloppy pronoun" must be a so-NP.
　　In the meantime, Otani and Whitman 1991 (O&W) appeared, in which the sloppy-identity reading in Japanese is discussed. I had been convinced by then, on the basis of the research reported in Hoji 1990, that what was considered in O&W as the sloppy-identity reading in Japanese is NOT regulated by syntactic conditions such as the c-command condition and that its (un)availability is subject to various pragmatic factors, and hence, cannot be regarded as constituting evidence for or against hypotheses about the grammar, if we adopt a categorical conception of grammar; see Hoji 2015: Chapter 3 for relevant discussion. The main goal of Paper 2 was to point that out.
　　As noted in Paper 2: Section 1, "[t]he main purpose of [Paper 2] is to demonstrate that the NOC [=Null Object Construction] in Japanese cannot be analyzed on a par with VPE [=VP Ellipsis] in English." At the end of its Section 2, Paper 2 states, "Having thus shown that both of the empirical bases for the VPE analysis of the NOC are invalid, I take it to be established that the NOC in Japanese cannot be analyzed as an instance of VPE in disguise, contrary to O&W." Since that was the main purpose of the paper, I could have stopped there. At the time of writing Paper 2, however, I did not have a clear understanding of the significance of the *Schema-based prediction in the terms of Hoji 2015 (see the Glossary provided at http://www.gges.xyz/hojiCUP/and Section 6.2 below), and I thought it was necessary to say something minimally coherent about the source of what Paper 2 calls "the sloppy-like reading" and the property of the "null object" that underlies the "sloppy-like reading."
　　Among the outstanding issues remaining in Papers 1 and 2 are: (i) whether we can show a clear correlation in Japanese between the effects of the c-command condition and those of the local disjointness condition, (ii) whether it is possible to make definite and testable predictions with regard to the availability of the sloppy-identity reading and have them supported by experimental results in a reproducible manner, and (iii) what formal mechanism underlies the local disjointness effects.
　　Papers 3-5 were written just about the same time. They address the issues just noted. Because of the page limit imposed on each of these papers, what could have been placed in one single and long paper were divided into the three papers, which resulted in some degree of redundancy among the papers, with regard to the discussion of the background issues and the general claims therein. An attempt is made in Paper 6 to synthesize the empirical results in Papers 3-5 as well as Paper 2.
　　One thing that I found unsatisfactory with the work reported in Paper 2 is that, while the paper successfully demonstrates something is possible, contrary to what is claimed in O&W, the paper does not demonstrate that something is clearly impossible due to the formal aspects of the grammar. In Papers 3-5 and Paper 6, I tried to make what would be called *Schema-based predictions in the terms of Hoji 2015. The desire to obtain robust informant judgments (about what is predicted to be impossible) led me to consider increasingly more involved sentence patterns and interpretive possibilities.
　　Papers 3 and 4, for examples, discuss the Mix-reading patterns in "comparative ellipsis" in Japanese. I got clear (enough) judgments on the relevant sentences myself, and so did my colleagues. But, I was concerned with how to substantiate our judgments, so to speak, by making further predictions and by obtaining informant judgments precisely in accordance with those new predictions; see the second paragraph in (4) below. What was crucially investigated was how the availability of the sloppy-identity reading is restricted by the LF c-command relation, the locality, and the lexical choices, all in accordance with the thesis pursued in Papers 3-6 that the sloppy-identity reading of a certain type is necessarily based on Formal Dependency (FD). It is in this context that Papers 3, 4 and 6 consider the Mix-reading paradigm. The ultimate test in this regard involves the local disjointness effects in the Mix-reading paradigm.
　　Paper 3 is a continuation of Paper 2 and, to a somewhat lesser degree, of Paper 1 as well. Its main points are:

(2)a.Principle B is a condition on Formal Dependency, rather than on co-indexation.
b.There are at least two types of sloppy identity readings and only one of them is based on Formal Dependency.
c.The so-called "interface between the Computational System and language use" contains the Formal Dependency System, as schematized in (i).
(i)

[The chart not provided here.]


d.Local disjointness effects that have been regarded as being due to Binding Condition B must be understood as arising from different sources, reflecting different components in the above diagram.
e.Kare can be marked [+Dep] in the terms of Hoji 2015.

　　(2a) is already suggested in Paper 1, but without empirical evidence. Paper 3's empirical evidence in support of (2a) draws from Heim 1992; see Paper 3: (8), (10), and footnote 4. The first half of (2b) is already demonstrated in Paper 2, where it is shown that the so-called Null Object Construction in Japanese (NOC) is not akin to English VP Ellipsis (VPE), with regard to the availability of the sloppy-identity reading. In Paper 2, the NOC was contrasted with "comparative ellipsis" in Japanese, which seems to share properties with English VPE. Paper 3 discusses the soo su ('do so') construction in Japanese and shows that it behaves like the NOC (and "comparative deletion," which is called the Non-CM-comparative in Paper 6, as opposed to "comparative ellipsis," which is called the CM-comparative in Paper 6). Paper 3 also elaborates on the main point in (2d), which is made in Paper 1, without specific reference to (2c-i).
　　Multiple sources of the sloppy-identity reading and those of local-disjointness effects, as discussed in Paper 3, should be understood along with multiple sources of BVA and those of DR. I wanted to obtain most robust generalizations with regard to these "phenomena" in Japanese. I had come to believe that we would have the best chance to do so if we focused on the informant intuitions that are crucially based on the (LF) c-command relation. What led to this belief includes my own experience over the years as a researcher-informant and the conceptual/theoretical reason addressed in Reinhart 1983. I thus tried to identify informant intuitions that are crucially based on the formal relation that is based on (LF) c-command.
　　The observation that led to (2e) was significant because it had been observed since the early 1980s (see Paper 1: Section 1 for some early references) that kare cannot be used as a bound variable. The validity of the empirical basis for (2e) as discussed in Paper 3, which involves the Mix-reading paradigm, however, has yet to be experimentally demonstrated, especially in a multiple-non-researcher-informant experiment in the terms of Hoji 2015. It is, however, interesting to note that Hoji et al. 1999 points out that it is possible for some speakers to have kare as B of BVA(A, B) even in the "reconstruction context," hence where it must be based on LF c-command. The relevant observation thus suggests that kare can be α of FD(α, β) at least for some speakers.
　　Paper 4 focuses on what formal relation Binding Principle B regulates. The evidence adduced there in support of (2a) involves the local-disjointness effects in the Mix-reading paradigm, as compared to the absence of the local-disjointness effects in the same "local context" in other "ellipsis constructions." As in the case of Papers 2 and 3, Paper 4 makes crucial use of "comparative ellipsis" in Japanese.
　　Paper 4 also suggests an account of why it recommended it in English clearly does not allow coreference while its Japanese counterpart readily allows it. This is closely related to (2d). I had been concerned with this issue for some time, starting with Hoji 1990. Some related discussion is also in Paper 1. As the exposition in Paper 1 may suggest, I came to be concerned with (2d) initially based on my observation that Japanese lacks local disjointness effects for coreference as observed in English (as in it recommended it, for example). It remains to be a challenge how to account for the clear effects of local disjointness for coreference in English and their absence in Japanese in such a way that the account makes predictions beyond local disjointness; see the second paragraph of (4) below. But the results of multiple-non-researcher-informant experiments in English and in Japanese seem to provide striking confirmation for the difference between the two languages in this regard. Hoji 2015 does not discuss the issue although the relevant experimental results are included at the website accompanying Hoji 2015, in the form of "raw data" for English and in the form of result charts for Japanese.
　　Paper 5 addresses the lexical condition on FD, along with the two structural conditions on it, the c-command condition and the anti-locality condition. The three conditions are empirically illustrated together and fairly systematically for the first time in Paper 5. Paper 5 considers only BVA, not the sloppy-identity reading, and also elaborates on (2c) and (2d).
　　Paper 6 was an attempt to grapple with the issues that had remained in Papers 2-5. I had realized, on the basis of the research that had resulted in Paper 1 and Papers 2-5, that the testability and reproducibility we had been able to attain was not nearly as robust as we wished. I thus tried to seek in Paper 6 a higher degree of testability (and reproducibility) with regard to the availability of the sloppy-identity reading. Around that time, I came to be concerned with the issue of testability more acutely than before, and the main methodological concern of Paper 6 was how we can pursue rigorous testability, not compatibility, in dealing with the sloppy-identity reading; see Appendix.
　　The general goal of Paper 6 is to illustrate how we can try to tease apart the contributions of the language faculty and those of the factors outside it. Following Hankamer & Sag 1976, Paper 6 recognizes two types of "ellipsis constructions" in Japanese and English, surface anaphora and deep anaphora, and classifies various "constructions" into one of these two types, on the basis of the operational tests that are designed in accordance with the hypothesized structural and lexical properties of the hypothesized formal object, FD; see below.