BINDING DEMONSTRATIVE PHRASES

DANIEL PLESNIAK, SCAMS 2021

1

MAIN CLAIMS

• 1. Being bound is not something that only happens to “pronouns”.

​

• 2. At least for some speakers, demonstrative phrases (e.g., ‘that professor’) can be bound.

​

• 3. Such binding is (a) cross-linguistically common, and (b) can be achieved in the same way as “regular” pronoun binding.

​

2

BINDING/BVA(X, Y)

• A rough definition (Hoji, 2015): BVA(α, β) is “The dependency interpretation […] such that the reference invoked by singular-denoting expression β co-varies with what is invoked by non-singular-denoting expression α.”

​

• An example: “Every boy loves his mother” with BVA(every boy, his) yields the reading that each boy loves his own mother, e.g., John loves John’s mother, Bill loves Bill’s mother, etc.

​

• “Binding” is also used to describe coreference relations, e.g., “John loves his mother” where ‘his’=‘John’s’.
• For purposes of time, I’ll not discuss those during the main part of the presentation; feel free to ask during questions.

3

DEMONSTRATIVE BINDING (I)

• Open the average paper on binding, and you are likely to see Y of BVA(X, Y) (the “bindee”) referred to as a “pronoun”.

​

• Demonstrative pronouns are rather well-studied when it comes to binding; lots of work in German, for example (see nice review in Hinterwimmer and Brocher, 2018)

​

• In semantics literature, there is a fair amount of discussion of the various quantificational properties of demonstrative phrases (Wolter 2006 is a seminal work), but there is comparatively less discussion in syntax.

​

4

DEMONSTRATIVE BINDING (II)

• Despite the pronoun-centric literature, it is generally accepted that in certain languages, “R-expressions” can be bound.
• Lasnik 1986 discusses a lack of certain Condition C effects in Thai, and Lee 2003 addresses BVA in particular with identical noun phrases in Zapotecan languages.
• Yet, as is something of a trend among such papers, in both cases, the newly observed pattern is argued to have a sort of “parochial” status; Lasnik 1986 argues that Thai simply has a differently set parameter from languages like English, while Lee 2003 argues that the R-expressions in question aren’t “real” NP’s, but just variables that get spelled out as an identical copy of their binder.

​

• But we can find (non-copy) bound R-expressions in English!
• (1) Every/each professor praised that professor’s student. With BVA(Every/each professor, that professor’s student)

5

PRECEDENCE ONLY?

• Still, the general tendency to find ways to render such examples as “second class citizens” has persisted.

​

• Hoji 1995 recognizes that sometimes binding relations can hold between two demonstrative phrases in English and Japanese, which is argued to be the result of a special operation called “Dem-Binding”, distinct from the standard “Argument Binding”.

​

• Ueyama 1998 builds on Hoji 1995, but also incorporated examples with full-fledged quantifiers as X of BVA(X, Y) (the “binder”).
• Ueyama 1998 argues that such cases are a subset of a general phenomenon of BVA being established under precedence, as opposed to via syntactic structure. That is, what matters is the binder coming before the bindee, not their structural positions.
• Purely c-command-based binding, in the style of Reinhart 1983, is argued to not be possible with demonstrative-phrase bindees.

6

VARIATION

• As it turns out, behavior of different bindees is subject to quite a lot of speaker-to-speaker variation.

​

• As we will be seeing, there are indeed some speakers for whom it seems demonstrative phrases must be bound under precedence. However, there are also speakers for whom they must be bound under c-command. And still others with different patterns!

​

• This is not at all unique to demonstrative phrases and can be found with pronouns.

7

EXAMPLE VARIATION IN “REGULAR” PRONOUNS

• In Korean, there are multiple ways to translate ‘his’ in ‘every professor praised his student’. We will focus on two:
• ku, deriving from a demonstrative, and casin, traditionally said to be a long-distance reflexive.

​

• Speakers generally accept sentences of the form “[every professor]_subj [ku/casin’s student]_obj praised” with BVA(every professor, ku/casin).
• In Korean: “모든 교수가 그/자신의 학생을 칭찬했다.” ([modeun gyosu-ga] [ku/casin-ui hakseng-ul] chingchan hessda)

​

• For many speakers, there is an asymmetry under scrambling, e.g., “[ku/casin’s student]_obj [every professor]_subj praised” with BVA(every professor, ku/casin) is judged * with ku but ok with casin.
• In Korean: 그/자신의 학생을 모든 교수가 칭찬했다. ([ku/casin-ui hakseng-ul] [modeun gyosu-ga] chingchan hessda])

​

• For those who have this pattern, ku seems to have a precedence requirement, while casin does not. But not everyone shares these judgements! Some allow scrambled binding of ku, for example.

8

GENERAL METHODOLOGY

• The data analyzed here is a portion of data gathered for my dissertation, which focuses on “dealing with” variation in binding judgements.
• Demonstrative phrases are not the focus of the dissertation, so the amount of data is limited, but it’s still quite revealing.

​

• The data was elicited from speakers of English, Korean, and Mandarin Chinese, in one-on-one ”controlled judgement elicitations”.
• We can consider these “experiments”, but we should be careful in their interpretation. The sample size is low, and the goal is not to establish what judgements are “typical” or might be most likely to generalize to the population. Rather, the goal is to check whether each person, as an individual, has a pattern of judgements that conforms to or contradicts the hypotheses in question.

​

• Participants judged whether various sentences could have various interpretations; most relevant for us were the items where it was asked whether sentences could have BVA(X, Y) with Y being a demonstrative phrase, and X preceding/not preceding Y and X c-commanding/not c-commanding Y.

9

SAMPLE ITEM

• Participants decide whether the the sentence can correspond to the (I)=referential reading, (II)=BVA reading, both, or neither.

​

• If participants found the sentence malformed, that was also noted.

10

CLASSIFICATION

• Consider the paradigm:

​

• There are basically four possible judgement patterns an individual could have:
• (I) An individual never accepts such BVA, no matter the configuration. Such an individual doesn’t seem to be able to bind demonstrative phrases at all. (Technically, they obey whatever negative generalization you like, just degenerately)
• (II) An individual accepts such BVA, but only in +Precedence cases. Such an individual person obeys Ueyama 1998’s generalization that binding of demonstrative phrases.
• (III) An individual accepts such BVA in the +C-command, -Precedence case(s), but still rejects it in the -C-command, -Precedence case(s). Such an individual violates the “precedence only” account of demonstrative phrase binding, and in fact shows sensitivity to c-command in demonstrative phrase binding.
• (IV) An individual accepts BVA in the -C-command, -Precedence case(s). Such an individual is apparently not obeying c-command or precedence-based restrictions on binding of demonstrative phrases.

11

A NOTE ON PRECEDENCE

• I am sure some would like to see evidence that �only c-command is relevant for BVA, and not �precedence, but this doesn’t appear to be true. �Consider the paradigm:

​

• The columns on the right are instances of “possessor binding” or “spec binding”, which have plagued c-command-based accounts of BVA since Reinhart 1983. Essentially, being a possessor, ‘every professor’ should not be able to enter into BVA(every professor, his) via c-command, yet many speakers accept (b).

​

• However, an extremely common pattern of responses (cross linguistically) is that participants accept BVA in (a)-(c), but not in (d). This is well explained under theories like Ueyama 1998’s; BVA(X, Y) requires either X to c-command Y or X to precede Y.
• There are some individuals for whom certain bindees resist precedence-based binding, but that does not contradict the basic hypothesis that either precedence or c-command are potential sources of BVA in general.

12

CLASSIFICATION (CONT.)

• To reiterate, four groups:
• (I) Those who can’t bind demonstrative phrases,
• Those who can bind demonstrative phrases…
• (II) …but must do so under precedence
• (III) ...and can do so with c-command alone (no precedence requirement)
• (IV)… and can do it “whenever” (no c-command or precedence apparently needed)

​

• Much of my dissertation is spent accounting for people of type (IV), many of whom have that pattern with binding “regular” pronouns too.
• These people are doing something “extra”, e.g., the operation commonly referred to in the syntax literature as “QR” (whether or not silent syntactic raising is really the best analysis of that operation).
• Crucially, when we bring in coreference-style binding as well as quantifier-to-quantifier scope considerations, we can very clearly identify these people via non-circular diagnostics.
• So we can primarily concentrate on groups (I)-(III)...

13

SAMPLE PARADIGM

• If you are a native speaker (or not!) and want to classify yourself on this particular paradigm:
• If you don’t accept any with BVA, type (I).
• If you accept (a) and/or (b) with BVA but nothing else, type (II).
• If you accept (c) with BVA but not (d), type (III), regardless of judgments on (a)/(b).
• If you accept (d) with BVA, type (IV), regardless of any other judgements.

​

​

​

​

​

• Judgments here vary a lot, albeit within certain predictable ranges.

14

RESULTS

• The table shows the results, �looking across all paradigms �in each of the three languages �considered:

​

• Descriptively most interviewed people could (at least sometimes), bind DemP, and the plurality could do it with c-command alone.

​

• As stated previously, we cannot per se generalize these proportions to the population, but we can certainly say: “while some individuals require precedence instead of c-command for binding DemP, it is not hard to find individuals who can bind DemP via c-command alone.”

15

NOTES ABOUT THE RESULTS

• The claim that demonstrative phrases can only participate in precedence-based binding, not c-command-based binding looks pretty falsified; it may be true for some people, especially some people in certain constructions, but it is not a general property of Universal Grammar.

​

• There do seem to be some, non-categorical, language-to-language sources of variation: a number of Mandarin Chinese speakers in particular rejected all attempts at DemP binding (and some of those who could accept it did so only in particular constructions).

​

• Even that is not unprecedented among “pronouns” though; for example, quite a few Mandarin speakers frequently rejected 他 ta ‘him’ in BVA, often wanting to use the reflexive ziji instead. This too is subject to inter-speaker variation.

​

16

A BASIC CONCLUSION

• It does not seem that (c-command-based) BVA is sensitive to a distinction between pronouns and phrases.

​

• A more plausible notion can be found in Hoji 1995 and Ueyama 1998, where one of the conditions on argument c-command-based-binding is that the bindee be “small enough” relative to the binder in terms of its semantic content; that is, to be able to bind a bindee, an individual must be able to effectively “suppress” the bindee’s content.
• This is relatively easy to do with pronouns, harder to do with demonstrative phrases (they normally have something like deictic content).
• This also ties in nicely with things ’the guy’ (sometimes discussed as epithets), as in ‘each engineer always forgets the guy’s password’, which is even harder to use as Y of BVA(X, Y), though not impossible for all speakers; not repeating the same head noun, it has even more semantic content that must be suppressed.

​

• Crucially though, “suppressing semantic content” is presumably not a syntactic operation, but something semantic/pragmatic/cognitive operation, suggesting that perhaps syntax itself imposes no particular restrictions on what can be bound.

17

FUTURE DIRECTIONS (I)

• I think it is, if nothing else, a good heuristic to start with the assumption that Y of c-command-based BVA(X, Y) can be, in principle, anything semantically appropriate, because that will allow us to highlight areas where that assumption breaks down.
• If we start out with restrictive hypotheses that it can only be, say, “pronouns”, and dismiss binding of anything other than a pronoun as “something else”, we risk missing out on some important patterns.

​

• One area things may break down in is with certain types of demonstratives.
• Like many languages, Japanese has two distal demonstratives, one of the form so- and one of the form a-. These combine with bound morphemes to form what are effectively pronouns (rather different from full-fledged DemP’s), e.g., soitsu/aitsu ‘that person’ (functions like ‘him’), soko/asoko ‘that place’ (functions like ‘it’).
• However, Hoji 2015 reports that only the so- type demonstratives can be Y of BVA(X, Y); the a- type must be referential.
• This alone is interesting, but the reality might be even more so.

​

​

18

FUTURE DIRECTIONS (II)

• Hoji (p.c.) notes that there is a small minority who can accept BVA(X, Y) with Y being an a- element. In fact, he has recently come himself to be able to accept it.

​

• However! So far there has not been found anyone with pattern type (III). That is, no one seems to accept asoko/aitsu as Y of BVA(X, Y) when BVA is clearly achieved via c-command.
• As noted, most do not accept it, but those who do seem to accept binding wherever, regardless of whether X c-commands Y.
• Thus there may be a property of a-type demonstratives that syntax in particular is sensitive to, when it comes to binding.

​

• While more work must be done in order to substantiate this phenomenon, this is precisely the sort of thing that would allow us to refine our hypotheses about the requirement imposed by syntax on Y c-command-based BVA(X, Y).

19

THANK YOU!

• References:
• Lasnik, H. 1986. On the necessity of condition C. In Essays on Anaphora. Boston: Kluwer Academics.
• Lee, Felicia. 2003. “Anaphoric R-Expressions as Bound Variables.” Syntax, 6 (1): 84–114.
• Hinterwimmer, S. and Brocher, A. 2018 An experimental investigation of the binding options of demonstrative pronouns in German. Glossa: A Journal of General Linguistics, 3(1), 77.
• Hoji, H. 1995. Demonstrative Binding and Principle B, NELS 25, pp. 255-271.
• Hoji, H. 2015. Language Faculty Science. Cambridge University Press.
• Reinhart, T. 1983. Anaphora and Semantic Interpretation. Chicago: University of Chicago Press.
• Ueyama, A. 1998. Two types of Dependency. Ph.D. Dissertation, University of Southern California.
• Wolter, L. 2006. That’s that; the semantics and pragmatics of demonstrative noun phrases. Ph.D. Dissertation, University of California at Santa Cruz.

​

20