LINGUIST List 10.1185

Wed Aug 11 1999

Disc: Review of Giannakidou: Author's response

Editor for this issue: Andrew Carnie <carnielinguistlist.org>


Directory

  1. A. Giannakidou, Final version of response to Simons

Message 1: Final version of response to Simons

Date: Tue, 10 Aug 1999 16:20:25 +0200
From: A. Giannakidou <A.Giannakidoulet.rug.nl>
Subject: Final version of response to Simons

Response to Mandy Simons's review of Giannakidou 1998: 
Polarity Sensitivity as (Non)veridical Dependency (Linguist list Vol. 
10-1152). 

I would like to thank Dr. Simons (henceforth S) for taking the time
to read and comment on my book. I should caution, however, that it would be 
difficult for me to consider her posting to the Linguist list a fair or 
balanced evaluation of my 1998 work (henceforth G 1998). Rather, 
it strikes me as a misfocused critical remark on G 
1998, without any evaluation of the central claims of the work, 
concentrating instead on aspects pertaining to points peripheral to 
the proposed analyses. For this reason, I would like to respond below with
some brief comments hoping to remedy any misconceptions 
that may have been caused. 

There are two worrisome features of the review. The first is that there are
 no evaluative remarks on points that are central to my 
proposal; in fact, most of the central points are missed. This fact is 
 striking, if we consider the novelty of the approach I defend in G 
1998, the synthetic character of the work (in the sense that it tries 
to build on intuitions and ideas already distributed in various places 
 in the relevant literature), as well as the fact that some of the 
issues I raise have not been seriously dealt with, or have been 
problematic, in the previous literature on polarity (with some 
exceptions mentioned in G 1998). Let me summarize here the 
most prominent novel points for the sake of clarity: (a) that (non)- 
veridicality, and not just downward entailment or negation, can 
explain the distribution of Greek polarity items (PIs), (b) that PIs 
may be of various kinds and that (non)veridicality, unlike its 
predecessors, can provide the basis for a unification of polarity 
contexts within a language and crosslinguistically, (c) that a 
theory based on (non)veridicality should be understood as an 
extension or refinement of the previous theories based on negation 
and downward entailment and not in conflict with these, (d) that 
grammar should allow for negative rules in the form of anti- 
licensing, (e) that PIs have restricted distribution because of their 
'special' lexical semantics (the sensitivity issue, see also Israel 
1996), (f) that polarity raises the issue of semantic well- 
formedness: sentences with unsanctioned PI are ungrammatical 
because they are semantically ill-formed (i.e. they cannot be 
interpreted by the semantic component of the grammar and are 
thus ruled out; with some provisos regarding sensitivity to 
pragmatic constraints too), (g) that semantic licensing doesn't 
always correspond to a syntactic scope condition, and (h) that 
negative concord is quantificational and we thus need QR to 
interpret it, and finally (i) that for the licensing of weaker PIs like 
Greek 'kanenas' and 'any' we need an LF c-command condition, 
and not an s-structure condition, as it is traditionally thought. The 
weight and consequences of these points for the analysis of 
polarity in particular and, more generally, for the division of labor 
between syntax and semantics are obvious to anyone who has 
ever worked on polarity or is concerned with the syntax-semantics 
interface. One can argue, , of course, for or against these points 
on empirical and formal grounds, but S's review does nothing of 
this sort. Rather, it overlooks most of the above points and
does not offer evaluation of the ones that are seen 
(with the exception of the variation point, part of point (b) above). 
This suggests that the value of the core proposals in G 1998 is 
missed. 

The second cause for concern about the review is that it focuses on 
minor points concerning some allegedly obscure definitions and it 
presents as problems things that are not problematic. Let me 
 go through the putative problems to illustrate. 

(A) I cite first S's passage. Note that the first sentence contains 
the characterization "far more problematic", which immediately 
follows S's summary of what I say in chapter 1. No evaluation of 
the many novel points of chapter 1 is made, but the choice of "far 
more problematic" implicates, with no justification, that what has 
been said in chapter 1 was already problematic. 

>"This part of the discussion is far more problematic. In making her 
>central claims as to the semantics of PIs,Giannakidou invokes several 
>novel notions which I found to be inadequately defined. To illustrate 
>some of the difficulties, consider the presentation of the sensitivity 
>semantics of APIs. Giannakidou proposes that APIs are expressions of 
>existential quantification, but of a special kind. She calls them 
>"dependent existential quantifiers". These are initially defined as 
>in (1) (p.70, Sec.2.3.5). (The definition is altered slightly as 
>I cannot reproduce in this format the symbols used in the 
original.) 
> 
>(1)	An existential quantifier is dependent iff the variable x which 
 it 
>	contributes does not introduce a discourse referent in 
>	[the actual world]. 
> 
>One problem is how to understand the term "discourse referent". 
>Giannakidou is assuming here GSV's semantic framework, but GSV 
>do not themselves use this term. They distinguish variables, 
>which are syntactic elements; pegs, which are "formal objects" 
>with which variables are associated; and the objects in the 
>domain of discourse. In her presentation of GSV's framework, 
>Giannakidou says that "since pegs serve no purpose other than 
>being 'intermediaries' between variables and discourse referents, 
>they can be dispensed with" (p.28). I infer from this that 
>Giannakidou is using the term "discourse referents" to refer to 
>members of the domain. (This in itself is a confusing use 
>of terminology.) 
>	Accordingly, I am uncertain how to understand the definition in 
(1). 
>In GSV, the use of a quantifier adds the associated variable to the 
>variables that are in active use, introduces a peg, and associates the 
>variable with a peg. The members of the domain are not affected by 
>linguistic items. Perhaps then we should set aside Giannakidou's 
>decision to dispense with pegs, and replace "discourse referent" in 
>definition (1) by "peg". However, it remains unclear how to interpret the 
>definition, as we cannot talk of pegs which belong to one world or another, 
>pegs being formal objects in the theory. 
>	 
>	Giannakidou claims that the distribution of APIs follows from their 
>being dependent existential quantifiers. However, as the definition 
>of this notion is not adequate, it is impossible to evaluate this claim. 
>... 

In this passage, S discusses my treatment of the 
sensitivity semantics of a particular class of PIs. To start with, she 
misses the point that G 1998 is one of the very few in the literature 
that address the sensitivity issue seriously, and the only one that 
actually spells out the intuition by making a concrete proposal for 
lexical sensitivity features. Instead, she concentrates on the 
tangential issue of the proper understanding of discourse referents 
(should they be understood as numbers, as pegs, or as individuals, 
 elements of D). The prose surrounding the definitions in G 
1998 (p. 69-71), as well as the preceding discussion concerning 
binding (p.66-69) make it clear that discourse referents are 
understood, like in many other places in the literature, as elements 
of D, an option also mentioned but not considered by S. That this 
is the way I implement discourse referents in G 1998 becomes 
evident also in the claim that regular existentials assert existence. 
 Now, if discourse referents are individuals the idea is crystal clear. 
 The difference between the variables associated with PIs and 
'regular' quantifiers is sortal: the variable contributed by the PI 
quantifier cannot be assigned a value according to the standard 
procedure by means of the assignment function g, whereas the 
variable contributed by regular quantifiers can. Rather, for PI- 
quantifiers to be assigned a value, certain conditions must be 
satisfied (for instance the dependent reference requirement for 
affective PIs or the variation requirement for free choice items). 
These constraints on the value assignment constitute by and large 
 the sensitivity semantics of PIs and are responsible for their 
limited distribution. As far as I know, G1998 is the first attempt to 
actually make this point. One can argue for or against it, but it hardly 
seems possible to plead inability to assess it. 

Next consider the following passage:

(B). S: As further illustration of the difficulties posed by Giannakidou's 
>definitions, consider her proposal for the sensitivity feature of 
>subjunctive relatives (p.91, Sec.2.5.2). (I here use E for the 
>existential quantifier.) 
> 
>(3)	[Op DP+SubjRel VP] has a truth value iff it is not known 
whether the 
>	following is true: 	Ex [NP(x) & SR(x)] 
> 
>The principal difficulty here is the phrase "is not known", which fails to 
>specify who is required to not know whether the existential statement is 
>true. Giannakidou perhaps intends to treat the requirement as a 
>presupposition; this is consistent with the proposal that failure to 
>satisfy the condition results in a truth value gap. However, this case is 
>unlike the normal case of presupposition, which requires some proposition 
>to be part of the common ground. Here we seem to have an 
>anti-presupposition, which requires that the common ground *not* contain 
>the existential proposition. 

Here we have a very similar type of criticism. S's difficulty with 
respect to the sensitivity semantics of subjunctive relatives arises from
an inadequate understanding of what "it is not known" in my 
definition on p.91 means. Again, from the surrounding discussion on 
pp. 89-92, it is clear that "it is not known" means mostly "it is not 
known to the speaker" and less frequently also "it is not part of the 
common ground", i.e. of the mutual speaker-hearer knowledge. 
This lack of knowledge could be understood as an anti- 
presupposition, as S suggests (though there are cases problematic 
 for this claim which I cannot go into here), but this is not at all 
crucial or in any way threatening for the point made in section 
2.5.2, which seems to be missed: that the use of a subjunctive 
relative is regulated by this semantic/pragmatic condition. 

Next, consider this passage from the review:

(C) S: Giannakidou first considers what she calls an "absolute 
notion" of 
>veridicality (pp.106-110, Sec.3.1.3). There are again some 
>difficulties with definitions: Giannakidou defines veridicality 
>only for monadic propositional operators and dyadic truth 
>functional connectives, although she is interested in 
characterizing 
>the veridicality of predicates such as "manage", "believe" and 
>"start", which do not belong to either of these classes. She also 
>wishes to characterize non-embedded simple clauses such as 
"Theodora 
>saw something" as veridical environments, although their 
>representation does not involve a propositional operator to which 
>veridicality can be attributed. However, her discussion indicates 
>that she has in mind a fairly intuitive notion of veridicality which 
>distinguishes, for instance, between "know", "believe" and 
negation. 
>"Know" (or alternatively, a clause embedded under "know") is 
>veridical, by virtue of the fact that "a knows that p" entails 
>that p. "Believe", on the other hand, is nonveridical, as the truth 
>of "a believes that p" entails neither that p nor that not p. 
>Negation is an antiveridical operator, as the truth of a negated 
>sentence (trivially) entails the falsity of the clause embedded 
>under negation. 

Here, S has seriously misunderstood me.
Presumably, it is a comment on defining nonveridicality for 
monadic propositional operators and dyadic connectives, although, 
 according to S, what I am interested in is in fact "predicates that 
do not belong to this class^�. By this she refers to the propositional 
 attitude verbs. First of all, it is entirely inaccurate to suggest that I 
am only interested in the (non)veridicality of these verbs, as the 
attentive reader of G 1998 can see. Moreover, after I introduce 
definitions 1 & 2 on pp.106-107, seven of the eight examples I give 
do not involve a propositional attitude verb, but the more familiar 
monadic and dyadic operators, among others, modal particles, 
negation, disjunction, and temporal connectives (pp. 107-109). 
This is totally in accordance with the definitions. (Admittedly, ex. 
(24) p. 108 would have been better placed in the next section.) 
Second, the definitions I am employing are modelled after Zwarts 
1995 (in fact, as I state, (21), p.107 is Zwarts's), and the term 
propositional operator is used in the literature not only to refer to 
type <t,t> functions but also to refer to elements and 
constructions that embed propositions (see e.g. Gamut 1991, 
Vol.2, p.16 where examples of modal and propositional attitude 
constructions like 'it is possible that' and 'I know that' are given as 
representative of propositional operators). Note crucially that on p. 
112, where relativized (non)veridicality is defined in definition 6, the 
term 'monadic' is dropped, and reference is made only to 
propositional operators in the spirit of the above usage. Here, S 
has simply misquoted me:

>	The revised notion of veridicality is more or less as follows 
>(p.112,Definition 6). I have paraphrased slightly for simplicity.
>
>(4)	Relativized veridicality for propositional operators
>	If Op is a monadic propositional operator then:
>(i)	Op is veridical just in case the truth of [Op p] in a context c 
>	entails
>	that p is true at every world in some epistemic model in c.
>	Otherwise, Op is non-veridical.

S's "paraphrase" has introduced exactly the error that exercises 
her so much: the word "monadic" does *not* appear in the original, 
the corresponding portion of which I quote here:

G 1998: 112 Definition 6 (Relativized (non)veridicality).
"... i. A propositional operator Op is veridical iff ..."

In other words, at no point in Def 6 do I refer to the adicity of the 
operators involved; instead, the definition is general and applies to 
any propositional operator at all (which I do not define here; see 
any introductory text on modal logic). 
	It is obvious from the semantics I give (p. 110-117) that I follow 
the standard assumption that propositional attitude verbs denote 
relations between individuals and propositions, and to the extent 
that these verbs take propositions as complements they can be 
properly characterized as propositional operators. 

S also makes the following invalid claim: 
she states that unembedded affirmative past assertions, which I 
want to analyze as veridical, 'do not involve in their representation 
any propositional operator to which veridicality can be attributed'. 
This is simply false. Under standard assumptions, unembedded 
affirmative past assertions contain not just one but two 
propositional operators, both of which are veridical -- the assertion 
operator and the past operator. Discussion of the latter is given in 
G 1998, and more extensively in Giannakidou and Zwarts 1998.

(D) In the final section of the review S has problems with defining 
truth in a context relative to individuals. I understand her worries 
that this relativization may lead to a very weak notion of truth, but I 
believe this notion of truth is needed, especially when one is 
dealing with propositional attitudes. In insisting on this I am hardly 
making a radical or novel claim, rather I am following a long- 
standing tradition which, recently is reflected in Farkas's (1992) 
notion of 'individual anchor' and which, besides polarity, has been 
proven useful in dealing with issues like mood selection in 
embedding contexts.

It appears, then, that many of the problems that S's review raises are 
either based on misunderstandings or are simply peripheral to the 
central claims of the work, and could probably be accommodated 
in footnotes in a future edition.

There is one exception, and I will close with it. S is asking (and 
here I am slightly paraphrasing S's comment): why is it that 
epistemic models are the only relevant models for (non)veridicality? 
 It is true that I do not give a comprehensive answer in G 1998 and I 
 will not attempt to sketch one here, but raising this question is quite
interesting and gives food for thought.


References
Farkas,D.F. 1992a. On the semantics of subjunctive 
 complements. Romance Languages and Modern Linguistic 
 Theory, John Benjamins, 69-104. 
Farkas,D.F. 1992b. Mood choice in complement clauses. 
 Approaches to Hungarian 4, JATE Szeged. 
Gamut,L.T.F. 1991. Logic, Language, and Meaning. Vol 2. 
 Chicago UPress. 
Giannakidou,A. and F. Zwarts. 1998. (Non)veridicality constraints 
 on tense/aspect combinations with temporal connectives. Ms. 
 University of Groningen. Presented at the Workshop on Mood 
 and Tense, Bergamo, July 1998.
Israel,M. 1996. Polarity sensitivity as lexical semantics. 
 Linguistics and Philosophy 19: 619-666. 
Zwarts,F. 1995. Nonveridical contexts. Linguistic Analysis 25, 3-4: 
 286-312. 
 

**************************
Anastasia Giannakidou
Fellow, Royal Dutch Academy of Sciences 
Dept. of Dutch Linguistics
University of Groningen
Postbus 716
9700 AS Groningen
The Netherlands

email: giannakilet.rug.nl
phone: +31 50 3635857 (off.)
 +31 50 3119266 (home)
******************************
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue