DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY, D LAZARIDOU

Tags: DISTRIBUTIVITY, definite article, Ohio State University, domain restriction, collective interpretation, Mouton de Gruyter, University of Massachusetts, Amherst, Giannakidou, Anastasia Giannakidou, Angelika Kratzer, Etxeberria & Giannakidou, partial distribution
Content: DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
DIMITRA LAZARIDOU-CHATZIGOGA University of Cambridge/Queen Mary, University of London [email protected]
Workshop (Co-)Distributivity 2015 26 February 2015 CNRS Pouchet
0 Overview This talk concerns two distributive quantifiers in Greek, kathe `every/each' and o kathe `lit. the every/each', as illustrated below:
1. Kathe pedhi zoghrafise mia alepu. kathe child drew.3SG a fox `Every/Each child drew a fox.' 2. To kathe pedhi zoghrafise mia alepu. the kathe child drew.3SG a fox `Each child drew a fox.'
I argue that kathe and o kathe are associated with two different types of distributivity, kathe is associated with slack distributivity, while o kathe with rigid distributivity, building on Beghelli and Stowell (1997) and Tunstall (1998). In addition to that, departing from previous analyses of the quantifier o kathe (cf. Giannakidou 2004), I offer a new syntactic and semantic proposal for o kathenominals that is compositional and treats o and kathe as two separate constituents that contribute independently to the meaning of o kathe-nominals. The account crucially relies on the semantics of the definite article in Greek as proposed in LazaridouChatzigoga (2009).
1 Introduction
Prima facie, it is not straightforward to tease the meaning of kathe and o kathe apart, given that in some contexts they seem to receive the same interpretation.1 Thus, kathe
1 Omitting the definite article with kathe here does not lead to ungrammaticality. This might be the reason, for which the descriptive literature alludes to the emphatic nature of the construction (Tsopanakis 1994: 309, Holton, Mackridge & Philippaki-Warburton 1997: 311, 313, TsamadouJacoberger 2002:248). Thus, o kathe-nominals differ from other co-occurrences of the definite article with determiners in Greek like demonstrative noun phrases in that the definite article in o kathe seems optional. Demonstratives necessarily have the definite article, as we see below: i. Afto to forema ine kitrino/ To forema afto ine kitrino. this the dress is yellow/the dress this is yellow `This dress is yellow.' ii. *Afto forema ine kitrino/ * Forema afto ine kitrino. this dress is yellow/ dress this is yellow It should be further noted that I will not treat occurrences of o kathe N like o kathe ashetos ehi apopsi ya tin sotiria mas `lit. the kathe ignorant has an opinion for our rescue', because they seem to receive an interpretation similar to minimizers and to be associated to marked intonation. For an analysis based on modality see Margariti (2011).
1
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
pedhi `kathe child' and to kathe child `the kathe child' above can refer to a specific set of children known to the interlocutors. For the time being, it helps if we think of kathe as having roughly the same meaning as every and o kathe as each.2 Let us consider the following sentences out of the blue, with no visual/perceptual hints as to the identity of the entities referred to: kathe is fine, while o kathe is unacceptable. Thus, o kathe requires a specific set of entities to quantify over.
3. Kathe anakalipsi krivi apo piso tis hronia entatikis dhulias. kathe discovery hides from behind her years intensive work `Every discovery hides behind it years of intensive work.' 4. #I kathe anakalipsi krivi apo piso tis hronia entatikis dhulias. the kathe discovery hides from behind her years intensive work `Each discovery hides behind it years of intensive work.'
3 can be used to make a characteristic statement about discoveries in general, while 4 requires a scenario, where different discoveries are presented to the public:
5. [Context: A public presentation of three recent scientific discoveries is made in order to raise awareness on ongoing research] I kathe anakalipsi krivi apo piso tis hronia entatikis dhulias. the kathe discovery hides from behind her years intensive work I proti oloklirothike meta apo dheka hronia erevnon, the first was.completed after of ten years studies i defteri kseperase ta ikosi, eno i triti eghine efikti the second overcame the twenty, while the third was.made possible mesa se molis pente hronia. within in only five years `Each discovery hides behind it years of intensive work. The first one was completed after ten years of research, the second lasted more than twenty, while the third one was possible only after five years.'
Crucially, for 4 to be felicitous in the context of 5, each one of the three discoveries must be associated with years of intensive work. It is not sufficient that only one or two of the three discoveries is associated with hard work. In contrast, kathe allows for some slack in its interpretation: 3 could be true of a scenario, where the property is not attributed of each one of the discoveries.
2 Empirical facts
2.1 Morphosyntax
(O) kathe has the following morphological and syntactic characteristics: · Kathe is not inflected (rare for Greek), thus has no overt agreement with the noun it combines with for number (singular/plural), gender (masculine/feminine/neuter) and case (nominative/accusative/genitive). · The definite article, if present, bears the relevant phi-features. · It is ungrammatical to combine kathe with a plural noun.
2 Anagnostopoulou (1994) was the first to suggest that o kathe intuitively corresponds to English each. 2
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
· (O) Kathe cannot occur without a noun, that is, it cannot be used pronominally.
6. (O)
kathe musikos epekse ena traghudhi.
the.NOM.MASC kathe musician.NOM.MASC played.3SG one song
`Every/Each musician played one song.'
7. O rithmos (tis) kathe melodhias itan dhiaforetikos.
the rhythm the.GEN.FEM kathe melody.GEN.FEM was different
`The rhythm of every/each melody was different.'
8. Piso apo (to) kathe traghudi iparhi mia istoria.
behind from the.ACC.NEUT kathe song.ACC.NEUT exists a story
`There is a story behind every/each song.'
9. *Kathe ghates efaghan ena pontiki.
kathe cats ate.3PL a mouse
10. *I kathe ghates efaghan ena pontiki.
the.PL.FEM kathe cats ate.3PL a mouse
11. *Kathe efaghe ena pontiki.
kathe ate.3SG a mouse
12. *I kathe efaghe ena pontiki.
the.SG.FEM kathe ate.3SG a mouse
As seen above, the order of constituents within (o) kathe-nominals is strict. The only possibility is o > kathe > Noun, as the following two ungrammatical orders show:
13. *kathe o musikos kathe the musician 14. *kathe musikos o kathe musician the
Based on this order and the non-availability of ellipsis, it is predicted that (o) kathe will not be able to float, as we confirm below, a fact to which we return later on.
15. *I pinguini (o) kathe efaghan ena psari. the penguins the kathe ate.3PL a fish 16. *I pinguini efaghan (o) kathe ena psari. the penguins ate.3PL the kathe a fish
2.2 Basic Semantics
I treat kathe as a Generalized Quantifier with the following denotation (Barwise and Cooper 1981). Thus, kathe builds a GQ that abstracts over two properties:
17. kathe = P. [Q. [x: [P(x) Q(x)]]
2.2.1 Scope
Previous work (Tsili 2001) has suggested that o kathe has a greater tendency than kathe to receive wide scope. Nevertheless, I follow Giannakidou (2001) and argue that both kathe and o kathe have similar scope properties. Kathe and o kathe lend themselves easily to a wide scope reading, (18a), without excluding narrow scope,
3
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
(18b), and both of them can receive inverse scope, as in 19b, (example taken from Giannakidou 2001:695):
18. (O) kathe musikos tu ensemble epekse mia klimaka. the kathe musician the ensemble played.3SG a scale `Every/each musician of the ensemble played a scale.' a. x (musician(x) y (scale(y) play (x,y))) b. y (scale(y) x (musician(x) play (x,y))) 19. Ja tin pirosvestiki askisi stis 4 i ora, tha topothetisume enan paratiriti brosta {se kathe eksodho/stin kathe eksodho}. `For the fire drill at 4 o'clock, we will station an observer in front of every/each exit.' a. # x [observer (x) FUT y [exit (y) station (we, x, y)]] b. y [exit (y) FUT x [observer (x) station (we, x, y)]]
Given the subtle judgements required for scope assignment and the difficulty to obtain reliable scope judgements without experimental data, we will leave this issue open, claiming only that kathe and o kathe have similar scope properties and that they behave alike in contrast to other universal quantifiers in Greek like olos, as will be discussed in detail in the next section.
2.2.1 Type of predicate
· (O) kathe is not compatible with collective predicates like mazevome `gather' (Dowty 1987), in contrast to olos `all', so (o) kathe is a distributive quantifier:
20. Ola ta pedhia mazeftikan stin avli. all the children gathered in.the yard `All the children gathered in the yard.' 21. *(To) kathe pedhi mazeftike stin avli. the kathe child gathered in the yard
We will discuss this more in detail in the section on distributivity.
2.2.2 Generic interpretation
· Kathe can be generic, that is, it can refer to a kind and it can appear with kindlevel predicates (Carlson 1977), as shown below (see also Giannakidou 1999, 2004).
22. Kathe ghata ehi tesera podhia. kathe cat has four legs `Every cat has four legs.' 23. Ehi eksafanisti kathe silogiki prospathia. has disappeared.3sg kathe collective effort `Any (sense of) collective effort has disappeared.'
As observed by Giannakidou (1999), kathe and o kathe seem to differ along similar lines: 22 can be a claim about cats in general, whereas 24 cannot:
4
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
24. I kathe ghata ehi tesera podhia. the kathe cat has four legs `Each cat has four legs.'
3 Distributivity
In this section I turn to distributivity, which is the main locus of differentiation of kathe and o kathe. I will start by establishing some terminology and facts about English and will then move on to Greek. As has been observed (cf. Vendler 1967), every induces distributive interpretations whereas all allows either distributive or non-distributive interpretations:
25. Every boy built a raft. (i) different raft for each boy (ii) *built single raft as a team 26. All the boys built a raft. (i) different raft for each boy (ii) built single raft as a team
Let's establish some further terminology: · In a sentence that involves a distributive Q and an indefinite, distributivity can be defined as a binary asymmetric relation holding between a distributive key (D-key) and a distributive share (D-share), (terms coined by Gil 1992 adapted from Choe 1987). · In the sentence below, `every boy' acts as the D-key, while the indefinite `a raft' is the D-share. In this case, it seems that the D-share falls under the scope of the D-key and thus a different raft has been built by each boy.3
27. [Every boy]D-key built [a raft.]D-share I will discuss here two refinements of distributivity that could prove useful when discussing the Greek data.
o Refinement of distributivity proposed by Beghelli & Stowell (1997) o Refinement of distributivity proposed by Tunstall (1998) 3.1 Refinement of distributivity proposed by Beghelli & Stowell (1997)
Beghelli & Stowell (1997) propose a refinement of distributivity along two dimensions focusing on the differences between all, every and each.
3.1.1 Strong vs. weak distributivity
The first dimension depends on whether the NP can have a collective interpretation or not:
3 There seems to be some judgement variation here, according to which the collective reading is not totally ruled out for some speakers. Related to distributivity, see also Lasersohn (1998), who reviews another line of thought that posits an abstract `distributivity operator', which is optionally attached to predicates.
5
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
o strong distributivity, attested in NPs with every and each o pseudo-distributivity/ weak distributivity, attested in NPs with all. 28. Strong distributivity a. DQPs headed by each/every are Strong Distributors. b. Strong distributivity is obligatory. c. Strong distributivity can arise under an inverse scope construal, e.g. where the distributee/distributive share is in Spec of AgrS-P and the distributor is in Spec of AgrO-P. 29. Pseudo-distributivity (weak distributivity) a. Plural definite and indefinite GQPs (including QPs headed by all) are pseudodistributors. b. Pseudo-distributivity is optional. c. Pseudo-distributivity cannot arise under an inverse scope construal, e.g. where the distributee is in Spec of AgrS-P and the distributor is in Spec of AgrO-P.4
· Collective vs. distributive interpretation
For pseudo-distributors like all, distributivity is optional. This means that they are compatible with collective predicates (Dowty 1987) like gather that can only be applied to groups or sums. Strong distributors on the other hand like every and each cannot appear with such predicates, since strong distributivity is obligatory:
30. All the boys gathered in the hall. 31. *Every boy gathered in the hall. 32. *Each boy gathered in the hall.
· Modification by different
When the distributive share is modified by different, nominals headed by every or each enforce a distributive reading, whereas all cannot enforce a similar reading. In the examples below, every or each-nominals receive a reading, according to which each boy read a book that was different from the one the rest of the boys read:
33. Every boy read a different book. 34. Each (of the) boy(s) read a different book. 35. #ll the boys read a different book.5
· Inverse scope
Finally, nominals headed by every or each can assume the distributor function, when they appear with inverse scope construals giving rise to strong distributive readings, in contrast to all:
4 DQP stands for Distributive Quantifier Phrase, Spec for Specifier, AgrS-P for Agreement SubjectPhrase and AgrO-P for Agreement Object-Phrase. 5 The # symbol indicates non-availability of the intended reading. The same holds for all the examples in this section that have the # symbol. Note that the anaphoric reading, according to which `a different N' means `an N which is not identical to the one mentioned before', is irrelevant here. The sentence seems subject though to variable judgements by English speakers. 6
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
36. A (different) boy read every book. 37. A (different) boy read each book. 38. #A (different) boy read all the books. 3.1.2 Optional versus obligatory distribution The second dimension concerns every versus each: o optional distribution, as with every o obligatory distribution, as with each Beghelli and Stowell use five different syntactic environments in order to establish this dimension. · Floating and binominal constructions Every cannot participate in floating constructions or binominal constructions (Safir and Stowell 1988), while each occurs in both constructions. Floating and binominal constructions provide unambiguous distributive construals for sentences with definite plurals in the subject position, where a collective construal would otherwise be possible. Thus, 39 could be true in a scenario where the penguins ate a fish all together, whereas 40 excludes this reading: 39. The penguins ate a fish. 40. The penguins will each eat a fish. 41. *The penguins will every eat a fish. 42. The men saw two women each. 43. *The men saw two women every. · Collective universal construals In this environment when the universal construal is headed by every, the requirement for distributivity seems to be relaxed. Thus, a verifying scenario for 44 would be that the boys lifted the piano together as a team. Each cannot be used with a similar collective interpretation: 44. It took every boy to lift the piano. 45. *It took each boy to lift the piano. Every, at least, in this context, can serve as a non-distributive quantifier and we could refer to this as every's tolerance to some kind of summation (Zimmermann 1993:177), which will be crucial for our account of kathe, as we will see in the next section. · Particle almost The particle almost can qualify any quantifier or numeral designating a fixed quantity understood as the end point of a scale including all and every, but not each: 46. One boy ate almost {all the apples/every apple}. 47. *One boy ate almost each apple. 7
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
· Particle not The particle not can combine with a variety of quantifiers including many, every and all, it cannot combine with each: 48. Not all the boys ate an ice cream cone. 49. Not every boy ate an ice cream cone. 50. *Not each boy ate an ice cream cone. 3.1.3 Criticism of their refinement · `Obligatoriness' is used twice as a criterion in two different stages of their argumentation: they use "obligatory" distribution to refer to the incompatibility of every/each with collective predicates like surround as opposed to all. When they further discuss the differences between every and each, they argue that every is "optionally distributive", while each is "obligatorily distributive", given that every seems to allow for a collective interpretation under certain conditions. · For the time being we adopt it as a working hypothesis in order to discuss some further data. · Their characterizations are not sufficiently fine-grained and, thus, we will provide an alternative refinement of distributivity, building on their insights. 3.2 Kathe and o kathe in light of Beghelli & Stowell (1997) Based on the empirical facts presented earlier if we establish an analogy between every and kathe and each and o kathe, we expect both kathe and o kathe to be strong distributors (first dimension) and kathe to be optionally distributive, while o kathe obligatorily distributive (second dimension). 3.1.1 Strong vs. weak distributivity Both kathe and o kathe satisfy all the criteria of strong distribituvity, as opposed to olos `all'. · Collective vs. distributive interpretation · Distributivity is obligatory: (o) kathe is ungrammatical on a collective interpretation with a predicate like gather, whereas olos allows it: 51. Ola ta pedhia mazeftikan sto hol. all the children gathered.3PL in.the hall `All the children gathered in the hall.' 52. *(To) kathe pedhi mazeftike sto hol. (to) kathe child gathered.3SG in.the hall · Modification by dhiaforetikos `different'
8
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
· Turning to singular indefinite QPs modified by the adjective dhiaforetikos `different', the Greek data are parallel to the English data. Both kathe and o kathe enforce distributive readings, whereas olos does not:
53. (To) kathe pedhi akuse
ena dhiaforetiko traghudhi.
the kathe child listened.3SG a different song
`Every/each boy listened to a different song.'
54. #Ola ta pedhia akusan
ena dhiaforetiko traghudhi.
all the children listened.3SG a different song
· Inverse scope · (O) kathe-nominals in the object position can assume the distributor function in an inverse scope construal, whereas nominals headed by olos cannot:
55. Ena (dhiaforetiko) pedhi akuse
(to) kathe traghudhi.
a different child listened.3SG the kathe song
`A different child listened to every song.'
56. #Ena (dhiaforetiko) pedhi akuse
ola ta traghudhia.
a different child listened.3SG all the songs
In 55, the object (to) kathe traghudhi `the kathe song' can be construed as the D-key and the subject ena dhiaforetiko pedhi `a different child' can function as the D-share. In 56, the object nominal cannot give rise to an inverse scope construal. · Both kathe and o kathe are related to strong distributivity, while olos is related to weak distributivity · Turning now to the second dimension and the possible differences between kathe and o kathe, Tsili (2001) has argued that kathe is optionally distributive, while o kathe is obligatorily distributive, without nevertheless providing detailed discussion of the five relevant tests.
3.1.2 Optional vs. obligatory distribution
· Floating and binominal constructions Neither construction is possible with either kathe or o kathe, as we saw above, repeated here for convenience:6 6 O kathenas `lit. the everyone'seems to express something similar to floating and binominal each: i. Ta pedhia zoghrafisan dhio zoa to kathena. the children drew.3PL two animals the eachone `The children drew two animals each.' Notice that the definite article is obligatory here. A tentative explanation for this is that to kathena `the each one' here refers back to the nominal ta pedhia `the children' and in order to do so the definite article is required to signal that the set to kathena is quantifying over is a familiar one. Related to this, note that Greek here adopts again an overt one strategy, as noted above in footnote 8. Finally, Dimitriadis (2006) discusses distance distributors in Greek and claims that distributive apo `from/of' is the equivalent to binomial each, a construction that does not necessarily involve kathe: i. Kathe pedhi pire apo pente mila. kathe child took.3SG from five apples `Each child took five apples.' ii. Ta pedhia evapsan apo tria avgha. the children painted.3PL from three eggs
9
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
57. *I pinguini (o) kathe efaghan ena psari. the penguins the kathe ate.3PL a fish 58. *I pinguini efaghan (o) kathe ena psari. the penguins ate.3PL the kathe a fish
· Collective universal construals
Kathe can sometimes receive a non-distributive interpretation, that is, it tolerates some kind of summation, while o kathe-nominals cannot:
59. Piran kathe metro ya na kseperasti
i krisi.
took.3PL kathe measure for SUBJ overcome.3SG.PASS the crisis
`They took every measure to overcome the crisis.'
60. *Piran to kathe metro ya na kseperasti
i krisi.
took.3PL the kathe measure for SUBJ overcome.3SG.PASS the crisis
To further illustrate this characteristic of kathe in a different context, consider the following example based on Landman (2000), where kathe-nominals appear as objects of a verb like sindhiazo `combine', whereas o kathe-nominals cannot:
61. S'afto to mathima tha prospathiso na sindhiaso kathe theoria in this the lesson FUT try.1SG SUBJ combine.1SG kathe theory ya tin oristikotita. for the definiteness `In this class I will try to combine every theory on definiteness.' 62. *S'afto to mathima tha prospathiso na sindhiaso tin kathe theoria in this the lesson FUT try.1SG SUBJ combine.1SG the kathe theory ya tin oristikotita. for the definiteness
· Particle shedhon `almost'
Even though Tsili (2001:793) argues on the basis of 63 that o kathe-nominals cannot be modified by shedhon `almost', whereas kathe-nominals can, google data as in 64 show that the evidence here is inconclusive and that we would need a more refined discussion of shedhon before concluding anything on the basis of this test7:
63. Ena aghori efaghe shedhon {kathe milo/*to kathe milo.} a boy ate.3SG almost kathe apple/ the kathe apple `A boy ate almost every apple.' 64. I kiria katefthinsi [...] ine i analisi tis fisis ke tu proorismu the main direction is the analysis the nature and the purpose tis ghinekas - thema pu djatrehi shedhon to kathe tefhos. the woman topic that runs almost the kathe issue `The main direction [...] is the analysis of the nature and the purpose of woman's nature, a topic that traverses almost each issue.'
`The children decorated three eggs each.' 7 For an alternative account of almost see Amaral (2007) and the discussion therein. 10
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
· Particle ohi `not' Both kathe-nominals and o kathe-nominals follow the same pattern: in subject position, modification by ohi `not' is ungrammatical for both, whereas in object position it is acceptable for both: 65. *Ohi (to) kathe pedhi efaghe paghoto. not the kathe child ate.3SG ice cream 66. Kalo ine na apokalipsume tin alithia, ala ohi (tin) kathe leptomeria. good is SUBJ reveal.1PL the truth, but not the kathe detail `It is good to reveal the truth, but not every/each detail.' 3.1.3 Summing up The tests that B & S use to differentiate every from each do not give any conclusive results for kathe and o kathe. The only crucial difference is that kathe tolerates some kind of summation, while this is not possible for o kathe. This tolerance crucially interacts with the second refinement of distributivity proposed by Tunstall (1998), to which I turn in the next section. 3.2 Kathe and o kathe in light of Tunstall (1998) Tunstall (1998) argues that every and each require multiple, or distributive, event structures, where the members of their restrictor set are associated with a number of different subevents. What this amounts to is that in the case of every there is a condition that the event must be at least partially distributive and in the case of each there is a condition that the event must be totally distributive.8 She defines the conditions for every and each (Tunstall 1998:99 and 100 respectively) as follows: 67. The Event Distributivity Condition A sentence containing a quantified phrase headed by every can only be true of event structures which are at least partially distributive. At least two different subsets of the restrictor set of the quantified phrase must be associated with correspondingly different subevents, in which the predicate applies to that subset of objects. 68. The Differentiation Condition A sentence containing a quantified phrase headed by each can only be true of event structures, which are totally distributive. Each individual object in the restrictor set of the quantified phrase must be associated with its own subevent, in which the predicate applies in that object, and which can be differentiated in some way from the other subevents. The requirement for partial distribution of the event with every aligns with the observations made about every being optionally distributive. Nevertheless, 67's allusion to `at least two' subsets seems too strict, given that we need to accommodate every's tolerance of some collective interpretation. With this caveat in mind, let us
8 As Tunstall (1998) notes, B. Gillon (1987) and Schwarzschild (1996) have used the term `intermediate' to describe the event structures called here partially distributive. I follow Tunstall's (1998) terminology, because I find the term more descriptive and transparent. 11
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
consider the event structures associated with kathe and o kathe.9 Consider the following sentence, which could depict a large event of photograph-taking e comprised by smaller events:
69. I Ino fotoghrafise
kathe zoo tu zoologiku kipu.
the Ino photographed.3SG kathe animal the zoologic garden
`Ino photographed every animal at the zoo.'
70. I Ino fotoghrafise
to kathe zoo tu zoologiku kipu.
the Ino photographed.3SG the kathe animal the zoologic garden
`Ino photographed each animal at the zoo.'
In 69-70 the event of taking photos of all the animals at the zoo allows for different interpretations with kathe and with o kathe. Scenarios A and B depict two of the partial distribution possibilities, whereas scenario C depicts a total distribution scenario. Finally, scenario D depicts a collective interpretation.
71. scenario A (partial distribution)
zebra
monkey
e1
fox
giraffe
e
hippo
e2
72. scenario B (partial distribution)
zebra
e1
monkey
e2
e
fox
e3
giraffe
hippo
e4
73. scenario C (total distribution)
zebra
e1
monkey
e2
e
fox
e3
giraffe
e4
hippo
e5
74. scenario D (collective)
zebra
monkey
fox
e
giraffe
hippo
Example 69, with kathe, can be true in a setting where Ino took four photos at a really small zoo, with a total of five animals, and say, for instance, the giraffe and the hippo were photographed together, while the remaining three animals, the zebra, the
9 See also Matthewson (2000) for an account of the distributive element pelp'ala7 in St'бt'imcets (Lillooet Salish) that builds on reference to the event structure, as proposed by Tunstall (1998). 12
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
monkey and the fox were each photographed separately (scenario B above). This cannot be a possible scenario for 70 though, since each animal must be associated with its individuated subevent of photo-taking. Thus, the partially distributive scenarios A and B are OK for kathe, but are ruled out for o kathe-nominals, for which only scenario C of total distribution is acceptable. Finally, scenario D may be accepted for kathe, because of the tolerance to collective interpretation. It should be noted though that it is not a preferred reading, given that kathe is a strong distributor in the sense described above. Scenario D would be better represented by the quantifier olos `all'. In any case, it is not a possible scenario for o kathe.
3.3 Slack and rigid distributivity
The results of the previous discussion on distributivity, can be distilled into the following claims: · On the basis of Beghelli & Stowell I argue that both kathe and o kathe are strong distributors, in contrast to olos, which is a weak distributor. · I inherit from Tunstall (1998) the importance of event structures and the need to differentiate between partial and total distribution. I argue thus that kathe is a slack distributor, while o kathe a rigid distributor. · The difference between kathe and o kathe lies in the different requirements attached to them. Kathe can be associated with partial distribution and allows for some collective interpretation, while o kathe requires total distribution. Here I introduce a modification to Tunstall's conditions, which I further simplify for the current purposes:
75. Slack distributor A sentence containing a QP headed by kathe is usually true of event structures, which are at least partially distributive. Kathe usually requires that there be at least two distinct subevents in the event structure. This distributivity condition is relaxed under certain conditions, which allow for a collective interpretation. 76. Rigid distributor A sentence containing a QP headed by o kathe can only be true of event structures, which are totally distributive. O kathe requires all the subevents to be distinct in the event structure.
Thus, kathe usually requires that there be at least two distinct subevents (for every object that is acted upon in one subevent we normally need to find one other object that is acted upon in another subevent), whereas o kathe requires all the subevents to be distinct (for every affected object we must check that all other objects are in another subevent). The requirement placed on kathe allows for some tolerance to summation or collective interpretation, as we have seen above. Considering again examples 69-70, the difference between kathe and o kathe is made clear if we continue the discourse with the modification ala ohi ksehorista `but not separately', according to which the subevents are not totally distributed to each and every animal in the zoo. O kathe-nominals are rendered deviant, while the modification works with kathe, a fact that suggests that kathe's distributivity requirement is slack:
13
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
77. I Ino fotoghrafise
kathe zoo tu zoologiku kipu,
the Ino photographed.3SG kathe animal the zoologic garden,
ala ohi ksehorista.
but not separately
`Ino photographed every animal at the zoo, but not separately.'
78. I Ino fotoghrafise
to kathe zoo tu zoologiku kipu,
the Ino photographed.3SG the kathe animal the zoologic garden,
#ala ohi ksehorista.
but not separately
In 78 each animal must be associated with a distinct subevent of picture taking, whereas in 77 we see that kathe might describe a scene with a single event.
The distinction between slack and rigid distributivity can furthermore shed some light on some facts that have to do with adverbial kathe. Up to now we have focused on adnominal kathe, but let us turn to some data with adverbial kathe. Adverbial kathe is able to quantify over plural individuals in the sense of Link (1983). Thus, in the following example the minimum unit kathe quantifies over is the plural object formed by tris katikus `three citizens', while o kathe is ungrammatical:
79. Iposhethikan na fitespun ena dhentro ya kathe tris katikus. promised.3PL SUBJ plant.3PL a tree for kathe three citizens `They promised to plant a tree for every three citizens.' 80. *Iposhethikan na fitespun ena dhentro ya tus kathe tris katikus. promised-3PL SUBJ plant.3PL a tree for the kathe three citizens
When we have a plural individual like three citizens quantification is only possible over sums, with no access to the sub-parts of the plural individual. I argue thus that for the rigid distributivity to be satisfied, o kathe requires atomic individuals to quantify over, given that access to each element of the set is essential. The plural individual comprised by three citizens does not allow the distributivity to go to the atomic level, thus rendering o kathe ungrammatical in these cases. Kathe on the other hand is a slack distributor and does not require access to the atomic level, as it is able to quantify over a sum of individuals.10 I will argue in the next section that the rigid distributivity facts are related to the presence of the definite article in o kathe. The fact that o kathe is a rigid distributor will be related to one of the conditions the definite article is associated with, the count-as-unique condition. In the next section I motivate a general proposal for the definite article in Greek and I apply it to o kathe-nominals.
4 The count-as-unique account
4.1 Syntactic proposal
In this section I present the syntactic part of my proposal. More specifically, I will motivate the following two claims: a) o and kathe do not form one constituent and b)
10 Interestingly, we see that the same seems to hold for English. Every is able to appear in phrases like every two years, whereas each is not. 14
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
o in o kathe has the meaning of the canonical definite determiner in Greek. Here is the basic structure I give to o kathe-nominals:
81. DP 3
D
QP
g3
oQ
NP
the g
4
kathe musikos
kathe musician
The first claim, namely, that o is positioned higher up in the tree of the derivation of o kathe-nominals and does not form one constituent with kathe, follows if we make standard assumptions (cf. Alexiadou 2014), according to which D combines with the extended projection of N to form a DP. In the account proposed by Giannakidou (2004), repeated below, the syntactic relation between o and the noun in the NP remains unexplained. Thus, for instance it remains unexplained how agreement takes place, since there is no relation between D and the NP, even though the noun determines the number, gender and case features of D. In the case below, for instance, the definite article agrees with the noun fititis `student' in being singular, masculine and in the nominative case.
82.
QP
wo
Q-det
NP
ru
4
D
Q-det
fititis
g
g
student
o
kathe
the kathe
With respect to the second claim, demonstrative data can show that o is a true determiner in this case (for details see Lazaridou-Chatzigoga under review). I will come back to the syntax when I compare my account to Giannakidou's in the next section.
4.2 The semantic proposal: the count-as-unique proposal
· In this section I propose that the definite article in o kathe-nominals has the denotation of the canonical definite article in Greek, as developed in Lazaridou-Chatzigoga (2009b). · First, I summarize the motivation for the count-as-unique proposal that argues that the definite article o is associated with two conditions, weak familiarity in the sense of Roberts (2003) and the count-as-unique condition, inspired by Badiou (1988), and I further claim that the semantic contribution of the definite article is inert, that is, o denotes the identity function. · Then I show how this proposal works for o kathe-nominals and I provide a derivation for these nominals.
15
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
4.2.1 Background assumptions: Two conditions for o: weak familiarity and count-as-unique
Based on observations in the literature on English and other languages (Birner & Ward 1994, Poesio & Vieira 1996, Farkas 2002, Roberts 2003, Farkas & deSwart 2007, Lцbner 2011, Schwarz 2013), I argue that both notions are relevant for the definite article o in Greek based on two sets of data. The first set of data would be possible counterexamples to a proposal that relies only on familiarity: a) associative definites, as in 83, b) first mention definites, as in 84, c) some definites, as in 85, where the entity `the grant' is familiar, but not unique and d) focused definites, as in 86:
83. Aghorasa ena vivlio htes. O sighrafeas ine katalanos.
bought.1SG a book yesterday the author is Catalan
`I bought a book yesterday. The author was Catalan.'
84. I Maria herete me tin prooptiki na pai taksidi stin Ispania.
the Maria is happy with the perspective to go.3SG trip to.the Spain.
`Maria is happy with the perspective of a trip to Spain.'
85. [Context: Professors Antoniu and Mpela are rivals in the Greek
Department, and each of them has received a major research grant for next
year.]
#Ta ipolipa meli tu tmimatos ine enthusiasmeni me tin ipotrofia.
`The other members of the department are very excited about the grant.'
86. TO atomo pu tha mporuse na me pisi na ghrafto ston silogho
the person that FUT could to me convince to register.1SG to.the club
molis parethitike.
just
quit.PASS.3SG
`THE person who could have convinced me to join that club just quit
himself.'
The second set of data involves counterexamples to a theory that relies only on uniqueness: a) incomplete definite descriptions, as in 87, b) anaphoric uses of definites, as in 88, c) definites under QPs, as in 89, d) definites that refer to individuals that are typically non-unique such as body parts or institutions, as in 90-91 and d) definites that are not unique, but are nevertheless familiar in the relevant sense, as in 92:
87. To trapezi ine ghemato vivlia.
the table is covered books
`The table is covered with books.'
88. Htes espase ena potiri krasiu. To potiri itan poli akrivo.
yesterday broke.3SG a glass wine.GEN the glass was very expensive
`A wine glass broke yesterday. The glass had been very expensive.'
89. Apo kathe podhilato elipe
mia aktina apo tin rodha.
from every bicycle missed.3SG a spoke from the wheel
`Every bicycle had a spoke missing from the wheel.' (Roberts 2003)
90. I Maria htipise ashima sto heri.
the Maria got.hit.3SG badly on.the arm
`Maria got badly hit on the arm.'
16
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
91. Htes hriastike na pao sto nosokomio. yesterday needed to go.1PL to.the hospital `Yesterday I had to go to the hospital.' 92. [Context: In a room with three windows] Kani poli zesti. Mporis se parakalo na aniksis to parathiro? does much heat can.2SG you please to open.2SG the window `It's very hot. Could you please open the window?'
4.2.2 Weak familiarity
With respect to familiarity, I follow Roberts (2003) and argue that the relevant notion for definites in Greek is weak familiarity and I will be using this label to encompass uses, where strong familiarity is satisfied. The examples, upon which I decided that weak familiarity is also the relevant notion for Greek are the following:
93. Kitakse tin ghata!
look.2SG the cat
`Look at the cat!'
94. Dhen mu aresi to kokino.
not me.GEN like.3SG the red
`I don't like the colour red.' (Poesio and Vieira 1998)
95. Maria, pighene kathise
stin ghonia tu
dhomatiou.
Maria go.IMPER stand.IMPER in.the corner the.GEN room
`Maria, go stand in the corner of the room.'
96. Ya na pate sto ghrafio tis
kirias Smith, protino
to SUBJ go.2PL to.the office the.GEN Ms
Smith,
na parete to treno.
SUBJ take the train
`To go to Ms. Smith's office, I suggest taking the train.'
suggest.1SG
4.2.3 The count-as-unique condition
The redefinition of uniqueness as the count-as-unique condition was inspired by the ontological insights of the philosophical work of Alain Badiou (1988). Relying on Roberts' (2003: 308) theory of definite NPs, I proposed the following conditions for definite NPs (to be modified in the next section):
97. Conditions for definite NPs Given a context C, use of a definite NPi presupposes that it has as antecedent a discourse referent xi, which is: a. weakly familiar in C (understood as in Roberts 2003) and b. counted-as-unique among discourse referents in C in being contextually entailed to satisfy the descriptive content of NPi Notice that the definition above includes an existential presupposition. The presupposition does not refer to existence in the actual world, but to existence of a discourse referent. Thus, existence is relativized to the discourse level.
17
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
4.2.4 The contribution of the definite article
I argue here that the definite article denotes the identity function and that it is associated with the two conditions below, which are pragmatic conditions and are found in a higher node, which we could call DefP for Definiteness in the spirit of Lekakou & Szendri (2012)11. The DefP is phonologically null and it is the place, where the nominal gets its e-type denotation. O is placed on the head of the DP and is semantically inert.
98. [[o]] = P.P Condition A: weak familiarity Condition B: the count-as-unique condition
The main syntactic structure is the following:
99. DefP 3
Def
DP
g3
шD
NP/QP
o
4
the
...
In the case of a canonical definite, where o combines with an NP of type , it gives back a type . Furthermore, it satisfies weak familiarity and the count-asunique condition. The contribution of the definite article is not truth-conditional, since it does not alter the semantic type of the expressions it combines with. Focusing on the DP, this is the structure I propose for a canonical definite in Greek:
100.
DP
3
D
NP
g
4
<,> o musikos
the musician
Here is the full-fledged version of the denotation I propose for o in canonical definites and the precise interpretation I give to the count-as-unique condition:
101. [[o]] = P.P Given a Context C, use of a nominal o NPi presupposes that it has as antecedent a discourse referent xi, which is tied to the following two conditions: (a) Condition A: xi is weakly familiar in C
11 In their account of polydefiniteness and close appositives, Lekakou & Szendri argue that the Greek DP is not of type e, but rather of type and that the e-type denotation is derived above the DP level. They argue thus that two separate functional heads are projected in Greek instead of one D head. Definiteness (DefP) is phonologically null. I follow here their rationale. This move is also motivated by data that involve the definite article with the free choice item opjosdhipote, which I do not discuss here, but formed part of the initial motivation for the analysis (see Lazaridou-Chatzigoga 2009a,b). 18
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
(b) Condition B: xi is counted as unique among discourse referents in C in being contextually entailed to satisfy the descriptive content of NPi and in the sense of receiving the unique event in the event structure, in which the predicate applies in xi Using the above denotation with a canonical definite, we obtain the interpretation discussed below:
102. I ghata kimithike sto pataki. the cat slept.3SG on.the mat `The cat slept on the mat.'
Use of i ghata `the cat' is successful only if there is a discourse referent that is weakly familiar in the context of utterance (condition A) and if that discourse referent is unique in satisfying the descriptive content of cat and in receiving the unique event in the event structure supplied by the predicate (condition B). Thus, if there are two cats when the above utterance is uttered or if both of them are sleeping, the utterance will be infelicitous. If there is only one cat that is weakly familiar in the discourse, then condition B is trivially satisfied in the sense that the referent will be counted as unique and there will only be one event in the event structure of cat-sleeping. In the next section, I will show how this proposal accounts for o kathe-nominals.
4.2.5 The count-as-unique proposal for o kathe-nominals
O kathe-nominals satisfy the weak familiarity condition, as seen below:
103. To therapeftiko hapi pu apokalite Fermargo kiklofori se mikra roz
the healing pill that is.called Fermargo comes in small pink
dhiskia...pula to kathe hapi stin timi ton pente irlandhikon liron.
tablets... sells the kathe pill in.the price the.GEN five Irish
pounds
`The healing pill, which is called Fermargo, comes in small pink tablets. She
sells each pill for the price of five Irish pounds.'
104. To kathe pedhi skeftike mia leksi apo alfa.
the kathe child thought one word from alpha.
`Each child thought up a word (that began) with alpha.'
Recall that kathe is a slack distributor, allowing for partial distribution, whereas o kathe is a rigid distributor. What the definite article in o kathe does is add the condition that the distribution needs to be total. The members of the o kathe-set require each to be given its own separate subevent in the event structure, which is not allowed to overlap with any other member of the set. In order to satisfy the count-U condition, we need to have access to the individuals one by one. Thus, the rigid distributivity facts of o kathe are related to the count-as-unique condition associated with the definite article. Here is the full-fledged denotation for o in o kathe-nominals:12
12 Note that the proposal advanced here relies on the availability of type shifting, which has been established in the formal semantics literature since the work of Partee (Partee and Rooth 1983 and Partee 1987). The definite article denotes the identity function over a property, whose type may vary between for canonical definites and <,t> for o kathe. Thus, the identity function can be
19
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
105. [[o]] = P.P 106. Given a context C, use of a nominal o kathe NPi presupposes that it has as antecedent a set of discourse referents Xi, which is tied to the following two conditions: (a) Condition A: Xi is weakly familiar in C (b) Condition B: for all x Xi, x is counted as unique among discourse referents in C in being contextually entailed to satisfy the descriptive content of NPi and in the sense of receiving its own subevent in the event structure, in which the predicate applies in x Let us see how this works with an example like the following: 107. I kathe ghata efaghe ena pontiki. the kathe cat ate.3SG a mouse `Each cat ate a mouse.' In the case above, Xi is the set of contextually relevant cats. Condition A is satisfied if we establish a set of cats that is weakly familiar in the discourse. Condition B will be satisfied if for each of the cats of the set established via condition A, the following holds: a) they satisfy the descriptive content of NPi (i.e. they are cats) and b) they are associated with their own subevent in the event structure, in which each one of them ate a mouse. In a scenario for instance with three cats, Neko, Tora and Sakana, the above utterance would be felicitous if the set that includes these three cats is weakly familiar in the discourse. In order to satisfy Condition B, which is linked to the rigid distributivity of o kathe, we have to consider each cat in the set as associated with a separate subevent of mouse-eating until we exhaust all the available possible values. If Neko, Tora and Sakana each ate a mouse, then the sentence is true. I argue that o and kathe do not form a constituent. The definite article in the D head of the DP is an identity function that needs to satisfy weak familiarity and the count-U condition (placed higher up in the tree, in the phonologically null DefP). Focusing on the DP, I repeat here the structure for o kathe with the semantic type of each node:
108.
DP <,t>
3
D
QP <,t>
g
3
<,t>,,t>> o
the Q g
NP 4
<,<,t>> kathe musikos
kathe musician
polymorphic. Subsequently, D is flexible between a type <,> in the case of canonical definites and a type <,t>,,t>> in the case of QPs. I thank an anonymous reviewer for bringing this issue to my attention. 20
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
The first operation that takes place is the quantification introduced by kathe, which quantifies over the predicate musikos `musician'. Kathe denotes a function from sets into generalized quantifiers and thus it builds a GQ, converting a predicative type to an argumental type <t>. The GQ kathe musikos `every musician' denotes the set of all sets X that contain every musician. Then, the definite article combines with the QP of type <t>. Given that the definite article denotes the identity function, it gives back the same type. The weak familiarity and the count-asunique conditions reside in the DefP, as we have argued above, thus they are found higher up in the structure. Focusing now on the semantic derivation of o kathe-nominals, kathe builds a GQ that abstracts over two properties: 109. kathe = P. [Q. [x: [P(x) Q(x)]] Going back to example 2, repeated here for convenience, I give the following semantic derivation for the phrase to kathe pedhi `the kathe child': 110. To kathe pedhi zoghrafise mia alepu. the kathe child drew.3SG a fox `Each child drew a fox.' 111. kathe pedhi w,g = kathe w,g (pedhi w,g) = [P. [Q.x [P(x) Q(x)]]] ([z. z is a child in w]) = [Q. [x [[z. z is a child in w] (x) Q(x)]] = [Q. x [ x is a child in w Q(x)]] 112. to kathe pedhi w,g = (to) (kathew,g (child w,g)) = (to) ([Q. x [x is a child in w Q(x)]]) = [Q. x [x is a child in w Q(x)]] x is weakly familiar in C and x is counted-as-unique The definite article does not alter the semantics of the o kathe-nominal. The two conditions are added separately, as pragmatic conditions. The above example would then be felicitous only if there is a set of children familiar in the discourse and if each one of the children in the context drew a fox. Each child is then associated with its own subevent, until rigid distribution is satisfied. 5 A previous analysis
The main point of departure from the previous literature is that whereas the definite article in o kathe has been treated as different from the canonical definite article, I propose to give a uniform analysis to both o kathe-nominals and the canonical definite article as developed above. The second main difference concerns the rigid distributivity facts of o kathe, which had remained undiscussed in the literature. I argue that the analysis that has been put forward in the literature by Giannakidou (2004) can only partially explain the data. This analysis relies on domain restriction (DR henceforth). In the remainder of this section I will focus on the proposed lexical ambiguity of o, on the specific syntactic implementation and on the failure to account for the rigid distributivity facts. DR is the widespread phenomenon, according to which the domain of a quantifier can be restricted in the following sense: in 113 below, everyone does not quantify over all the individuals in the world, but rather over a contextually restricted
21
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
set of individuals, i.e. the dinner guests who had rhubarb pie for dessert (example from von Fintel 1994:33):
113. The dinner guests had rhubarb pie for dessert. Everyone developed a rash.
In fact, given that a specific set of people has been introduced beforehand in the discourse, we cannot interpret everyone as quantifying over all the individuals in the world. How we represent or encode DR in the grammar is currently under debate and opinions vary as to whether contextual restriction is part of the syntax/semantics or not and where exactly we should place the covert domain variables, on the nominal or on the quantifier.13 Giannakidou (2004) and, more recently, Etxeberria and Giannakidou (2010) place o kathe within the broader discussion on DR. While one line of proposals argues that DR happens on the nominal argument via covert domain variables at LF (Stanley and Szabу 2000, Stanley 2002), Giannakidou (2004) argues that we also need to allow for the possibility of overt domain restriction. Based on the behavior of o kathe, she argues that DR happens on the quantificational determiner (Q-det) itself and not on its nominal argument (see also Etxeberria 2005, Martн 2009 for similar proposals for overt domain restriction in the Basque definite determiner and the Spanish indefinite algunos respectively). Thus, o kathe is seen as an embedding of a quantifier under the definite article forming the complex Q-det o kathe. Giannakidou (2004:121) provides the following analysis and argues that what the definite article does is restrict the domain, providing the C(ontextual) variable, as seen in the formula below:
114. [QP o D + kathe Q-DET [ NP fititis N]] o kathe fititis = [ kathe (C)] (student) `each student'
[[o kathe]]= C P Q {x: C(x)=1 & P(x)=1} {x: Q(x)=1}
`each'
115.
QP
wo
Q-det
NP
ru
4
D
Q-det
fititis
g
g
student
o
kathe
the each
In Etxeberria & Giannakidou (2010) the analysis is replaced by a simpler one, according to which the definite article in o kathe is a modifier of the Q-determiner, yielding a Q-determiner with a contextually restricted domain. They explicitly state (Etxeberria & Giannakidou 2010:118) that this meaning of the definite article should be seen "as an additional meaning that the definite determiner can have in a given language".
13 Another line of proposals follows a situation-based view of domain restriction (Cooper 1995, Recanati 1996, Schwarz 2012, Kratzer, 2004), according to which situation arguments, rather than covert domain variables, seem to be responsible for implicit quantifiers. This is not the place to decide upon this issue, but see Kratzer (2004) for an overview. 22
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
According to the view defended in this paper, a significant drawback against both Giannakidou (2004) and Etxeberria & Giannakidou (2010) is that they assign to the definite article in o kathe a different function than its usual one. In their account, o in o kathe is a domain restrictor, while the definite article in regular definites maintains its semantics (whatever that is). Their account assumes a lexical ambiguity in the definite article in Greek. However, I have argued that the contribution of the definite article in o kathe amounts to the contribution of the canonical definite article in Greek, which is an identity function that is associated with two conditions, weak familiarity and the count as-unique condition. Giannakidou's proposal that o kathe is a complex Q-det is partly based on data like 116, which is argued to illustrate the non availability of o kathe in polydefinites:
116. *o kathe o fititis the kathe the student
According to her account, o kathe does not spread, because it does not build a DP, but rather a complex Q-det. Thus, if we do not have a definite in the first place, we do not expect o kathe to participate in polydefinites. The above ungrammaticality might though be explained on different grounds. As shown in 2.1, for 116 we merely have to allude to the ellipsis facts and the fact that o kathe-nominals must include an overt noun. The above example does not suffice to justify the claim that o kathe does not spread; rather the relevant examples are the ones I have discussed in section 2.1, which crucially involve adjectives. As far as the syntax of o kathe is concerned, I have argued that o does not form a constituent with kathe, given standard assumptions about o being in the D head of the DP and, furthermore, given that o and kathe have distinct semantic/pragmatic contributions, which I have shown above. The nature of the composition of D with Q in their account is not explained in detail, but it is described as an adjunction or incorporation procedure. It seems a rather ad hoc stipulation that is not motivated independently. Even if we abstract away from the specific syntactic implementation of Giannakidou's account, the main concern with her analysis is that it does not capture the contribution of the definite article in o kathe, because it does not crucially refer to the rigid distributivity facts. Giannakidou assumes that the definite article in o kathe contributes only DR (cf. Westerstеhl 1984), but we have seen that the importance of context is only part of the contribution of o. Thus, Giannakidou's account might be able to explain the weak familiarity facts, but it cannot explain the rigid distributivity facts. Going back to example 70, repeated here for convenience, her account cannot explain why for instance to kathe zoo `the kathe animal' would be unacceptable in a scenario of partial distribution:
117. I Ino fotoghrafise
to kathe zoo tu zoologiku kipu.
the Ino photographed.3SG the kathe animal the zoologic garden
`Ino photographed each animal at the zoo.'
If, according to Giannakidou, the only contribution of the definite article is domain restriction, then in a scenario with a contextually restricted set of animals available, the unavailability of o kathe if rigid distributivity is not satisfied remains unexplained.
23
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
6 Conclusion I have argued that kathe and o kathe are associated with two different types of distributivity, kathe is associated with slack distributivity, while o kathe with rigid distributivity, building on Beghelli and Stowell (1997) and Tunstall (1998). I offered a new syntactic and semantic proposal for o kathe-nominals that is compositional and treats o and kathe as two separate constituents that contribute independently to the meaning of o kathe-nominals. The account crucially relies on the semantics of the definite article in Greek as proposed in Lazaridou-Chatzigoga (2009). References Alexiadou, Artemis. 2014. Multiple determiners and the structure of DPs. John Benjamins. Amaral Matos, Patrнcia. 2007. The meaning of approximative adverbs: evidence from European Portuguese, Ph.D. dissertation, Ohio State University. Anagnostopoulou, Elena. 1994. Clitic Dependencies in Modern Greek. PhD dissertation, University of Salzburg. Badiou, Alain. 1988. L'Etre et l'йvйnement. English translation: Being and event. trans. O.Feltham, London, Continuum, 2005. Barwise, Jon, and Robin Cooper. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4: 159-219. Beghelli, Filippo, and Tim Stowell. 1997. Distributivity and Negation. In Ways of Scope Taking, ed. Anna Szabolcsi, 71-107. Dordrecht: Kluwer. Birner, Betty, and Gregory Ward. 1994. Uniqueness, familiarity, and the definite article in English. In Proceedings of the Annual Meeting of the Berkeley Linguistic Society, 93-102. Berkeley, CA. Berkeley Linguistics Society. Carlson, Greg. 1977. Reference to kinds. New York: Garland. Ph.D. dissertation, UMass, Amherst. Choe, Jae-Woong. 1987. Anti-quantifiers and a theory of distributivity. Ph.D. dissertation, University of Massachusetts, Amherst. Dimitriadis, Alexis. 2006. Distribution over Symmetric Events. In: Tsoulas, G. (ed.), Proceedings of the 7th International Conference on Greek Linguistics, York, UK. Dowty, David. 1987. Collective Predicates, Distributive Predicates, and All. In ESCOL '86: Proceedings of the Third Eastern States Conference on Linguistics, eds. Fred Marshall, Ann M. Miller, Zheng Sheng Zhang, 97-115. Columbus: Ohio State University. Etxeberria, Urtzi, and Anastasia Giannakidou. 2010. Contextual domain restriction and the definite determiner. In Context-Dependence, Perspective and Relativity, eds. Franзois Recanati, Isidora Stojanovic, and Neftalн Villanueva, 93-126. Mouton de Gruyter, Mouton Series in Pragmatics 6. Farkas, Donka. 2002. Specificity distinctions. Journal of Semantics 19:1-31. Farkas, Donka and Henriлtte de Swart. 2007. Article choice in plural generics. Lingua 117: 1657-1676. von Fintel, Kai. 1994. Restrictions on quantifier domains. Ph.D. dissertation. University of Massachussets, Amherst. Giannakidou, Anastasia. 1999. Affective dependencies. Linguistics and Philosophy 22: 367­421. Giannakidou, Anastasia. 2004. Domain restriction and the arguments of quantificational determiners. In Semantics and Linguistic Theory 14, 110-128. Cornell Linguistics Club: Ithaca, NY. Gil, David. 1992. Scopal quantifiers: Some universals of lexical effability. In Meaning and grammar crosslinguistic perspectives, eds. Michel Kefer and Johan van der Auwera, 303-345. Berlin: Mouton de Gruyter. Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. Ph.D. dissertation. Garland Press, New York. Heim, Irene, and Angelika Kratzer. 1998. Semantics in generative grammar. Malden, MA, Blackwell Publishers. Holton, David, Peter Mackridge and Irene Philippaki-Warburton. 1997. Greek: A Comprehensive Grammar of the Modern Language. London: Routledge. Kadmon, Nirit, and Fred Landman. 1993. Any. Linguistics and Philosophy 16.4: 353-422. Kamp, Hans. 1981. A Theory of Truth and Semantic Representation. In Formal Methods in the Study of Language, eds. Jeroen A.G. Groenendijk, Theo M.V. Janssen and Martin J.B. Stokhof, 277-322,. Amsterdam: Mathematical Centre. Kartunnen, Lauri. 1976. Discourse referents. In Syntax and Semantics 7, ed. James McCawley, 363-385. New York: Academic Press. Keenan, Edward. 1996. The Semantics of Determiners. In The Handbook of Contemporary Semantic Theory, ed. Shalom Lappin. 41-63. Blackwell. Kratzer, Angelika. 2004. Covert quantifier restrictions in natural languages. Talk given at Palazzo Feltrinelli in Gargnano June 11, 2004. Landman, Fred. 2000. Events and Plurality, Kluwer, Dordrecht. Lasersohn, Peter. 1998. Generalized Distributivity Operators. Linguistics and Philosophy 21.1: 83-93.
24
DISTRIBUTIVE QUANTIFIERS IN GREEK: SLACK AND RIGID DISTRIBUTIVITY
WORKSHOP (CO)-DISTRIBUTIVITY 2015
Lazaridou-Chatzigoga, D. (2007) Free Choice Items and definiteness: Evidence from Greek In Proceedings of Sinn und Bedeutung 11, E. Puig-Waldmueller (ed.), Barcelona: Universitat Pompeu Fabra, p.403-417. ______ (2009a). Greek Generic Noun Phrases Involving the Free Choice Item opjosdhipote and the Definite Article, Proceedings of the 2007 Workshop in Greek Syntax and Semantics at MIT, MITWPL 57 (MIT Working Papers in Linguistics), p.123-137. ______ (2009b). On definiteness and the co-occurrence of the definite article with other determiners in Modern Greek. unpublished PhD thesis. Universitat Autтnoma de Barcelona. Lazaridou-Chatzigoga, D. (under review) Distributivity in Greek: the co-occurrence of the definite article with the universal distributive quantifier kathe Lekakou, Marika, and Kriszta Szendri. 2012. Polydefinites in Greek: ellipsis, close apposition, and expletive determiners. Journal of Linguistics 48: 107-149. Link, Godehard. 1983. The logical analysis of plurals and mass terms: A lattice-theoretical approach. In Meaning, Use and Interpretation of Language, eds. Rainer Bauerle, Christoph Schwarze, and Arnim von Stechow, 302-323. Berlin: de Gruyter. Margariti, Anna Maria. 2011. Modality and quantification: their relation in respect with the "kathe-DP" case. Handout given at the 6th Athens Postgraduate Conference of the Department of Philology, National and Kapodistrian University of Athens, 13-15 May 2011. Martн Martinez, Luisa. (2009) Contextual Restrictions on Indefinites: Spanish algunos and unos, in A. Giannakidou and M. Rathert (eds.) Quantification, Definiteness and Nominalization, Oxford Studies in Theoretical Linguistics 22, Oxford University Press, 108-132. Matthewson, Lisa. 1999. On the interpretation of wide-scope indefinites. Natural Language Semantics 7: 79­134. Matthewson, Lisa. 2000. On Distributivity and Pluractionality. In Proceedings of SALT X, eds. Brendan Jackson and Tanya Matthews, 98-114. Ithaca, NY: CLC Publications. Matthewson, Lisa. 2001. Quantification and the nature of crosslinguistic variation. Natural Language Semantics 9: 145­89. Partee, Barbara H. and Mats Rooth. 1983. Generalized conjunction and type ambiguity. In Meaning, Use, and Interpretation of Language, eds. Rainer Bдuerle, Christoph Schwarze and Arnim von Stechow, 361-383. Berlin: Walter de Gruyter. Partee, Barbara. 1987. Noun phrase interpretation and type-shifting principles. In Studies in Discourse Representation Theory and the Theory of Generalized Quantifiers, eds. Jeroen Groenendijk, Dick de Jongh and Martin Stokhof, 115­143. Dordrecht: Foris. Poesio, Massimo, and Renata Vieira. 1998. A corpus-based investigation of definite description use. Computational Linguistics 24(2): 183-216. Recanati, Franзois. 1996. Domains of discourse. Linguistics and Philosophy 19:445-475. Roberts, Craige. 2003. Uniqueness in definite noun phrases. Linguistics and Philosophy 26, 287-350. Schwarz, Florian. 2012. Situation pronouns in Determiner Phrases. Natural Language Semantics, December 2012, Volume 20, Issue 4: 431-475. Schwarz, Florian. 2013. Two Kinds of Definites Cross-linguistically. Language and Linguistics Compass 7 (10), 534-559 Stanley, Jason, and Gendler Szabу, Zoltбn. 2000. On Quantifier Domain Restriction. Mind and Language 15: 219­ 261. Tsili, Maria. 2001. The quantificational phrases (o) kathe NP, (o) kathenas > universality and distributivity (In Greek). In Studies in Greek Linguistics 21, 783-794. Thessaloniki, Aristotle University of Thessaloniki.Tunstall, Susanne Lynn. 1998. The Interpretation of Quantifiers: Semantics and Processing. Ph.D. dissertation. University of Massachusetts, Amherst. Vendler, Zeno. 1967. Each and every, any and all. In Linguistics in Philosophy, ed. Z.Vendler, 70-96. Cornell University Press, Ithaca and London. Westerstеhl, Dag. 1984. Determiners and context sets. In Generalized quantifiers in Natural Language, eds. J. van Benthem, and A. ter Meulen. Foris, Dordrecht. 45-71. Zimmermann, Thomas Ede. 1993. On the Proper Treatment of Opacity in Certain Verbs. Natural Language Sematics 1(2). 149-179.
25

D LAZARIDOU

File: distributive-quantifiers-in-greek-slack-and-rigid-distributivity.pdf
Title: DISTRIBUTIVE QUANTIFIERS IN GREEK
Author: D LAZARIDOU
Author: Dimitra Lazaridou Chatzigoga
Published: Thu Feb 26 07:19:22 2015
Pages: 25
File size: 0.34 Mb


The Shape Shifter, 1 pages, 0.06 Mb

, pages, 0 Mb
Copyright © 2018 doc.uments.com