SINFO
INTE
CO
Faculty of Arts,
University of Ljubljana
Department of English &
Department of Comparative
and General Linguistics
SinFonIJA 8, the 8th
International Conference
on Syntax, Phonology and
Language Analysis
BOOK OF ABSTRACTS
Ljubljana, 24th-26th September 2015
SINFONIJA 8, THE 8TH INTERNATIONAL CONFERENCE ON SYNTAX,
PHONOLOGY AND LANGUAGE ANALYSIS: BOOK OF ABSTRACTS, LJUBLJANA,
24TH-26TH SEPTEMBER 2015
SINFONIJA 8, 8. MEDNARODNA KONFERENCA O SKLADNJI, FONOLOGIJI
IN JEZIKOVNI RAZČLEMBI: ZBORNIK POVZETKOV, LJUBLJANA, 24.-26.
SEPTEMBER 2015
Organizers: / Organizatorji:
Department of English & Department of Comparative and General Linguistics, Faculty of Arts, University of
Ljubljana
Oddelek za anglistiko in amerikanistiko & Oddelek za primerjalno in splošno jezikoslovje Filozofske fakultete
Univerze v Ljubljani
© University of Ljubljana, Faculty of Arts, 2015. All rights reserved.
© Univerza v Ljubljani, Filozofska fakulteta, 2015. Vse pravice pridržane.
Editors: / Uredniki:
Gašper Ilc, Frančiška Lipovšek, Tatjana Marvin, Andrej Stopar
Issued by: / Izdal:
Department of English & Department of Comparative and General Linguistics
Oddelek za anglistiko in amerikanistiko & Oddelek za primerjalno in splošno jezikoslovje
Published by: / Založila:
Znanstvena založba Filozofske fakultete Univerze v Ljubljani
Ljubljana University Press
For the Publisher: / Za založbo:
Branka Kalenić Ramšak, dekanja Filozofske fakultete
Branka Kalenić Ramšak, Dean of the Faculty of Arts
This publication is available online and free of charge. / Publikacija je dostopna na spletu in brezplačna.
CIP - Kataložni zapis o publikaciji
Narodna in univerzitetna knjižnica, Ljubljana
81’367(082)
INTERNATIONAL Conference on Syntax, Phonology and Language Analysis (8 ; 2015 ; Ljubljana)
Book of abstracts [Elektronski vir] / SinFonIJA 8 - the 8th International Conference on Syntax,
Phonology and Language Analysis, Ljubljana, 24th-26th September 2015 ; [organizers] Faculty of Arts,
Department of English & Department of Comparative and General Linguistics ; [editors Gašper Ilc ... et al.]. El. knjiga. - Ljubljana : Znanstvena založba Filozofske fakultete, 2015
Način dostopa (URL): http://www.anglistika.net/sinfonija
ISBN 978-961-237-768-7 (pdf)
1. Ilc, Gašper 2. Filozofska fakulteta (Ljubljana). Oddelek za anglistiko in amerikanistiko
3. Filozofska fakulteta (Ljubljana). Oddelek za primerjalno in splošno jezikoslovje
281294080
SinFonIJA 8, the 8th International
Conference on Syntax, Phonology and
Language Analysis
Book of Abstracts
Faculty of Arts, University of Ljubljana
Department of English & Department of Comparative and
General Linguistics
Ljubljana, 24th-26th September 2015
Table of Contents
1 Conference Details ............................................................................................................................................... 2
1.1 Conference Call................................................................................................................................................. 2
1.2 Invited Speakers ................................................................................................................................................ 2
1.3 Organizers ......................................................................................................................................................... 2
1.4 Scientific Committee ......................................................................................................................................... 2
2. Conference Programme ....................................................................................................................................... 3
3. Abstracts ............................................................................................................................................................. 7
The Intensional Profiles of Hungarian Deverbal (Proto-) Imperative Sentence Types ........................................... 7
Hungarian object agreement with possessed direct objects: An experimental approach ...................................... 12
Information structure constraints on the position of bare indefinites in Russian .................................................. 15
On the locality of movement and phases............................................................................................................... 19
Discontinuous Foci and Unalternative Semantics ................................................................................................. 21
Vocalic length as evidence for the incorporated-free particle distinction in Czech .............................................. 23
Parametric variation in the functional and the substantive lexicon ....................................................................... 27
Subjunctive obviation as a semantic failure .......................................................................................................... 28
On a use of German discourse particle denn in the antecedent of conditionals .................................................... 33
Strategies of participle agreement with conjoined subjects in Bosnian/Croatian/Serbian .................................... 38
Two positions for verbal modifiers: evidence from derived particle verbs ........................................................... 41
Recursion Restricted ............................................................................................................................................. 45
Upper bounded and un-bounded no more ............................................................................................................. 49
Experimental evidence for neg-raising in Slavic................................................................................................... 53
The Relationship between (In)Alienable Possession and Deverbal Nominalizers in Hungarian .......................... 57
Reversed Coordinated Multiple Wh-Questions in Japanese .................................................................................. 60
On answers to negative questions ......................................................................................................................... 64
Resumption in Slavic relative clauses ................................................................................................................... 67
Tense and Modality in the Nominal Domain ........................................................................................................ 71
Agreement-by-Correspondence with spreading .................................................................................................... 75
The connection between metaphony and unstressed vowel reduction in Italian dialects ...................................... 78
Alignment and locality in the typology of language games .................................................................................. 82
Floating Universal Quantifier as a Base-Generated Functional Head ................................................................... 86
In the South Slavonic Garden: Landscaping the Landscape of Arguments andNon-Arguments .......................... 90
Wh-in-situ in a multiple wh-fronting language...................................................................................................... 94
Complementizer doubling and clausal topics in Italo-Romance ........................................................................... 98
Locative expressions in Slovenian Sign Language (SZJ) ................................................................................... 101
Symmetry in Syntactic Theory:English and Polish NPN forms.......................................................................... 106
Types and degrees of vowel neutrality ................................................................................................................ 109
Possessives and the NP or DP? Wrangle ............................................................................................................ 113
Pragmatic abilities in bilinguals: the case of scalar implicatures ........................................................................ 117
Stating the Obvious: Unifying the restrictions on the subjects of imperatives and directive subjunctives ......... 122
Two (non-) islands in Slovenian: A study in experimental syntax...................................................................... 127
Adverbial Left-Branch Extraction and the Structure of AP in Slavic ................................................................. 131
Indexical demonstratives and identificational focus in Hungarian ..................................................................... 136
Emphasis and the syntax of particle verbs .......................................................................................................... 140
A compositional analysis of French -eur nominalsin combination with ambiguous adjectives .......................... 144
Events, individuals, their subparts, and two types of multiplicative adjectives in Polish ................................... 148
Lexical strata and vowel (dis)harmony: The Turkish transformation of a Balkan hypocoristic ......................... 152
Numeral phrases as subjects and agreement with participles and predicative adjectives.................................... 155
On the discourse properties of dedicated impersonal pronouns in episodic sentences........................................ 159
1
1 Conference Details
1.1 Conference Call
We are pleased to announce that the 8th Syntax, Phonology and Language Analysis conference
(SinFonIJA 8) will take place at the Faculty of Arts at the University of Ljubljana, in Ljubljana,
Slovenia, from 24-26 September 2015. SinFonIJA is a formal linguistics conference organized each
year by a different institution in the area of former Yugoslavia and Austria-Hungary. The theme of the
conference is wide and spans all areas of formal linguistics.
1.2 Invited Speakers
•
•
•
•
Željko Bošković, University of Connecticut, USA
Guglielmo Cinque, University of Venice, Italy
Peter Jurgec, University of Toronto, Canada
Penka Stateva, University of Nova Gorica, Slovenia
1.3 Organizers
•
•
•
•
Tatjana Marvin, Department of Comparative and General Linguistics, Faculty of Arts,
University of Ljubljana
Franja Lipovšek, Department of English, Faculty of Arts, University of Ljubljana
Gašper Ilc, Department of English, Faculty of Arts, University of Ljubljana
Andrej Stopar, Department of English, Faculty of Arts, University of Ljubljana
1.4 Scientific Committee
Gašper Ilc, University of Ljubljana; Tatjana Marvin, University of Ljubljana; Penka Stateva,
University of Nova Gorica; Moreno Mitrovic, University of Cambridge; Andreas Bluemel; Karl
Franzens, University Graz; Peter Jurgec, University of Toronto; Željko Bošković, University of
Connecticut; Sabina Halupka-Rešetar; Rajesh Bhatt, UMass Amherst; Markéta Ziková, Masaryk
University; Katalin É. Kiss, Research Institute for Linguistics of the Hungarian Academy; Luka Crnic,
The Hebrew University of Jerusalem; Remus Gergel, University of Graz; Nadira Aljović, University
of Zenica; Boban Arsenijević, University of Niš; Matic Pavlič, University of Ljubljana; Balazs
Suranyi, Research Institute for Linguistics of the Hungarian Academy of Sciences; Adrian Stegovec,
University of Connecticut; Dalina Kallulli, University of Vienna; Ljiljana Progovac, Wayne State
University; Frančiška Lipovšek, University of Ljubljana; Martin Prinzhorn, University of Vienna;
Branimir Stanković, University of Niš; Andrej Stopar, University of Ljubljana; Adam Szalontai,
Eötvös Loránd University; Rok Žaucer, University of Nova Gorica; Sašo Živanović, University of
Ljubljana; Mojmír Dočekal, Department of Linguistics and Baltic languages, Brno; Tobias Scheer,
Université de Nice; Marko Hladnik; Daniel Büring, University of Vienna; Iliana Krapova, University
of Venice; Andrew Nevins, UCL; Petra Mišmaš, University of Nova Gorica; Lanko Marušič,
University of Nova Gorica; Veronika Hegedus, Research Institute for Linguistics of the Hungarian
Academy of Sciences; Luka Szucsich, Humboldt University; Berit Gehrke, CNRS researcher, Paris
Diderot; Natasa Milicevic; Tanja Milicev, University of Novi Sad; Arthur Stepanov; Jonathan Kaye;
Anna Gazdik, Eötvös Loránd University, Budapest, University Paris VII.
2
2. Conference Programme
DAY 0 (Wednesday, 23/9/2015): Warm-up party: Emonska klet, Plečnikov trg 1 at 19.00
DAY 1 (Thursday, 24/9/2015), Room 2
8.00-9.00
Registration
9.00-9.30
Opening ceremony
Session 1 – Chair: Janez Orešnik
9.30-10.20
Guglielmo Cinque
(invited speaker)
10.20-10.45
Negin llkhanipour
10.45-11.05
Coffee break
Session 2 – Chair: Mojmír Dočekal
11.05-11.30
Olga Borik
11.30-11.55
11.55-12.20
Parametric variation in the functional and the
substantive lexicon
Tense and modality in the nominal domain
Information structure constraints on the position
of bare indefinites in Russian
Emphasis and the syntax of particle verbs
Andreas Trotzke and Stefano
Quaglia
Matic Pavlič
Locative expressions in Slovenian Sign
Language (SZJ)
12.20-14.00
Lunch break
Session 3 – Chair: Marcel Den Dikken
14.00-14.25
Marijana Marelj
In the South Slavonic garden: Landscaping the
landscape of arguments and non-arguments
Two positions for verbal modifiers: evidence
from derived particle verbs
On answers to negative questions
Experimental evidence for neg-raising in Slavic
14.25-14.50
Éva Dékány and Veronika
Hegedűs
14.50-15.15
Hana Gruet-Skrabalova
15.15-15.40
Mojmír Dočekal and Jakub
Dotlačil
15.40-16.00
Coffee break
Session 4 – Chair: Martin Krämer
16.00-16.25
Pavel Caha and Markéta Ziková
16.25-17.15
17.30-19.30
19.00
Vocalic length as evidence for the incorporated–
free particle distinction in Czech
Agreement-by-Correspondence with spreading
Peter Jurgec
(invited speaker)
Poster session
Conference buffet dinner
DAY 1 (Thursday, 24/9/2015) Parallel session – Phonology, Room 13
Session 5 – Chair: Mary Ann Walter
14.00-14.25
Mary Ann Walter
Lexical strata and vowel (dis)harmony: The
Turkish transformation of a Balkan hypocoristic
14.25-14.50
Martin Krämer
The connection between metaphony and
unstressed vowel reduction in Italian dialects
14.50-15.15
Péter Rebrus and Miklós
Types and degrees of vowel neutrality
Törkenczy
15.15-15.40
Martin Krämer and Barbara
Alignment and locality in the typology of
Vogt
language games
3
Poster Session (17.30-19.30)
Yasuyuki Fukutomi
(alternate)
Lucas Tual (alternate)
Marcin Wągiel (alternate)
Sarah Zobel (alternate)
Gabor Alberti, Judit
Kleiber and Veronika
Szabó
Irina Burukina
Mojmír Dočekal
Wiktor Pskit
Enikő Tóth and Péter
Csatár
Reversed coordinated multiple wh-questions in Japanese
A compositional analysis of French -eur nominals in combination with
ambiguous adjectives
Events, individuals, their subparts, and two types of multiplicative
adjectives in Polish
On the discourse properties of dedicated impersonal pronouns in
episodic sentences
The intensional profiles of Hungarian (proto-) imperative sentence types
Russian possessive pronouns typology: Overt and invisible elements
On the non(exhaustive & contrastively-focused) constituent negation in
Slavic
Symmetry in syntactic theory: English and Polish NPN forms
Indexical demonstratives and identificational focus in Hungarian
4
DAY 2 (Friday, 25/9/2015), Room 2
Session 6 – Chair: Daniel Büring
9.00-9.50
Penka Stateva
(invited speaker)
9.50-10.15
Nicola Munaro
10.15-10.40
Pragmatic abilities in bilinguals: the case of scalar
implicatures
Complementizer doubling and clausal topics in
Italo-Romance
Floating universal quantifier as a base-generated
functional head
Dong-Yi Lin
10.40-11.00
Coffee break
Session 7 – Chair: Penka Stateva
11-00-11.25
Daniel Büring
11.25-11.50
Eva Csipak and Sarah Zobel
11.50-12.15
12.15-12.40
Discontinuous foci and unalternative semantics
On a use of German discourse particle "denn" in
the antecedent of conditionals
Subjunctive obviation as a semantic failure
Stating the obvious: Unifying subject restrictions
in matrix and embedded directive clauses
Francesco Constantini
Adrian Stegovec
12.40-14.30
Lunch break
Session 8 – Chair: Pavel Caha
14.30-14.55
Jacek Witkoś and Dominika
Dziubała-Szrejbrowska
14.55-15.20
Nermina Čordalija, Nedžad
Leko, Amra Bešić, Ivana
Jovović, Nevenka Marijanović,
Lidija Perković, Midhat Šaljić,
Dženana Telalagić and Tamara
Butigan
15.20-15.45
András Bárány and Ádám
Szalontai
15.45-16.05
Coffee break
Session 9 – Chair: Željko Bošković
16.05-16.30
Arthur Stepanov, Manca Mušič
and Penka Stateva
16.30-17.20
Marko Hladnik
(invited speaker)
19.00
Conference dinner
Numeral phrases as subjects and agreement with
participles and predicative adjectives
Strategies of participle agreement with conjoined
subjects in Bosnian/Croatian/Serbian
Hungarian object agreement with possessed
objects: An experimental approach
Two (non-)islands in Slovenian: A study in
experimental syntax
Resumption in Slavic relative clauses
5
DAY 3 (Saturday, 26/9/2015), Room 34
Session 10 – Chair: Jan-Wouter Zwart
9.00-9.50
Željko Bošković
(invited speaker)
9.50-10.15
Aida Talić
On the locality of movement and phases
Adverbial left-branch extraction and the structure
of AP in Slavic
Wh-in-situ in a multiple wh-fronting language
10.15-10.40
Petra Mišmaš
10.40-11.00
Coffee break
Session 11 – Chair: Franja Lipovšek
11.00-11.25
Judit Farkas and Gabor Alberti
The relationship between (in)alienable possession
and deverbal nominalizers in Hungarian
Possessives and the "NP or DP?" wrangle
11.25-11.50
Branimir Stanković
11.50-12.10
Break
Session 12 – Chair: Guglielmo Cinque
12.10-12.35
Marcel Den Dikken and Éva
Dékány
12.35-13.00
Jan-Wouter Zwart
13.00-13.15
Closing ceremony
Recursion restricted
Eliminating external merge
6
3. Abstracts
The Intensional Profiles of Hungarian Deverbal
(Proto-) Imperative Sentence Types
Gábor Alberti, Judit Kleiber, and Veronika Szabó
University of Pécs, Faculty of Humanities, Department of Linguistics
[email protected], [email protected],
[email protected]
In the (S)DRT-based [2] ℜeALIS framework [1], a clause performed in a context is
assumed to convey an infon (a piece of information) that belongs to an intensional profile,
which is an element of set P((M×P(I)×R×T×P({+,–,0}))*): the power set of the set of finite
sequences of a particularly specialized set of level labels. The clause is to be interpreted
against the (possible-world-like but finite) components of this intensional profile, called
worldlets in ℜeALIS, in order to obtain its truth conditions and other semantic and/or
pragmatic well-formedness conditions in the given context.
M is the set of modal labels that say whether an infon serves to someone as some kind
of belief (B), or desire (D), or intension (I). Set I provides degrees for expressing the intensity
of the given modality, from “maximum” (M) up to “some” (s). Associated with the modality
B, for instance, this scale ranges from sure knowledge to weak conjecture. The muss/soll/will
triplet of German epistemic modal auxiliary verbs can be regarded as evidence for the
existence of at least three non-maximal degrees [4:53]. The power set P(I) of I is used in the
formula (1) because certain modal words may be associated with more than one degree of
intensity of a given modality. Set R is responsible for referring to the host of the given infon,
who can primarily be the addresser (AR) or the addressee (ae). Worldlets are also assigned
polarity values, which are members of the eight-element powerset P({+,–,0}) of the two
traditional polarity values “true” (+) and “false” (–) and a not so accustomed value “nonspecified”. The Kleene-star in the formula above manifests the “reciprocal” character of
ℜeALIS by offering, instead of quintuples of the above-discussed labels, finite series of such
quintuples. Finally, the power set symbol in the initial position of the formula requires some
7
explanation. The point is that an infon can be simultaneously associated with more series of
worldlet labels (in the human mind), which can be understood as a “prism effect”.
The content of the components in Figure 1, for instance, applied to the Hungarian
declarative sentence in (1), is as follows, from left to right: “(2) I, (the addresser: AR) know
that Péter moved to Mari (I refrain from telling lies or bluffing). (3) I think that you (the
addressee: ae) do not know this. (4) I think that you would like to be aware of this fact at a
later point τ+ in time (otherwise, I would not have uttered the sentence, since it is important
for me to be relevant). (5) (Being also cooperative) I intend to help you to acquire the infon in
question.”
This analysis is based on the Gricean maxims of conversation. The visual
representation is essentially a conglomerate of (S)DRT boxes [2], but, instead of parts of
segmented logical formulas, it is immediately the referents (constants/variables) contained
that are placed in the partially ordered boxes, augmented with the aforementioned level labels.
(1) Péter Marihoz költözött.
Péter
Mari.Ade move.Past.3Sg
‘Péter moved to Mari’s.’
(2) 〈Β,Μ,AR,τ,+〉
(3) 〈Β,gr,AR,τ,+〉〈B,M,ae,τ,0〉
(4) 〈Β,gr,AR,τ,+〉〈D,M,ae,τ,+〉〈B,M,ae,τ+,+−〉
(5) 〈I,gr,AR,τ,+〉〈B,M,AR,τ+,+〉
Figure 1. The intensional profile of the Hungarian declarative sentence.
In our talk we would like to provide a comparative overview of the different
Hungarian sentence types with imperative verb morphology given in (a-f) below, whose
intensional profiles are presented in the table below (by providing the above-demonstrated
level labels. Our pragmasemantic analyses are chiefly based on Szücs’s empirical
observations and systematization [7], but other results are also considered [3] [6].
8
a. Költözzön
Péter Marihoz!
move-IMP.3.SG P.
M.-ADE
‘Péter should move to Mari’s.’
b. Köööltözzön Péter Marihoz!
move-IMP.3.SG P.
M.-ADE
‘Péter can move to Mari’s, I do not mind.’
a'. ??Költözzek
Marihoz!
move-IMP.1.SG M.ADE
‘Let me move to Mari’s.’
b'. ??Köööltözzek
Marihoz!
move-IMP.1.SG. M.ADE
‘Let me move to Mari’s.’
c. Hadd költözzön
Péter Marihoz!
let move-IMP.3.SG P.
M.ADE
‘Let Péter move to Mari’s.’
c'. Hadd költözzek
Marihoz!
let move-IMP.1.SG. M.ADE
‘Let me move to Mari’s.’
d. [Hadd pletykáljanak]e, [odaköltözöm
Marihoz]e'!
let
gossip-IMP.3.PL there.move-1.SG Mari-ADE
‘Let there be gossip, I do not mind, I will move to Mari’s.’
e . Költözzön
csak Péter Marihoz!
f. Költözzön
már Péter Marihoz!
M.ADE
move-IMP.3SG already P.
M.Ade
Move-IMP.SG only P.
‘Let Péter move to Mari’s.’
‘I want Péter to decide to move to Mari’s at long last.’
It is common in all types (see the first two rows of the table under the heading) that the
addresser of the chosen speech act is sure that the result phase ϕres(e) of the given eventuality
e does not hold (i.e., Péter and Mari still live in different flats, that is, Péter has not moved to
Mari’s yet) and more or less assumes that the addressee is also aware of this fact (the certainty
of her or his assumption is given as ‘nM’, that is, ‘non-maximal’). By performing the basic
imperative type (a), the addresser longs for the result state (ϕres(e)) and wants the addressee to
intend the action e. The addressee’s stimulated intention is optimally efficient if (s)he
coincides with the agent of the action (“Move to Mari’s.”). It is, however, definitely excluded
that the addresser and the Agent coincide (a'). It also must be noted that (a) is the only
imperative in the strict sense proposed in [3:411], (b-f) can rather be classified as protoimperatives.
9
Type (b) differs from the basic type only in intoning the first syllable of the verb stem
in a peculiarly lengthened way. The effect is that now it is not the addresser who longs for the
given action but the addressee or the agent of the action. As for intentions, the addresser
remains neutral, and does not want the addressee to do anything against e. It is, again,
definitely excluded that the addresser and the Agent coincide (b'). Type (c) is associated with a
third “distribution” of intentions among the three straightforwardly interested participants: the
addresser, the addressee and the agent of the action in question. Now it is the addressee who is
assumed not to long for ϕres(e) while the addresser and the Agent long for it. The latter two
participants preferably coincide (c') while this time it is the coincidence of the addressee and
the Agent that is excluded. In the speech act defined by (e), the addresser is definitely against
ϕres(e), which is now assumed to be longed for very much by the agent (or, perhaps, the
addressee). Example (f) presents a new distribution of intentions again: the addresser thinks
that someone, preferably the agent, longs for ϕres(e) very much, and (hence) wants this person
to realize her or his wishes.
Instead of entering into further details, let us conclude the abstract with this
generalization: checking whether the speaker, the hearer and the given situation are suitable
for serving as the addresser, the addressee and the context of the linguistically defined speech
act (à la Oishi [5]) simply requires a truth-conditional investigation primarily into the
addresser’s mind’s certain worldlets (e.g., what (s)he longs for and assumes certain other
persons to long for). The task boils down to get to the worldlets in which certain polarity
values must then be checked.
10
References
[1] Alberti, G., and J. Kleiber. (2014). ℜeALIS: Discourse Representation with a Radically New Ontology. In
Complex Visibles Out There. Olomouc Modern Language Series 4, 513–528.
http://olinco.upol.cz/assets/olinco-2014-proceedings.pdf.
[2] Asher, N., and A. Lascarides. (2003). Logics of Conversation. Cambridge: CUP.
[3] Gärtner, H.-M., and B. Gyuris. (2012). Pragmatic markers in Hungarian: Some introductory remarks. Acta
Linguistica Hungarica 59: 387–426.
[4] Leiss, E. (2014). Modes of modality in an Un-Cartesian framework. In Certainty-uncertainty – and the
attitudinal space in between. Amsterdam: Benjamins, 47–62.
[5] Oishi, E. (2014). Discursive functions of evidentials and epistemic modals. In Certainty-uncertainty – and
the attitudinal space in between. Amsterdam: Benjamins, 239–262.
[6] Péteri, A. (2012). The Hungarian Imperative Particle Hadd. Acta Linguistica Hungarica 59: 439–463.
[7] Szücs, M. (2010). A hadd problémaköre [The problem of hadd ‘let’]. LingDok9. U. Szeged, 193–210.
11
Hungarian object agreement with possessed direct
objects: An experimental approach
András Báránya, Ádám Szalontaib
a
University of Cambridge, bResearch Institute in Linguistics of the Hungarian Academy of
Sciences
a
[email protected], [email protected]
Certain direct objects (DOs) in Hungarian trigger object agreement (OBJ in (1)) in
addition to subject agreement, while others do not (SBJ in (2)). Possessed DOs are exceptional,
because they sometimes require object agreement and sometimes do not.
(1)
Mari olvas-sa /
*olvas-ø
Petőfi vers-é-t.
M.
read-3SG.OBJ
read.3SG.SBJ P.
‘Mari read Petőfi’s poem.’
poem-3SG.POSS-ACC
While DOs with nominative possessors always trigger object agreement, (1), DOs with
dative possessors differ, (2): datives can extract to a pre-verbal position and some speakers
allow subject agreement in such cases, with a non-specific interpretation of the DO. Szabolcsi
(1994) links extraction to non-specificity of the possessed noun, Bartos (1999) and É. Kiss
(2000) argue that such DOs are NumPs/NPs but not DPs.
(2)
Chomsky-nak nem
olvas-t-ad /
%olvas-t-ál
Ch.-DAT
NEG
read-PAST-2SG.OBJ
read-PAST-2SG.SBJ
‘You haven’t read Chomsky’s poem/any poem of Chomsky’s.’
vers-é-t.
poem-3SG.POSS-ACC
It is unclear which varieties or groups of speakers allow the %-marked option in (2).
Most authors merely state that there is variation (Szabolcsi 1994, Bartos 1999, É. Kiss 2000,
Coppock 2013). In this paper, we present the results of two experiments that aim to assess and
describe: (i) the acceptance of the presence and absence of object agreement with nominative,
dative and null possessors, (ii) the effects of word order and modifiers on the possessed DO
and (iii) possible demographic effects on variation.
This question is of theoretical interest because the variation in agreement relates to the
syntactic representation of different possessors and their influence on object agreement, as
well as the interpretational properties of possessed direct objects. In Standard Hungarian,
where all possessed DOs require object agreement, agreement cannot be determined
12
semantically. In varieties that do allow for variation, agreement arguably correlates with the
interpretation of the DO. This paper tests this claim, and investigates the influence of
syntactic and semantic properties on agreement.
Experiment design
Two acceptability judgement experiments were conducted. The first (174 subjects; 130
female, 44 male; mean age: 30.8) tested the acceptability of subject and object agreement in
constructions involving DOs with either a dative or a nominative possessor. In addition, the
experiment surveyed the effects of semantic and syntactic modification of the DO through
negation, weak and strong quantifiers and the pre-verbal extraction of the dative possessor as
well by varying the post-verbal word order of the dative possessor and the possessum. The
second experiment (65 subjects; 57 female, 8 male; mean age: 21.9) expanded these tests to
include null possessors, the effects of the focus marker csak ‘only’, as well as repetitions to
test the consistency of speakers’ judgements. Subjects were surveyed online, and were asked
to judged sentences on a 7 point Likert scale. Target items were composed of variations of 4
sentences. The results were analyzed using repeated measures ANOVA.
Results and Discussion
Subject agreement was always judged unacceptable, while object agreement in
constructions with nominative possessors was always acceptable (Figure 1). Object agreement
with dative and silent possessors were judged significantly worse than those with nominative
possessors, but significantly better than all sentences with subject agreement. Negation and
the presence of modifiers did not have a significant effect on acceptability scores, nor did
extraction or word order variation (Figure 2). A characteristic of judgements given to object
agreement with dative possessors was their wider spread across the scale as compared to the
more clear cut judgements of nominative possessors, or subject agreement. The second set of
experiments showed that subjects are for the most part not consistent in judging these
constructions, but also that lexical effects like the choice of predicate may induce varying
levels of uniformity. While object agreement with dative possessors was more accepted in the
North-East of Hungary (Hajdú-Bihar) than the South (Baranya), the uneven geographic
spread of subjects means that this result must be regarded as preliminary.
13
Given the predominant opinions in the literature,
the lack of acceptability of DOs with
Figure 2 A/D: post-verbal possessor > NP, B/E: pre-
Figure 2 Subject and object (OBJ) and subject only
(SBJ) agreement with dative and nominative
possessors
verbal possessor, C/D post-verbal NP > possessor;
A/B/C: object and subject agreement (OBJ), D/E/F:
subject agreement only (SBJ)
Given the predominant opinions in the literature, the lack of acceptability of DOs with
subject agreement (cf. (2)) is surprising. None of the variation of syntactic structure
(extraction of the possessor, quantificational modifiers) led to significant differences between
judgements. As the interpretation of possessive structures ranges from definite to indefinite,
semantic properties are unlikely to trigger object agreement. If, following Bartos (2000) and
É. Kiss (2000), DPs trigger object agreement, the results suggest that possessive DOs are
treated as the same syntactic category, independently of the position of the possessor and
quantifiers in the noun phrase.
References
Bartos, H. (1999). Morfoszintaxis és interpretáció: A magyar inflexiós jelenségek szintaktikai háttere. PhD
thesis. Budapest: ELTE.
Coppock, E. (2013). A semantic solution to the problem of Hungarian object agreement. Natural Language
Semantics 21, 345–371.
Kiss, É. K. (2000). The Hungarian Noun Phrase is like the English Noun Phrase. In: Approaches to Hungarian 7.
Ed. by G. Alberti and I. Kenesei. Szeged: JATEPress, 119–149.
Szabolcsi, A. (1994). The Noun Phrase. In: The Syntactic Structure of Hungarian. Ed. by F. Kiefer and K. É.
Kiss. Vol. 27. Syntax and Semantics. New York: Academic Press, 179–274.
14
Information structure constraints on the position of
bare indefinites in Russian
Olga Borik
Universitat Autònoma de Barcelona
[email protected]
It is generally assumed in the literature that bare nominals (BNs) in Russian can get
both a definite and an indefinite interpretation (although see Dayal 2004 for a different view).
This observation is empirically supported by the data, but what is also known is that indefinite
nominals do not freely appear in all syntactic positions. In particular, Geist (2010) argues that
BN topics in Russian are excluded from preverbal subject position because they cannot get a
specific reference. In other words, since only strongly referential, i.e. specific, indefinites can
function as topics (at least according to Reinhart 1981), BNs are excluded from topic position.
In this talk, I focus on the distribution of bare singulars (BSgs), to avoid unnecessary
complications that BPls with a generic reference (cf. Carlson 1977) could cause. I will argue
that the generalization suggested by Geist (2010) only partially holds. It is, indeed, the case
that indefinite BSgs are excluded from the preverbal subject/topic position, but this
distributional constraint cannot be due to the fact that BSgs cannot get a specific interpretation
– there is empirical evidence that BSgs can be interpreted as strong indefinites in Russian. I
will propose a different explanation that rules out BSsg in subject/topic position, based on the
claim that BSgs are intrinsically ambiguous between a specific and a non-specific reading and
can only serve as topics after disambiguation.
Let me first clarify some theoretical and/or terminological assumptions I am making in
this paper. First of all, the notion of specificity, which has various uses in the literature on
indefinites, is understood here as semantic/pragmatic notion in terms of referential intentions
of the speaker (cf. von Heusinger 2011). More precisely, if the speaker has a particular
referent in mind that he intends to use for an indefinite expression, this expression is specific.
This notion of specificity can be roughly described as ‘known to the speaker, unknown to the
hearer’ and is broader than some other characterization of specificity, for instance, in term of
partitivity (Enç 1991) or as referential vs. quantificational (Fodor & Sag 1982).
Just as the notion of specificity, the notion of topic is notoriously difficult to define. In
this paper, topic is an information structural notion, i.e. it is a component of a sentence which
15
is classified as a categorical judgment as opposed to a thetic judgment (Ladusaw 1994, based
on Kuroda 1972) and hence, is not meant to have any syntactic implications. Thus, a term
‘topic position’ simply means a position where a nominal argument (usually a subject, and
hence it is a subject position) is understood as a topic.
Consider now the empirical data. The example in (1) illustrates that Russian BSgs can,
indeed, have an indefinite reading (contra Dayal 2004). A BSg in (1) is fully acceptable in a
distributive context, where the nominal phrase cannot be interpreted as either a kind or a
definite.
(1) V kazhdom dome igral
rebenok.
in every
house play.IMP.PST
child
‘A child (a different one) was playing in every house’
The next example illustrates that indefinite BSgs are, indeed, banned from the
preverbal subject/topic position. (2a) sets a context which requires the subject in (2b) to be
interpreted as a non-specific indefinite , and a BSg in (2b) is akward.
(2) a. V komnate bylo neskol’ko malen’kih detej.
in room
were several
small
children.
‘There were several small children in the room.’
b. #Devočka podošla ko mne i
sprosila…
girl.NOM. came.up to me and asked…
However, if we look at BSgs in object position, we can see that they exhibit regular
scopal ambiguities just as expected of canonical indefinites. This is illustrated in (3): BSgs
can get either a weak, non-specific (corresponding to a narrow scope, 3a) or a strong, specific
(corresponding to the wide scope, 3b) interpretation.
(3) Vasja hochet
zhenit’sja
na
Vasja wants
marry
to
‘Vasja wants to marry a movie star…
kinozvezde,
movie-star
a. no ne mozhet
najti podhodjaschuju.
but not can
find suitable
…but cannot find a suitable one.’
want > Indef
b. no my eje esče ne videli.
but we her yet not saw
but we haven’t met her yet.’
Indef > want
16
In fact, (3) is three-way ambiguous: the object a movie star can be non-specific,
specific for the subject or specific for the speaker. Formal accounts of indefinites in terms of
choice functions can capture this three-way ambiguity (but I omit the formal details).
Crucially, (3) illustrates that the reason why a BSg cannot appear in a topic position in (2b)
cannot be explained by the general ban on a specific interpretation of indefinite BSgs. Thus,
Geist’s generalization that indefinite BSgs cannot appear in topic position seems to be correct.
Her explanation, however, does not seem to hold.
An alternative explanation that I would like to propose instead is based on the inherent
ambiguity of BSgs between a specific and a non-specific interpretation. Note that this is not a
structural ambiguity, i.e., we do not have to postulate that the two types of indefinites are
associated with different syntactic structures. Rather, it is a discourse-related ambiguity which
can be resolved either by contextual or by specific sentential information. In the absence of
this information, there is simply no means to determine if the BSg has a specific or a nonspecific reference. Since only specific indefinites are allowed in topic position (Reinhart
1981), BSgs cannot appear as sentential topics unless they are disambiguated and established
as referring specifically.
One of the means for such a disambiguation is a use of various referential indicators
within the nominal phrase. In particular, there are various pronominal specificity markers in
Russian that can be used to determine the type of reference of a nominal phrase (for
discussion, see Haspelmath 1997, Bylinina & Testelec 2004, Yanovich 2005, Geist 2008,
Ionin 2013, etc.). An important observation for my purposes is that if the subject in (2b)
appears with an overt specificity marker, the sentence becomes acceptable:
(4) Odna/ kakaja-to devočka podošla ko mne i sprosila…
one/ some
girl.NOM came.up to me and asked…
If, however, the subject combines with a marker that yields a non-specific reading, it is
ruled out from the topic positon:
(5) #Kakaja-nibud’
some (unspecific)
devočka podošla
girl.NOM came.up
ko mne i
sprosila…
to me and asked…
Another possible linguistic indicator of a specific reading is the more detailed,
descriptive reading of the noun phrase, achieved, for instance, by means of restrictive and
non-restrictive relative clauses (Fodor & Sag 1982). Thus, the modified subject in (6) is
perfectly acceptable:
17
(6) (same context as in 2a)
Devočka so slomannoj kukloj v ruke podošla ko mne i
sprosila…
girl.NOM with broken
doll in hand came.up to me
and asked…
‘A girl holding a broken doll in her hand came up to me and asked… ‘
Examples (4) to (6) show that disambiguated nominals behave in accordance with the
general expectations: specific one can serve as topics, non-specific ones are ruled out. The
problem with BSgs in topic position arises only when they can be interpreted as non-specific.
Various specificity markers are employed to establish the reference of BSgs unambiguously.
References
Bylinina, Elisaveta & Yakov Testelets. (2004). On Sluicing-based Indefinites. In S. Franks, F. Gladney, and M.
Tasseva-Kurktchieva (eds.) Proceedings of FASL 13. Michigan Slavic Publishers. Ann Arbor, MI.
Dayal, Veneeta. (2004). Number marking and (in)definiteness in kind terms. Linguistics & Philosophy 27.
Enç, Mürvet. (1991). The semantics of specificity. Linguistic Inquiry 22: 1–25.
Fodor, Janet and Ivan Sag (1982). Referential and quantificational indefinites. Linguistics and Philosophy 5:
355–398.
von Heusinger, Klaus. (2011). Specificity. In: K. von Heusinger, C. Maienborn, P.Portner (eds.). Semantics: An
International Handbook of Natural Language Meaning. Vol.2. Berlin: de Gruyter, 1024–1057.
Haspelmath, Martin. (1997). Indefinite pronouns. Oxford: Oxford University Press.
Geist, Ljudmila (2008): Specificity as Referential Anchoring: Evidence from Russian. In: Grønn, Atle (Ed.):
Proceedings of SuB12. Oslo.
Geist, Ljudmila. (2010). Bare Singular NPs in Argument Positions: Restrictions on Indefiniteness. International
Review of Pragmatics 2(2): 19–227.
Ionin, Tanya (2013). Pragmatic variation among specificity markers. In Stefan Hinterwimmer and Cornelia Ebert
(eds), Different Kinds of Specificity Across Languages. Springer, 75–103.
Ladusow, William. (1994). Thetic and categorical, stage and individual, weak and strong. In Proceedings of
SALT IV, ed. Mandy Harvey and Lynn Santelmann, 220–229. Ithaca, NY: Cornell U. DMLL.
Reinhart, Tanya (1981). Pragmatics and linguistics. An analysis of sentence topic. Philosophica 27: 53–94.
Yanovich, Igor (2005). Choice-functional series of indefinites and Hamblin semantics. Proceedings of SALT 15.
UCLA.
18
On the locality of movement and phases
Željko Bošković
University of Connecticut
[email protected]
It is well-known that extraction from clausal complements of nouns is disallowed, as
illustrated by (1). The ban in question, which goes back to Ross (1967), is referred to as the
Complex NP Constraint (2).
(1) *Howi did you hear [NP rumors [CP that [IP a dog bit John ti]]]?
(2) The Complex NP Constraint
Extraction from complex NPs is disallowed, where complex NP is a noun modified by a
clause.
While extraction from complex NPs is disallowed, extraction from such VPs is
allowed. In other words, while the Complex NP Constraint holds, there is no such thing as the
Complex VP Constraint.
(3) Howi did you [VP think [CP that [IP a dog bit John ti]]]?
Previous research on the locality of movement has focused on (3), putting (1) aside. I
argue that that that move has been misguided since (1) represents a pervasive pattern found in
many contexts, (3) being highly exceptional. Thus, prepositions, adjectives, and ergative
verbs (in constructions without expletives) pattern with nouns in the relevant respect, i.e. we
have the Complex PP Constraint, illustrated by Spanish (4), the Complex AP Constraint,
illustrated by (5), and the Complex Ergative VP Constraint, illustrated by (6).
(4) a. se
acordó
de [que [Pedro preparaba
la comida]]
CLITIC.3P (S)HE.remembered prep that Pedro prepared.IMPERFECT the food
‘She just remembered that Pedro used to cook the food.’
b. *¿cómoi se
acordó
de
how CLITIC (S)HE.remembered of
[que [Pedro preparaba
la comida ti]]
that Pedro prepared.IMPERFECT the food
(5) *Howi/Whyi are you [proud [that John hired Mary ti]]?
(6) a. *Howi did it appeal to Mary [that John fixed the car ti]?
b. *Howi did it depress Mary [that John was fired ti]?
19
I will furthermore show that the generalization of the Complex NP Constraint is even
broader; it in fact holds for all types of complements of the lexical heads in question.
After establishing the generalized version of the Complex NP Constraint, I will
explore several ways of deducing it, which have broad consequences. Among other things, the
deductions have consequences for the proper formulation of phases, the determination of what
is sent to spell-out, the proper formulation of the Phase-Impenetrability Condition, labeling,
and successive-cyclic movement. New approaches to all of these will in fact be proposed.
I will also discuss an exception to the Complex NP Constraint. I will show that Bantu
languages do not observe the Complex NP Constraint and provide an account of their
exceptional behavior in this respect.
20
Discontinuous Foci and Unalternative Semantics
Daniel Büring
University of Vienna
[email protected]
Discontinuous Focus is exemplified in (1-a), where, according to Question-Answer
Congruence, the focus should be the property ‘be arrested by the police’, just like in (1-b).
(1) What happened to John? —
a. The police aRRESted John.
John was arrested by the poLICE.
Standard F-marking approaches like Selkirk (1995), Schwarzschild (1999) are forced
to analyze such cases as double foci (the policeF arrestedF John). While technically possible,
this means that (1-a) and (2) have identical F-marking.
(2) (The secret service abducted John. —) Nonsense! The poLICE aRRESted John.
This paper explores a different route. Following Büring (2015), focus alternatives are
not introduced by an F-marked node, but constrained at branching nodes, where the
prosodically weak daughters are marked as non-focus or ‘un-alternatives’. Essential steps in
deriving these unalternatives for the answers in (1) are sketched below:
1a
1b
R focus of form ‘R John’ (R ̸=‘arrested’)
[s(trong)aRRESted][w(eak)John]
[wthe police][saRRESted John] d ‘the police’ is not narrow focus
[warrested][sby the poLICE]
d ‘arrested’ is not a narrow focus
[wJohn][swas arrested the poLICE] d ‘John’ is not a narrow focus
The middle column in the table indicates whether the metrical w/s assignment is
default (right strong in all cases in (1)) or reversed. These two configurations introduce two
different kinds of constraints. In (1b), with only default weights, the constraints are merely
that neither V nor the subject are narrow foci; this is compatible with object-, VP- or S-focus,
as it should (this framework does not represent focus ambiguities; to each metrical structure
there is exactly one set of F-restrictions).
In (1a), the metrical weight reversal between V and object leads to a stronger
restriction: it requires that the weak daughter be part of the background (i.e. ‘in’ every focus
alternative). Combined with the S-level restriction that the subject not be a narrow focus, we
21
derive that (1a) is either narrow V focus or subject+V focus. There is no formal difference
between discontinuous foci and ‘regular’ foci.
Among the empirical advantages of this treatment is that it has no problems with examples
like (3).
(3) Where’s Kim’s homework?/What happened to Kim’s homework?
Someone/Smone STOLE it.
You’re/Ya SITTING on it.
I/∅ forgot it on the BUS.
In all of these, the pronominal subject is prosodically weak; it does not receive phrasal
stress. In fact, it can even be further reduced to clitic status (smone, ya), or zero in (3-c). This
kind of reduction is generally possible within a larger focus (and in the background, of
course), but not for a focus, even in pre-final position:
(4) a. (First, JO STOLE my homework and now) YOU are/*ya SItting on it.
b. (First my brother misplaced my homework and now) *(I) forgot it on the BUS.
This suggests that (3) are not cases of double focus, or more generally, that the subjects
themselves are not F-marked.
A final, even trickier class of discontinuous foci discussed in this paper is illustrated in
(5).
(5) a. (Why is this guy so sad? —) Maybe his WIFE left him.
b. (What happened to Kim’s homework? —) The DOG ate it.
Evidently, neither VP nor V can be assumed to be F-marked in these cases (they are
accent-less), yet a subject+V focus reading is clearly available. Again, the F-restricton
framework is equipped to handle these cases. First, since the object is pronominal,
[strongate][weakit] is in fact the default prosody, meaning only it as a narrow focus is excluded.
Second, [strongsubject][weakVP], too, is default prosody here, since this is a case of an
integrated or thetic structure, analogous to intransitive S-focus cases like our DOG
disappeared or a DOG is barking. Therefore, this structure merely excludes it and ate it as
narrow focus (it also excludes narrow V focus, given the way F-constraints propagate up the
tree, as I discuss in the talk); it is compatible with either the subject, or the entire clause, or,
discontinuously, subject+V as focus.
References
Büring, Daniel. (2015). Unalternative Semantics. Paper presented at SALT 25.
Schwarzschild, Roger. (1999). GIVENness, AvoidF and Other Constraints on the Placement of Accent. Natural
Language Semantics 7(2): 141–177.
Selkirk, Elisabeth O. (1995). Sentence Prosody: Intonation, Stress, and Phrasing. In John A. Goldsmith, ed., The
Handbook of Phonological Theory. London: Blackwell, 550–569.
22
Vocalic length as evidence for the incorporated-free
particle distinction in Czech
Pavel Caha and Markéta Ziková
ÚČJ, Masarykova Univerzita
[email protected]
1. Introduction ∗
In Norwegian (and a number of other Germanic languages), prepositional particles
alternate between two states; we will call them free and bound (for a lack of a better term). In
Norwegian, the free state is characterized by two properties; (i) the particle can be separated
from the verb, and (ii) the verb precedes the particle, see (1a). The second state will be refered
to as bound, and it is characterized in Norwegian by the fact that the particle (i) cannot be
separated from the verb and (ii) the verb follows the particle.
(1)
Norwegian
a.
kast katta ut!
throw the cat out
‘throw the cat out’
FREE
b.
ut-kast
BOUND
out-throw
'a discard/a draft'
Czech prepositional particles are only found in what superficially resembles the bound
state. In all of their uses, they are both (i) inseparable and (ii) preverbal. This is shown in (2).
Specifically, in the grammatical example (2a), the particle vy- ‘out’ precedes the verb and it is
adjacent to it. Any permutation that violates either property (i) or (ii) leads to
ungrammaticality. (2b,c) are failed attempts to separate the prepositional particle from the
verb, and (2d) shows that the particle cannot follow the verb.
(2)
a.
b.
vy-hoď
tu kočku
out-throw
the cat
‘throw the cat out!’
*hoď tu kočku
vy
throw the cat
out
c.
d.
*vy
tu kočku
hoď
out
the cat
throw
*hoď-vy
tu kočku
throw out
the cat
The widely accepted conclusion drawn from data such as (2) is that the prefix and the
verb form a word -- a complex head. A popular analysis is shown in (3) below. SC stands for
“small clause,” where the particle is predicated of the direct object. The label is used for
∗
This contribution is funded by grant no. 14-04215S (Morphophonology of Czech: Alternations in Vowel
Length) issued by the Czech Science Foundation.
23
convenience, as a shorthand for whatever the exact structure of the predicative structure is.
The rough constituency and head movement of the prefix are the important bits.
(3)
rootP
root
P
vy
'out'
root
hoď
throw
SC
DP
P
kočku
'cat'
vy
'out'
The goal of this paper is to show that contrary to the appearances based on (2), Czech
prefixes do alternate between the ‘free’ and ‘bound’ state. However, the language specific
reflex of the two states is different in Czech than in Norwegian: it is revealed through an
alternation in vocalic quantity (first discussed from a theoretical perspective in Scheer 2001
and Ziková 2012). Specifically, in roughly the same environments where Norwegian particles
are found in the free state, Czech prefixes have a short vowel (4a). Where Norwegian particles
are bound, Czech particles have a long vowel (4b).
(4)
a.
vy-hoď tu kočku!
out-throw the cat
‘throw the cat out!’
SHORT
b.
vý-hoz
LONG
out-throw
‘a throw-off ’ (in handball)
In the paper, we investigate in detail the precise conditions that govern the length
alternation in Czech. Building on this, we establish a syntactico-semantic parallel between the
conditions that govern the Czech length alternation and the Norwegian separable/inseparable
alternation. The existence of a common logic behind the two alternations can be captured if a
unified explanation can be provided for them, a possibility we set out to explore. Specifically,
we put forth the hypothesis that the unifying mechanism for the alternations in (1) and (4) is
particle movement: in a particular context, the particle moves from inside the VP (where it is
base-generated) to a Spec of a dedicated functional head F (proposed first independently in
Svenonius 2004).
The proposal for (1a) is shown in (5a), drawing on Taraldsen (2000). In Norwegian,
removing the particle from the VP (full arrow) leads to particle stranding when a subsequent
step of VP movement (dashed arrow) caries the root to the left of the particle.
24
(5)
Taraldsen (2000):
a.
FP
PP
ut
b.
FP
F
F is an eventive/verbal projection
(absent in nouns)
nP
VP
VP
n
PP
V
PP
V
P
kast
P
kast
ut
Ø
ut
Naturally, then, even though Taraldsen (2000) does not discuss this explicitly, we
account for the bound state of the particle in (1b) by claiming that structures with bound
particles lack F. With F missing, the particle must remain in its base position. As a
consequence of its VP internal position, it precedes the verb and it cannot be separated from
it.
In sum, Taraldsen’s account of the free/bound distinction has two independent
ingredients: particle movement and verb movement. Their interaction yields the observed
surface patterns in Norwegian. However, depending on how verb movement works in a
particular language, particle movement (the underlying cause of the distinction between the
two states) may have different effects across various languages.
We propose that in Czech, verb movement proceeds differently, and the verb never
crosses the particle, so there is no effect on the order. The way we implement this technically
is to say that verb movement in Czech moves the whole FP, and so the verb root can never
move to the left of the particle, as shown in (6a).
(6)
a.
FP
b.
PP
FP
vy
VP
PP
P
F is an eventive/verbal projection
(absent in nouns)
n/aP
F
VP
V
PP
hoď
throw
P
vý
out
vý
out
25
n/a
V
hoz
throw
Ø
In fact, the only effect of prefix movement is that the prefix gets structurally far away
from the root. We propose that the Czech phonology cares about this difference, and the
length distinction arises as its consequence. In sum, the Czech verbal prefix is subject to the
same movement from a ‘bound’ position to a ‘free’ position, where ‘bound’ and ‘free’ refer to
a contrast in structural distance.
Importntly, there are independent reasons for a theory along these lines. Specifically,
we look at the behavior of prefixes when they are used adnominally as prepositions. What we
find is that when these morphemes are used as a run-of-the-mill ‘free’ preposition, they are
short, see (7a). However, prepositions can also be ‘bound’ in formations that resemble
English forms such as underground. A Czech equivallent is shown in (7b), and what we see is
that when the preposition is ‘bound,’ it is long. So (7) independently confirms that whether
the preposition forms a prosodic word with the root (it is ‘bound’) or not (it is ‘free’)
correlates with its quantity.
(5)
a.
na
břeh-u
on
bank-loc
‘on the bank’
b.
ná-břež-í
on-bank-PLACE
‘a river side’ (lit. the (PLACE) on bank)
Hence, our proposal that Czech verbal prefixes alternate between ‘free’ and ‘bound’
(in terms of their structural position) then allows us not only to capture the common logic
between the Norwegian and Czech verbal alternation, but it also allows us to capture a
parallel across the adverbal and adnominal uses of the particular morphemes.
References
Scheer, Tobias. (2001). The Rhythmic Law in Czech: Vowel-final Prefixes. In Gerhild Zybatow et al. (eds.):
Current Issues in Formal Slavic Linguistics. Frankfurt am Main: Lang, 37–48.
Svenonius, Peter. (2004). Slavic prefixes inside and outside VP. In Peter Svenonius (ed.): Nordlyd 32(2), Special
issue on Slavic prefixes. Tromsø, 205–253.
Taraldsen, Tarald. (2000). V-movement and VP-movement in derivations leading to VO-order. In: Peter
Svenonius (ed.): The Derivation of VO and OV. Amsterdam: John Benjamins, 97–122.
Ziková, Markéta. (2012). Lexical Prefixes and Templatic Domains: Prefix Lengthening in Czech. In Markéta
Ziková & Mojmír Dočekal (eds.): Slavic Languages in Formal Grammar: Proceedings of FDSL 8.5,
Brno 2010. Frankfurt am Main: Peter Lang, 325–338.
26
Parametric variation in the functional and the
substantive lexicon
Guglielmo Cinque
Università Ca’ Foscari, Venezia, Italy
[email protected]
An influential conjecture concerning parameters is that they can possibly be “restricted
to formal features of functional categories” (Chomsky 1995,6; cf. also Borer 1984,2f).
In Rizzi (2009, 2011) such features are understood as instructions triggering one of the
following syntactic actions: (1) External Merge; (2) Internal Merge (Move); (3)
Pronunciation/Non pronunciation (the latter arguably dependent on Internal Merge – Kayne
2005).
In the talk I will discuss what appears to be a particularly pervasive source of variation
among languages in the domain of the lexicon (both functional and substantive) and consider
whether this source can be reduced to one of the above actions or should be added to the
inventory.
References
Borer, Hagit. (1984). Parametric syntax: case studies in Semitic and Romance languages. Dordrecht: Foris.
Chomsky, Noam. (1995). The Minimalist Program. Cambridge, Mass.: MIT Press.
Kayne, Richard. (2005). Some Notes on Comparative Syntax, with Special Reference to English and French. In
G.Cinque and R.Kayne, eds., The Oxford Handbook of Comparative Syntax. New York: Oxford
University Press. 3–69.
Rizzi, Luigi. (2009). Some elements of syntactic computation. In D.Bickerton and E.Szathmáry, eds., Biological
Foundations and Origin of Syntax. Cambridge, Mass.: MIT Press. 63–88.
Rizzi, Luigi. (2011). The Elements of Variation: format, locus and acquisition of parametric properties. Talk
given at the 50 Years of Linguistics at MIT, 9 Dec. 2011.
27
Subjunctive obviation as a semantic failure
Francesco Costantini
University of Udine
[email protected]
It is a well-known fact that null subjects in Romance subjunctive clauses appear to
have peculiar interpretative properties. In sentence (1), for instance, the embedded subject is
obviative with respect to the attitude subject, that is, it cannot be de se.
(1) Pietro pensa che pro parta
domani.
Pietro thinks that pro leaves-SUBJ tomorrow
‘Pietro thinks he is leaving tomorrow.’
# if ⟦pro⟧ = Pietro
The phenomenon, which has been called ‘subjunctive obviation’ or ‘subject disjoint
reference effect’, has been studied at least since the 1980s within the generative framework
and different theories have been worked out to explain it. In spite of an abundant literature on
the phenomenon (Bouchard 1984, Picallo 1985, Farkas 1992, Kempchinsky 1987, Avrutin
and Babyonyshev 1997, Manzini 2000, Schlenker 2005, a.m.o.), obviation still remains in
many respects mysterious. In particular, data originally discussed by Ruwet (1984) appear to
be difficult to account for (examples (2) and (3)).
(2) Pietro sospetta che pro possa
aver fatto molti errori.
Pietro suspects that pro may-SUBJ have made many mistakes
‘Pietro suspects he might have made many mistakes.’
(3) Pietro teme che pro abbia
fatto molti errori.
Pietro thinks that pro has-SUBJ made many mistakes
‘Pietro suspects he has made many mistakes.’
?/ if ⟦pro⟧ = Pietro
?/ if ⟦pro⟧ = Pietro
Sentences (2) and (3) show that modals and tense/aspect auxiliaries appear to affect the
interpretation of the embedded subject, marginally allowing for the de se reading.
To account for these facts Schlenker (2005) proposes that the distinction between de se
and de re can be extended to event variables and hypothesizes that the event variable cannot
be bound de se in sentence (1) because the infinitive can convey this reading and ‘blocks’ the
subjunctive.
(4) Pietro pensa di PRO partire domani.
Pietro thinks C PRO leave-INF tomorrow
‘Pietro plans to leave tomorrow.’
28
⟦pro⟧ = Pietro
As for sentences (2), the embedded eventuality is incompatible with the event de se
reading, so that the infinitive does not ‘block’ the subjunctive under the de se reading.
Promising though it may be, Schlenker’s theory is admittedly ‘extremely preliminary’.
Example (3) still appears to be difficult to account for. Moreover, the theory resorts to
sentence-level blocking, which is at least controversial (Embick-Marantz 2008).
In this talk I refine Schlenker’s proposal and I show that, by reformulating it, it is
possible to account for the facts at issue with no need to call on competition and blocking. To
do so, I notice that Schlenker relates the idea of ‘event de se’ to an idea originally discussed in
Higginbotham (2003) w.r.t. verbs like remember and imagine. Comparing sentences like (5)
and (6), Higginbotham points out that the former, though not the latter, expresses a way of
remembering ‘from the inside’, that is, for sentence (5) to be true the speaker must remember
the event of saying as something of which she was the agent.
(5) I remember PRO saying John should finish his thesis by July.
(6) I remember my saying John should finish his thesis by July.
The idea of remembering ‘from the inside’ clearly concerns the notion of selfknowledge (in the sense used in philosophy of language), that is, the self-ascription of mental
states, such as beliefs, wishes, emotions, sensations, etc. (Shoemaker 1996, Burge 1988, 1996,
2007, Recanati 2007). Self-knowledge has been viewed as something different from
knowledge about the world ‘external’ to oneself, as it relies on introspection. Because of this,
self-knowledge is non-inferential and is highly epistemically secure. Moreover, in selfascribing a mental state, the subject is authoritative. Hence, under normal circumstances selfknowledge is endowed with the presumption of truth. To illustrate, if the speaker truthfully
utters sentence (7)A, replying as in (7)B would be infelicitous.
(7) A: I feel pain.
B: #How do you know that you feel pain?
From a syntactic viewpoint, the notion of self-knowledge can be captured by resorting
to the notion of evidentiality, that is, the grammatical category indicating the source and the
reliability of information (Chafe and Nichols 1986, Rooryck 2001). As for self-knowledge,
the source of information is introspection, and the information is completely reliable.
It has been shown (Izvorski 1997, Rooryck 2001, Simons 2007) that embedding
predicates, like believe, think, etc. function as indirect or inferential evidentials. The sentence
in (8), for instance, indicates that the information expressed in the embedded clause is not
29
completely reliable and the source of the information is not the speaker’s own perceptual
experience.
(8) I think it is raining.
Moreover, sentence (8) implicates that the speaker does not know that it is raining.
Thus, belief verbs introduce a clausal implicatures as in (9) (Gazdar 1979, Levinson 1982).
(9) a believes φ +> ¬ [a knows φ]
This suggests that propositions expressing self-knowledge cannot be embedded under
attitude predicates exactly because of a clash between evidential sources: embedding
predicates introduce the embedded propositional content either as indirect or inferred; on the
contrary, self-knowledge is introspective and reliable. Hence, I hypothesize that the
interpretation of pro in sentences like (1) derives from a clash between the evidential nature of
attitude predicates and the fact that the embedded clause expresses self-knowledge (as for (1),
a plan – cf. Copley 2008 on futurate sentences). To show this, suppose (by reductio ad
falsum) that pro is de se. The truth-conditions of (1) can be represented as in (10) (where ca
represent the bearer of the attitude, p a plan):
(10) ⟦Pietro pensa che pro parta domani⟧c, w, t = 1 iff
∀w' compatible with what Pietro thinks at w at t, ca is committed to p in w'
By (9), (10) implicates that the subject of the attitude does not know that he is
committed to a plan in the actual world. But this is absurd (in fact, a form of Moore’s
paradox), because the epistemic access to one’s own plans is direct. Thus, pro cannot be de se
in (1).
As for sentence (2), the embedded clause does not express introspective knowledge
(despite epistemic modals are relative to the attitude bearer’s knowledge – the judge, in the
sense of Lasersohn 2005, Stephenson 2007). The acceptability of reply in (11)B shows this
point (in contrast with ex. (7)). Thus, the status of sentence (2) is also predicted.
(11) A: I may have done many mistakes.
B: How do you know that?
As for example (3), past eventualities can be recollected directly through one’s own
memory (Higginbotham’s ‘from the inside’) or by inferring (see the contrast between (5) and
(6)). Remembering ‘from the inside’ is introspective, so one would expect that an attitude
30
towards a past eventuality cannot be de se in this case. But ‘inferential’ remembering is not
introspective, so the hypothesis correctly predicts sentences like (3) can be de se. To illustrate,
imagine a scenario whereby Pietro is talking about his first school day. He does not remember
how he went to school on that day, but he remembers that he used to go to school by car as a
schoolboy. In this context, sentence (12) appears to be acceptable under the de se reading.
(12) Pietro suppone che pro sia
andato a scuola in auto.
Pietro supposes that pro is-SUBJ gone to school by car
‘Pietro suppose that he has gone to school by car.’
 if ⟦pro⟧ = Pietro
The hypothesis pursued here also correctly predicts that obviation can occur in
indicative clauses as long as a clash between epistemic sources arises (see (13)), as well as in
environments where no infinitival competitor is available (see examples in (14), where (14)b
is ungrammatical because epistemic modals cannot embed infinitival clauses (cf. Epstein
1984)).
(13) Pietro ha saputo che pro ha
il mal di testa.
Pietro has known that pro has-IND a headache
‘Pietro has come to know that he has a headache.’
(14) a. Pietro pensa che sia
probabile che pro abbia il mal di testa.
Pietro thinks that is-SUBJ probable that pro has-SUBJ the headache
‘Pietro thinks that it is probable he has a headache.’
# if ⟦pro⟧ = Pietro
# if ⟦pro⟧ = Pietro
b. *Pietro pensa che sia
probabile pro avere
il mal di testa.
Pietro thinks that is-SUBJ probable pro have-INF the headache
Both facts were unnoticed and could not be explained under previous accounts.
References
Avrutin, Sergey, and Maria Babyonyshev. (1997) . Obviation in subjunctive clauses and Agr: evidence from
Russian. Natural Language and Linguistic Theory 15: 229–267.
Bouchard, Denis. (1984). On the Content of Empty Categories. Dordrecht: Foris.
Burge, Tyler. (1988). Individualism and Self-Knowledge. The Journal of Philosophy 85: 649–663.
Burge, Tyler. (1996). Our Entitlement to Self-Knowledge: I. Proceedings of the Aristotelian Society 96: 91–116.
Burge, Tyler. (2007). Foundations of Mind. Oxford: OUP.
Chafe, W. and Nichols J. (1986). Evidentiality: the linguistic coding of epistemology. Norwood, NJ: Ablex.
Copley, Bridget. (2008). The Plan’s the Thing: Deconstructing Futurate Meaning. Linguistic Inquiry 39: 261–
274.
Embick-Marantz. (2008). Architecture and blocking. Linguistic Inquiry 39: 1–53.
Epstein, Samuel David. (1984). Quantifier-PRO and the LF representation of PROarb. Linguistic Inquiry 20:
499– 505.
Farkas, Donka. (1992). On Obviation. In Ivan A. Sag, Anna Szabolcsi (eds.), Lexical Matters. Stanford
University, CSLI. 85–109.
Gazdar, Gerald. (1979). Pragmatics: Implicature, Presupposition and Logical Form. New York, Academic Press.
Higginbotham, James. (2003). Remebering, Imagining, and the First Person. In Alex Barber (ed.), Epistemology
of Language, Oxford: Oxford University Press. 496–533.
Izvorski, Roumyana. (1997). The Present Perfect as an Epistemic Modal. In SALT VII, Aaron Lawson (ed.).
31
Ithaca, NY: Cornell University. 222–239.
Kempchinsky, Paula. (1987). The subjunctive disjoint reference effect. In Studies in Romance Linguistics, Carol
Neidle, Rafael Nuñez Cedeño (eds.). Dordrecht: Foris. 123–140.
Lasersohn, Peter. (2005). Context dependence, disagreement, and predicates of personal taste. Linguistics and
Philosophy 28: 643.
Levinson, Stephen. (1983). Pragmatics. Cambridge: Cambridge University Press.
Manzini, Maria Rita. (2000). Sentential complementation. The subjunctive. In Lexical Specification and
Insertion, Peter Coopmans, Martin Everaert, Jane Grimshaw (eds). Amsterdam/ Philadelphia: John
Benjamins.
Picallo, Carme. (1985). Opaque domains. Ph.D. diss., CUNY.
Recanati, François. (2007). Perspectival Thought. Oxford: Oxford University Press.
Rooryck, Johan, 2001. Evidentiality. Glot International 5/4-5, 125–133, 161–168.
Ruwet, Nicolas. (1984). Je veux partir / *Je veux que je parte: on the distribution of finite complements and
infinitival complements in French. Cahiers de Grammaire 7: 75–138. Reprinted in Nicolas Ruwet,
1991. Syntax and Human Experience. Chicago/London: The University of Chicago Press.
Simons, Mandy. (2007). Observations on embedding verbs, evidentiality, and presupposition. Lingua 117, 1034–
1056.
Shoemaker, Sydney. (1968). Self-Reference and Self-Awareness. Journal of Philosophy 65: 555–67.
Schlenker, Philippe. (2005). The Lazy Frenchman’s approach to the subjunctive (speculations on reference to
worlds, presuppositions, and semantic defaults in the analysis of mood). In Romance Languages and
Linguistic Theory 2003, Twan Geerts, Ivo van Ginneken, Haike Jacobs (eds.).
Amsterdam/Philadelphia: John Benjamins. 269–309.
Stephenson, Tamina. (2007). Judge dependence, epistemic modals, and predicates of personal taste, Linguistics
and Philosophy 30: 487–525.
32
On a use of German discourse particle denn in the
antecedent of conditionals
Eva Csipaka, Sarah Zobelb
a
University of Göttingen, bUniversity of Tübingen
[email protected],
[email protected]
In this talk we analyze the use of the German discourse particle denn in antecedents of
conditionals (henceforth: CD) illustrated in the following example.
(1)
A: Alex kommt um drei. Dann picknicken wir.
A: 'Alex arrives at three. Then we'll have a picnic.'
B: Wenn denn die Sonne scheint.
B: 'If it is denn sunny.'
(bare antecedent)
So far, the discourse particle denn has been mainly described in the literature as only
being able to occur in questions (cf. e.g. Thurmair 1991, Bayer 2012; but cf. Kwon 2005 who
observes
CD,
but does not analyze it in detail). Discourse particle denn occurs much more
frequently in questions than in conditional antecedents: on the basis of the „Gesprochene
Sprache“ corpus (corpus of spoken German), we estimate that discourse particle denn in
questions comprises about 95% of all (spoken) particle uses (in addition, denn can be used as
a causal complementizer, an archaic variant of comparative als, Engl. 'than'; and a regional
variant of temporal dann, Engl. 'then').
The goal of our talk is to shed light on the meaning contribution of CD to the utterance
it occurs in, describe its usage restrictions, as well as delimit it from alternative linguistic
means that seem to have a similar function, specifically the particle überhaupt and the
intonation pattern verum focus. Our results contribute to the question of how discourse
particles help navigate and structure the discourse, as discussed in a. o. McCready (2006),
Zimmermann (2011), Rojas-Esponda (2014), and Matthewson (2015).
In everyday communication, speakers make many background assumptions which are
not generally up for discussion. Hence addressees need strategies to put these assumptions "on
the table" for discussion if they conflict with their expectations or private beliefs, e.g. "Hey,
wait a minute!" in the case of presupposition (von Fintel 2004). Another strategy to challenge
33
all kinds of background assumptions is to use an antecedent containing CD, see B's answer in
(1).
Intuitively, B's reaction conveys two things: B treats it being sunny as a necessary
precondition for having a successful picnic (we use necessary precondition not in the logical
sense, but in terms of world knowledge: when the weather is good a picnic is more fun).
CD
additionally signals that B doubts more so than A that it will be sunny in the actual world.
We identify two conditions on the use of CD: these conditions are not themselves contributed
by CD; they are the necessary background for CD to be used felicitously and must both the
satisfied for CD to be acceptable. We also give a description of CD's meaning contribution.
(2)
Condition 1 on the use of CD: a conditional antecedent wenn denn p is only acceptable
if the speaker is not committed to the truth of the antecedent proposition p:
[[denn]]c(p) is defined iff DoxcS⊈p
CD
can only occur felicitously if the speaker is not committed to the truth of the
antecedent proposition p, independently of whether the question whether p is already settled
in the actual world. In (1), p is the propositional content of “die Sonne scheint” (Engl. 'it
is/will be sunny'). A and B cannot yet know whether it will be sunny at the time of the picnic.
However,
CD
can also be used in contexts where the speaker could know whether p, but
simply does not, cf. (3).
(3)
Context: A and B are talking about a party that neither of them attended, they know
that Maria and Peter were both invited and that they fight whenever they meet.
A: Maria und Peter haben sich sicher wieder gestritten.
A: 'Maria and Peter definitely had another fight.'
B: Wenn sie denn beide da waren.
(bare antecedent)
B: 'If they denn both attended.
In (3), B could easily find out whether the antecedent proposition p holds in the actual
world, i.e., whether both Maria and Peter attended the party (this fact is settled in the actual
world), but in order to use CD felicitously, B may not be committed to either p or not p.
Condition 1 correctly predicts that CD is bad in factual conditionals (because the speaker is
already committed to the truth of p). This is borne out, see (4).
(4)
A: Maria kommt zu spät.
A: 'Maria is late.'
B: Wenn Maria (*denn) zu spät kommt, fangen wir einfach an.
B: 'If Maria is (*denn) late, we will just start.'
The second condition on the use of
CD
update the Common Ground with p.
34
is that there is a (previous) tacit proposal to
(5)
Condition 2 on the use of CD: there is a previous tacit proposal to update the Common
Ground with p by speaker α:
[[denn]]c(p) is defined iff ∃α∃q [α proposes to update common ground with q & p is a
precondition for q & p is tacitly proposed]
Condition 2 limits the distribution of
CD
in two important ways:
CD
is predicted to be
bad out of the blue or in contexts where there was no previous tacit proposal of p, but it is also
predicted to be bad in contexts where p was asserted, i.e., where an overt proposal to add p to
the Common Ground was made. These predictions match the data. Note as well that the
update with p can have been proposed either by another speaker, as in (1) & (3), or by an
earlier timeslice of the speaker herself. The latter cases, i.e., self-qualification by the speaker,
are the predominant cases found in written corpora, see (6).
(6)
Dramatischer ist – wenn die umlaufenden Zahlen denn zutreffen – die Abwanderung
von Milliardären.
‘More dramatic is the emigration of billionaires – if the circulating numbers are
denn correct.’ (Die Zeit, 12.08.1996 [DWDS])
Conditions 1 and 2 on the use of
CD
capture in which contexts an antecedent wenn
denn p can be used. Now we provide the contribution of CD to the utterance it occurs in. It is
well-known that discourse particles typically do not contribute to the truth conditions of the
proposition they occur with, but rather that they add not-at-issue meaning in the sense of
Simons et al. (2010). This is also the case for CD.
(7)
(8)
Informal proposal for the meaning contribution of CD:
[Context: A discourse participant α tacitly proposes to update the Common Ground
with p (Condition 2). The speaker cS can infer: for α, the probability that p is true in
the actual world w0 is "high enough" (higher than a threshold S used for assertions).
The speaker cS is not committed to p(w0) (Condition 1).]
For cS, the probability that p(w0)=1 is in fact lower than the threshold S; this is the notat-issue content conveyed by CD.
[[denn]]c(p): prob(cS, w0, p) < S
(not at issue)
Regarding the scopal behavior of
CD,
Conditions 1 and 2 require that
CD
takes scope
inside the antecedent below the complementizer wenn (Engl. ‘if, when’) and above the
remaining lexical material, which makes up the antecedent proposition. In other words, the
proposition that is commented on is always only the antecedent proposition even if the
antecedent is part of a full conditional. Two considerations support this: First,
CD
only
comments on the truth of the antecedent proposition in the actual world, but never on the
consequent or the conditional relation between the antecedent and the consequent. Second, we
argue that Condition 1 is satisfied by hypothetical wenn (≈ ‘if’) which conveys that the
35
question whether p holds in the actual world is not settled (see von Fintel & Iatridou 2002 on
“iffiness”).
In a preliminary corpus study using the ZEIT corpus (225.8 mio. tokens, DWDS
corpora), we annotated 265 naturally occurring examples of antecedents containing
CD.
In
particular, we were interested in the relative position of the antecedent with respect to a
possibly co-occurring conditional consequent. Since Condition 2 requires that
CD
follow a
tacit proposal of p, and in written corpora, predominantly examples like (6) occur, we
expected antecedents containing denn to occur predominantly after their consequents.
Results: 45.7% of all antecedents were positioned (possibly parenthetically) inside the middle
field of the consequent; in these cases, the antecedent was mostly positioned directly after a
presuppositional element. 29.4% were extraposed. Hence, 75.1% of all antecedents occur
towards the end of a co-occurring consequent. Only 20.8% were positioned in the pre-field; of
these most did not qualify an element in the consequent. Only 4.2% of the cases were used
bare as an afterthought. In sum, the results so far fit our analysis; further results are pending.
Alternative means to express a similar speaker's attitude are antecedents containing the
particle überhaupt (Rojas-Esponda 2014) and antecedents occurring with verum focus (Höhle
1992), see (9) – focus is indicated by upper case.
(9)
a. Dramatischer ist – wenn die umlaufenden Zahlen überhaupt zutreffen – die …
b. Dramatischer ist – WENN die umlaufenden Zahlen zutreffen – die …
In cases like (6) & (9), denn, überhaupt, and verum focus seem to make the same
contribution (the speaker emphasizes that she is not committed to the truth of the antecedent
in the actual world). This is not generally the case. For space reasons, we only show that their
distributions differ: unlike
CD,
überhaupt cannot be used to self-qualify one's utterance, see
(10).
(10)
(11)
Alex gewinnt ein Auto, wenn die Münze (denn/#überhaupt) Kopf zeigt.
Alex will win a car if the coin (denn/#überhaupt) comes up heads.
A: Alex kommt um drei. Dann picknicken wir. B: #WENN die Sonne scheint.
Propositions that occur with verum focus need to be given in the context, but the
nature of the givenness is different from that for
CD:
in (1), B's utterance with
acceptable, but the corresponding (11) with verum focus is pragmatically odd.
36
CD
is
References
Bayer, Josef. (2012). From modal particle to interrogative marker: a study of German denn. In L. Brug al. (eds.),
Functional Heads. Oxford: Oxford University Press. 13–26.
Höhle, Tilman. (1992). Über Verum-Focus im Deutschen. In J. Jakobs (ed.), Informationstructur und
Grammatik, Linguistische Berichte, Sonderheft 4/1991–92.
Kwon, Min-Jae. (2005). Modalpartikeln und Satzmodus Untersuchungen zur Syntax, Semantik und Pragmatik
der deutschen Modalpartikeln. LMU Munich.
Matthewson, Lisa. (2015). Discourse particles in Gitksan. Handout talk at 37th annual meeting of DGfS.
University of Leipzig.
McCready, Eric. (2006). Japanese Yo: Its Semantics and Pragmatics. Sprache und Datenverarbeitung 30(1): 25–
34.
Rojas-Esponda, Tania. (2014). A discourse model for "überhaupt". Semantics and Pragmatics 7: 1-45.
Simons, Mandy, Judith Tonhauser, David Beaver, and Craige Roberts. (2011). What projects and why. In
Proceedings of SALT 20. 309–327.
Thurmair, Maria. (1991). Zum Gebrauch der Modalpartikel “denn” in Fragen. In Linguistische Arbeiten 260:
377–387. Tübingen: Niemeyer.
von Fintel, Kai. (2004). Would you Believe it? The King of France is Back! (Presuppositions and Truth-Value
Intuitions). In M. Reimer & A. Bezuidenhout (eds.) Descriptions and Beyond. Oxford University
Press. 269–296.
von Fintel, Kai and Sabine Iatridou. (2002). If and When If -Clauses Can Restrict Quantifiers. Ms. MIT.
Zimmermann, Malte. (2011). Discourse particles. In K. von Heusinger, C. Maienborn, & P. Portner (eds.), HSK
33.2. Berlin, New York: Mouton de Gruyter. 2012–2038.
37
Strategies of participle agreement with conjoined
subjects in Bosnian/Croatian/Serbian
Nermina Čordalija, Nedžad Leko, Amra Bešić, Ivana Jovović, Nevenka
Marijanović, Lidija Perković, Midhat Šaljić, Dženana Telalagić and Tamara
Butigan
University of Sarajevo
[email protected]
Following Marušič et al. (2015) we will show that Bosnian/ Croatian/ Serbian, like
Slovenian, has three distinct strategies of subject-predicate agreement when the subject
consists of conjoined noun phrases. Gender and number agreement forms of participles in the
predicate may be computed in three different ways depending on the features of conjoined
noun phrases in the subject. Participles may agree in gender and number with the subject
phrase as a whole (that is agreement with the maximal projection - Boolean Phrase), or with
the conjunct which is closest to the participle, or with the conjunct which is hierarchically the
highest conjunct.
In order to prove this claim, we performed a controlled experimental study of the
morphosyntax of agreement between conjoined subjects and participles in Bosnian/ Croatian/
Serbian. We conducted two types of studies: written elicitation and spoken elicitation, and in
both we registered variability in elicited production. We will present results of our
experiments documenting the existence of three distinct grammars of conjunct agreement in
Bosnian/ Croatian/ Serbian: agreement with the highest conjunct, agreement with the closest
conjunct, or agreement with the Boolean Phrase itself. The experiment was conducted at the
University of Sarajevo. All participants in the experiment were native speakers of Bosnian/
Croatian/ Serbian from Sarajevo, students at the University of Sarajevo. The studies involving
written elicitation of participial agreement were conducted using a fill-in-the-blanks style
questionnaire, administered to a group of 30 students, immediately after oral elicitation of the
same test material. The questionnaire consisted of 108 test examples, composed of 54
examples with preverbal conjoined noun phrases and 54 examples with postverbal conjoined
noun phrases.
We first investigated possible patterns of participial agreement with uniform gender
conjuncts when both conjuncts are plural, and when uniform gender plural subjects occur both
38
preverbally and postverbally. Such subjects largely elicit participial agreement that
corresponds to the gender of the two conjuncts. However, default masculine agreement occurs
even when both conjuncts are the same gender (feminine, or neuter), demonstrating that the
‘resolution rule’ of masculine agreement is attested even in uniform gender conjunctions. But
this was registered predominantly in oral elicitation (44 examples oral, vs. 11 examples
written).
The default masculine agreement is more prominent when both conjuncts are not of the
same gender, but rather of mixed gender – feminine and neuter. In preverbal written
elicitation with [Fem + Neut] conjuncts the majority of participle forms were masculine (78),
there were 58 neuter forms and 20 feminine forms. This clearly shows that participants in our
experiment used three strategies of agreement: agreement with the maximal projection Boolean Phrase (78 masculine participle forms), agreement with the nearest conjunct (58
neuter participle forms), and agreement with the conjunct which is hierarchically the highest
conjunct (20 feminine forms, which is 13%). In preverbal written elicitation with [Neut +
Fem] conjuncts again the majority of participle forms were masculine (96), there were 58
feminine forms and 22 neuter forms. Very similar results were obtained in preverbal oral
elicitation with [Neut + Fem] conjuncts: 82 examples with masculine agreement, 57 examples
of feminine agreement, and 21 examples of agreement with the highest, neuter conjunct.
However, in preverbal oral elicitation with [Fem + Neut] conjuncts there were less examples
of agreement with the highest, feminine conjuncts (only 5 examples), with most examples of
agreement with the nearest, neuter conjunct (87 examples), and 71 examples of masculine
agreement.
The total number of elicited examples, both written and oral, with [Neut + Fem] and
[Fem + Neut] conjuncts in preverbal position was 655. The dominant form of agreement was
the default masculine agreement (327 examples, or (roughly) 50%), then the agreement with
the closest conjunct (260 examples, or (roughly) 40%), and the least represented was the
agreement with the highest conjunct (68 examples, or (roughly) 10%). The percentage of 10%
is not small, so these examples cannot be treated as performance errors, and therefore we
conclude that agreement with the highest conjunct is the third strategy of agreement used by
native speakers. This contradicts the claim by Bošković (2009) that highest conjunct
agreement in preverbal position is not possible in Bosnian/ Croatian/ Serbian.
Then we investigated possible patterns of participial agreement with [Fem + Neut] and
[Neut + Fem] conjuncts in postverbal position. In postverbal written elicitation with [Fem +
Neut] conjuncts there were no examples of agreement with the furthest, neuter conjunct.
39
There were only four examples of default, masculine agreement, and 156 examples of nearest,
feminine conjunct agreement. In postverbal written elicitation with [Neut + Fem] conjuncts
again the majority of examples (172) were examples of agreement with the nearest, neuter
conjunct, there were three examples of default, masculine agreement, and four examples of
agreement with the furthest, feminine conjunct. In postverbal oral elicitation with [Fem +
Neut] conjuncts there were 147 examples of agreement with the nearest, feminine conjunct,
six examples of default, masculine agreement, and four examples of agreement with the
furthest, neuter conjunct. In postverbal oral elicitation with [Neut + Fem] conjuncts there
were 162 examples of agreement with the nearest, neuter conjunct, six examples of default,
masculine agreement, and six examples of agreement with the furthest, feminine agreement.
The total number of elicited examples, both written and oral, with [Neut + Fem] and
[Fem + Neut] conjuncts in postverbal position was 670. The dominant form of agreement was
the agreement with the nearest conjunct (637 examples, or (roughly) 95%), then the default,
masculine agreement (19 examples, or (roughly) 3%), and the least represented was the
agreement with the furthest conjunct (14 examples, or (roughly) 2%). The percentage of 2% is
too small and therefore these examples should be treated as performance errors rather than a
separate agreement strategy. So our results confirm the claim by Marušič et al. (2015) that a
postverbal conjunction will not allow agreement with the second/ last/ farthest conjunct.
References
Bošković, Željko. (2009). Unifying first and last conjunct agreement. NLLT 27: 455– 496.
Marušič, Lanko, Andrew Nevins, and William Badecker. 2015. The Grammars of Conjunction Agreement in
Slovenian. Syntax 18(1): 39–77.
40
Two positions for verbal modifiers: evidence from
derived particle verbs
Éva Dékány & Veronika Hegedűs
Research Institute for Linguistics of Hungarian Academy of Sciences
[email protected], [email protected]
Aims and claims
We argue that contrary to the received view, not all verbal particles and resultatives are
introduced into the syntactic structure in the same way, as complements to the verb. Using so
far unnoticed Hungarian data, we argue that particles and resultatives involving a degree
component are merged as modifiers above the verb. The analysis supports the view that both
the agent and the theme are merged as specifiers in the extended vP.
The problem
Hungarian verbal particles immediately precede the verb in neutral sentences but they
occur in a postverbal position in clauses involving focus or negation.
(1)
János fel-bicikliz-ett
a
the
John up-bike-PST.3SG
‘John biked up the mountain.’
hegy-re.
mountain-to
(2)
János nem/JÁNOS
bicikliz-ett
fel a hegy-re.
John not/John
bike-PST.3SG up the mountain-to
‘John did not bike up the mountain. / It was John who biked up the mountain.’
There is a group of so far neglected particle verbs, however, that feature a non-
separable particle. The particles of feltételez ‘assume’, kivitelez ‘carry out’, kivételez ‘show a
favour toward’, bevételez ‘enter as income’, szemrevételez ‘inspect’, utánvételez ‘collect
(value) upon delivery’, kivonatol ‘précis’, kifogásol ‘take objection to’, befolyásol
‘influence’, and felvételizik ‘take an entrace exam’ are preverbal in non-neutral clauses, too.
(3)
János nem/JÁNOS
fel-vételiz-ett
az
egyetem-re.
John not/John
up-exam.take-PST.3SG the
university-to
‘John did not take an entrance exam. / It was John that took an entrance exam.’
41
It is tempting to treat these particle verbs as lexicalized forms composed in the lexicon.
The particles of these verbs are, however, visible for syntax. Verbs that feature these particles
do not combine with other verb modifiers in neutral sentences, e.g. preverbal bare objects, and
ordinary resultatives or ordinary verbal particles, contrary to what would be expected on a
lexicalist account.
(4)
*A cég kész-re/meg ivitelez-t-e
a terv-et.
the firm ready-to/PRT carry.out-PST-3SG the plan-ACC
‘The firm carried out the plan.’
(5)
*Vörös-re felvételiz-t-em
a szem-em.
red-to entrance.exam.take-PST-1SG the eye-POSS.1SG
‘I got red eyes by taking entrance exams/an entrance exam.’
If the particle and the verb are combined in syntax, however, then it is unexpected that
particles with a highly bleached semantics (exhaustive ki, exhaustive szét, durative el, and
durative át) and certain resultatives (e.g. halálra ‘to death’, betegre ‘sick’) grammatically
combine with the particle verbs under discussion (particle doubling and co-occurrence of a
particle and a resultative pre-verbally in a neutral sentence do not normally happen).
(6)
[after 5 exams] má-ra
ki-felvételiz-t-em
mag-am-at
today-for out-entrance.exam.take-PST-1SG self-1SG-ACC
‘I got exhausted with entrance exams for the day...’
(7)
beteg-re felvételiz-t-em
mag-am-at
sick-to entrance.exam.take-PST-1SG self-1SG-ACC
‘I got myself sick by taking entrance exams’
Theoretical background
We take particle verbs in general to be constructed in syntax. We adopt Surányi’s
(2009) proposal that Hungarian verbal modifiers (including particles and resultatives) are
merged in the VP. They move to a low, vP-internal position where they semantically
incorporate into the verb, and then they move on to their surface position in Spec,TP. As for
derivational morphology, we adopt a DM approach, whereby acategorial roots are embedded
under categorizing heads: v for verbs and n for nouns.
42
The inseparability of the particle
Verbs with inseparable particles involve a nominalization (-t or -ás) under the
outermost v categorizer head. The base verb and the non-parting particle are merged in
syntax, and the particle undergoes movement to the low vP-internal position for semantic
incorporation. The vP constructed this way is embedded under one or more n nominalizing
heads, and the resulting noun is finally embedded under a v verbalizer head (see 8). The
clausal projections are erected on top of this v. Particles are known to be able to move out of
vP, so the reason that the particles of befolyásol ‘influence’, felvételizik ‘take an entrace
exam’, etc. are immobile must be due to the presence of the nominalizer between the core vP
and the outermost verbalizer. We suggest that n is a phase head, so in order to part from the
core verb, the particle would have to move through the specifier projected by n. This
movement is ruled out by a conspiracy of two independently known facts. PPs cannot occupy
specifier positions in the extended NP; if they are to sit in nominal Specs, they must be
attributivized by the suffix -i or the functional head való ‘being’. PPs involving case markers
and particles cannot be attributivized, however (*kert-ben-i/való pad ‘garden-ine-I/being
bench’, *el-i/való séta ‘away-I/being walk’).
(8)
[v [n [n [n [vP fel ve(sz) ] n=t ] n=el ] n=i ] v=z ]
Particle doubling & co-occurrence with resultatives
Verbal modifiers, including particles and resultatives, are normally merged as
complements to the verb. A verb constructed like (8), however, has no complement position
available for VMs: the complement position of the outermost v head is occupied by a
nominal. This rules out examples like (4) and (5). The particles (exhaustive ki, exhaustive
szét, durative el, and durative át) and resultatives (halálra ‘to death’, betegre ‘sick’) that our
particle verbs can exceptionally combine with have the semantic component ‘degree’ in
common. We argue that they are modifiers rather than complements of the verb, and so their
presence does not depend on the availability of the complement position. The ki ‘out’ of (6) is
merged above the complex verb felvételiz ‘take an entrance exam’, and the verb and the
particle form a complex predicate. As both felvételiz and ki introduce an argument of their
own, the complex predicate has two arguments. This accounts for the obligatory presence of
the fake reflexive. Given that there is no complement position available to either felvételiz or
the complex predicate kifelvételiz, the theme magam ‘self-acc’ must be introduced as a
specifier (above the complex predicate kifelvételiz). The theme in (6) has the same syntactic
43
and thematic properties of ordinary objects, and we argue that this is best captured if ordinary
objects, too, are merged as specifiers. Our analysis thus supports the analysis of themes in
Bowers (1993), Hale and Keyser (1993), and Den Dikken (to appear).
No particle doubling with ordinary particles
If verbal modifiers in general are merged as complements, but particles and
resultatives with a ‘degree’ meaning are merged as modifiers above the verb, then they could,
in principle, co-occur (e.g. exhaustive ki and ordinary el ‘away’ could both combine with run
or read). We argue that this is excluded for case-theoretic reasons. The verb and the two
particles all introduce an argument of their own. This makes for 3 arguments in total, with
only two case assigners available.
References
Bowers, John. (1993). The syntax of predication. Linguistic Inquiry 24: 591–656.
Dikken, Marcel den. to appear. Raising the subject of the object-of relation. MIT Working Papers in Linguistics.
Hale, Kenneth & Samuel Jay Keyser. (1993). On the argument structure and the lexical expression of syntactic
relations. In The view from building 20: essays in linguistics in honor of Sylvain Bromberger, ed.
Kenneth Hale & Samuel Jay Keyser. Cambridge, MA: MIT Press. 53–109.
Surányi, Balázs. (2009). Verbal particles inside and outside vP. Acta Linguistica Hungarica 56: 201–249.
44
Recursion Restricted
Marcel den Dikken & Éva Dékány
Research Institute for Linguistics, Hungarian Academy of Sciences
[email protected], [email protected]
1 The contrast between Hungarian (1a,b), on the one hand, and (1c–e), on the other,
suggests that ki is grammatical per se as a caseless (‘nominative’) possessor but that under
certain circumstances it ‘outgrows’ the DP-internal caseless possessor position, and can only
be realised in the DP-peripheral dative possessor position, as in (2). Semantic definiteness is
not a factor in this: (1a,b) do not form a natural class as opposed to (1c–e) in terms of
semantic definiteness. The restriction in question is syntactic in nature. The central hypothesis
of this paper is that what underlies the pattern in (1) is a restriction on self-embedding
recursion structures created through External Merge, stated in (3). This restriction is the exact
counterpart of the familiar c-command cum phasemate requirement imposed on identical
copies of a single category under Internal Merge.
(1) a. mindenki háza ‘everywho house.POSS, i.e., everyone’s house’
b. valaki háza ‘somewho house.POSS, i.e., someone’s house’
c.*aki háza ‘A-who house.POSS, i.e., whose house (RELATIVE)’
d.*ki háza? ‘who house.POSS, i.e., whose house (INTERROGATIVE)’
e.*ki-ki háza ‘who-who house.POSS, i.e., everyone’s house (DISTRIBUTIVE)’
(2) a. mindenkinek a háza ‘everywho.DAT the house.POSS, i.e., everyone’s house’
b. valakinek a háza ‘somewho.DAT the house.POSS, i.e., someone’s house’
c. akinek a háza ‘A-who.DAT the house.POSS, i.e., whose house (REL)’
d. kinek a háza? ‘who.DAT the house.POSS, i.e., whose house (INT)’
e. kinek-kinek a háza ‘who.DAT-who.DAT the house.POSS, everyone’s house (DIST)’
(3) Restriction on recursion
A phasal category of type α can be embedded in a phasal category of the same type where
there is a c-command relation between the heads of the two instances of α only if the two
instances of α are separated by a phase head.
2 For possessed noun phrases, we adopt a syntax in which there is no phase head
between the possessed noun phrase’s outer D–head and the phrase which harbours caseless
possessors in its specifier position: (4). (The label ‘PossP’ serves expository purposes only;
the details of the syntax of possession will be laid out in the paper but do not matter here.)
PossP itself is arguably a phase, and PossP may potentially be separated from D by projec45
tions belonging to the functional sequence of the possessum; however, no phase head occurs
in between PossP and D.
(4) [DP DATIVE POSSESSOR [D (...) [PossP CASELESS POSSESSOR [Poss [POSSESSUM]]]]]
↓
NO PHASE
3 The combination of (3) and (4) delivers the pattern in (1)–(2) as follows. For (2), the
size of the possessor is immaterial: as the heads of the possessed DP and the possessor in
SpecDP are not in a c-command relation, dative possessors as large as full DPs will not cause
a violation of (3). For (1), with caseless possessors in SpecPossP, on the other hand, size
matters. The presence of a (formally identical with the definite article) in the relative pronoun
aki indicates that, while semantically indefinite, the relative pronoun is as large as a DP. The
ungrammaticality of *aki háza (1c) then follows from (3), as the structure in (5) illustrates.
For interrogative *ki háza (1d), we adopt a structure that represents its [+WH] feature in D
(‘typing’ features are on phase heads), as in (6), which once again runs afoul of (3). The illformedness of *ki-ki háza (1e) falls out on an analysis of reduplication of ki as involving
movement of ki into the DP domain: distributive ki-ki is itself as large as a DP, preventing it
from occupying the caseless possessor position (see (7)). The grammaticality of mindenki
háza and valaki háza (1a,b) indicates that mindenki and valaki are smaller than DP — valaand minden- are modifiers of ki, adjoined to the QP of ki: (8).
(5) *[DP D [PossP [DP a- [QP ki]] [Poss=-a [n/NP ház]]]]
(6) *[DP D [PossP [DP [+WH] [QP ki]] [Poss=-a [NP ház]]]]
(7) *[DP D [PossP [DP ki [QP ki]] [Poss=-a [n/NP ház]]]]
(8) [DP D [PossP [QP {vala-, minden-} [QP ki]] [Poss=-a [n/NP ház]]]]
(cf. (1c))
(cf. (1d))
(cf. (1e))
(cf. (1a,b))
4 The recursion restriction also explains the distribution of possessed noun phrases as
caseless possessors of larger noun phrases: (9) and (10) are grammatical; (11–12) are not.
(9)
(10)
(11)
(12)
az én kalapom széle
a te kalapod széle
az ő kalapjuk széle
*az ő kalapja széle
*János kalapja széle
the I hat.1SG rim.POSS
the you hat.2SG rim.POSS
the (s)he hat.3PL rim.POSS
the (s)he hat.POSS rim.POSS
János hat.POSS rim.POSS
‘the rim of my hat’
‘the rim of your hat’
‘the rim of their hat’
‘the rim of his/her hat’
‘the rim of János’ hat’
The descriptive generalisation is that when kalap bears φ-feature inflection
(underlined) cross-referencing its possessor, its projection can serve as the caseless possessor
46
of a larger possessed noun phrase. The shared φ-features of the possessor in SpecPossP and
the Poss-head allow the XP–YP structure to be labelled, by φ (cf. Chomsky 2014). Absent φagreement, the XP–YP structure must be included in an extended projection that completes
the structure. For PossP, it is D that facilitates the completion of the extended projection. So
in the absence of φ-agreement between the possessor and the Poss-head, a DP must be erected
on top of PossP. With ő/János kalapja dominated by a DP node, (3) prevents it from being in
the caseless possessor position.
5 The account of (5)–(7) carries over to *ez háza versus ennek a háza ‘the house of
this’. When a demonstrative is used independently (i.e., ez by itself), there is a silent nominal
head that it combines with (Dékány 2011: Hungarian free-standing demonstratives are
portmanteaux of N, Dx and D). This silent head needs licensing by D, which prevents the
demonstrative from being used as a caseless possessor. The DP approach also applies to
Hungarian case-concordial demonstratives in adnominal position: base-merged as full DPs in
a position c-commanded by D and not separated from D by a phase head, they must raise to
SpecDP to avoid violating (3); this explains that case-concordial demonstratives must precede
D (ezt a házat ‘this.ACC the house.ACC’). The non-case-concordial demonstratives, e and eme,
by contrast, never occur in pre-determiner positions and do not have free-standing,
independent uses. They have no phrasal distribution. They are exponents of the Deixis-head
(Dx) in the extended projection of the noun.
6 In that idiot of a doctor and Hungarian csoda egy napok ‘wonder a days’, predicate
inversion (of idiot around doctor, and of csoda around napok) requires phase-extending
movement of the head of the small clause to a SC-external head (Den Dikken 2006). The in
situ subject of the small clause (doctor, napok) thus seems shielded by the head of the
extended phase from the D–head of the complex DP. That *that idiot of the doctor and *csoda
egy a napok ‘wonder a the days’ are nonetheless ungrammatical is because in these complex
DPs, head movement continues overtly up to D: this extends the single phase to DP, and turns
D and the SC-subject into phase-mates, preventing the SC-subject from being a DP (as per
(3)). Head movement to D also explains that such DPs resist an outer definite article (Hungarian
*a csoda egy nap).
7 For English-type the man’s coat, with the possessor in SpecPossP below D,
satisfaction of (3) requires that the NOT form a constituent with man because that would make
the man a DP in an immediate self-embedding recursion configuration. So the must be a spellout of the outer D–head. The non-constituency of the man accounts for its non-extractability.
The placement of the definite article immediately preceding the possessor in the outer D–head
47
also accounts for the fact that the man’s coat is outwardly definite, hence cannot serve as the
associate of there, whereas a man’s coat can (as in there is a man’s coat on the chair). In
English-type the man’s coat, outward definiteness comes from D and the possessor is smaller
than DP. The complex possessed noun phrase and the possessor cannot both be DPs, as per
the recursion restriction in (3).
References
Chomsky, Noam. (2014). Problems of projection: Extensions. Ms., MIT.
Dékány, Éva. (2011). A profile of the Hungarian DP. Ph.D. dissertation, University of Tromsø.
Dikken, Marcel den. (2006). Relators and linkers: The syntax of predication, predicate inversion, and copulas.
Cambridge, MA: MIT Press.
48
Upper bounded and un-bounded no more
Mojmír Dočekal
Masaryk university
[email protected]
Introduction
Slavic constituent negation (CN) seems to share the following properties with CN in
other languages (see e.g. Vicente (2006), for Slavic Borschev et al. (2006) and Jasinskaja
(2010)): (i) it associates with contrastively focused constituent; (ii) it triggers existential
presupposition; (iii) it has exhaustive interpretation. The properties (ii)+(iii) is believed to be
shared by CN and cleft constructions (see Büring (2011), Büring and Kriz (2013)). Although
such properties are exhibited by many CN examples, they don’t hold for all Slavic CN
constructions at all: I conducted a corpus search (material: Czech national corpus, SYN2010,
the biggest representative corpus of contemporary Czech) and then classified 200 randomly
selected sentences with CN. The result is that only 47% of sentences (94/200) exhibit the
properties (i) – (iii) – the prototypical example is in (1); 39% (78/200) + 14% (28/200) do not
have at least the properties (i) and (iii), as exemplified by (2) – discussed in detail later – and
by (3) – CP constituent negation. All to-date existing accounts of Slavic CN (Borschev et al.
(2006), Jasinskaja (2010), a.o.) focus only on examples like (1) and their formal analysis
cannot explain the semantic/pragmatic properties of examples like (2) and (3). The goal of our
talk is to improve the existing analyses and give a unified CN semantics for all types of Slavic
CN.
(1) Jsem politik a ne odborník přes morálku.
I-am politician and not expert on morality
‘I am politician but not expert on morality.’
(2) Ne více než dva lidé vypovídali pravdivě.
no more than two people testified truthfully
‘No more than two people testified truthfully.’
(3) Ne aby
vás napadlomočit
z
oken!
not COMP you cross-your-mind to-piss from windows
‘Don't you dare to piss out of windows!’
Proposal.
Let’s build on the beautiful and most formally advanced CN semantics in Jasinskaja
(2010): J. claims that CN (and sentential) negation share core semantics: CN is a property (P)
49
and generalized quantifier – GQ – (Q) taking operator which returns truth iff the property P is
not in the set of sets denotation of Q: ⟦CN⟧ = λPλQ[Q(λx¬P(x))]. With the exhaustification
understood in the sense of Groenendijk and Stokhof (1984) – see (4), this gives correct truth
conditions for (1) like in (5-a), equivalent to (5-b), with focus just on the exhaustified CN of
(1).
(4) EXH = λQλP[Q(P) ∧ ¬∃P'[Q(P')∧P≠P'∧∀x[P'(x)→P(x)]]]
(5) a. EXH(λP[P(SPEAKER)])(λx[x ∈ C ∧ politician(x)]) ∧ EXH(λP[P(SPEAKER)])(λx[x ∈ C
∧ ⌐morality_expert(x)])
b. … ∀x[[x ∈ C ∧morality_expert(x)] → x ≠ SPEAKER]
Such semantics would be empirically inadequate for examples like (2) and (3) –
neither of the types is interpreted exhaustively (discussed later) and the constituents
(comparative numerical NP in (2) and the whole CP in (3)) are not contrastively focused.
Theoretically J. ⟦CN⟧ would give us wrong truth conditions for (2) like ∀x[[x ∈ C ∧
set_of_two_and_more_people(x)] → x ≠ truthfull_witness(x)] – if we would allow type
shifting of the comparative numerical NP to set. Therefore I propose that the Czech
constituent negation in examples like (2) is interpreted as denial of the string ‘no more than
two’, and building on Guerts’ (1998) theory of denial that the string is both used and
mentioned, while the negation retains its usual truth reversing force (⌐of classical logic). The
logical form of (2) is then ⌐[<the degree properties of two and more, expressed by Czech
‘more than two’> people testified truthfully]. And for reasons discussed immediately, I
believe that its final truth-conditions are in fact ∃!x[#x ≤ 2 ∧ people(x) ∧ truthful_witness(x)].
This interpretation is not-exhaustive (the exhaustive interpretation would have truthconditions ∃!x[#x = 2 ∧ people(x) ∧ truthful_witness(x)]) and in this respect Slavic CN
behaves totally differently from English CN (see Nouwen (2008) and his discussion of
exhaustified interpretation of CN like no more than ten marbles ≈ ...#x = 10...). I claim that
the non-exhaustive interpretation of CN in examples like (2) or (6) – the most common
instantiation of the non-exhaustified CN in the corpus – is a result of even lower scope of CN
than in (1), namely schematically [[[neg] [on all]] [tournaments]] for (6) [[[neg] [more than
two]] [people]] for (2) and its denial interpretation. This is semantically interpreted as
negation of the <denotation, form> pair which leads to the interpretation of the main logical
operator as ≤ in (2) ([⌐ [ ˃ 2 ]] → ≤ 2) and as ∃ in (6) ([⌐∀] → ∃). For (6) I assume that the
LF is then (6-b) instead of (6-a) and not (6-c) which would be the predicate logic equivalent
of (6-a). The evidence for the low scope comes from the lack of FCI licensing (Czech FCI
jakkoliv ‘in any manner’) in constructed example like (8), based on a grammatical example
50
from ČNK: note that syntactical bracketing [¬[∀[NP]]] would be downward entailing and FCI
licensing, but as (8) shows, the actual scope is [¬∀][NP] resulting in [∃][NP] semantics – the
unlicensing of the FCI in (8) explained. Another evidence for the low scope comes from a
case assignment pattern: Czech cardinal numerals from the interval [1,4] don't assign genitive
but act as syntactical adjectives, agreeing with the case of a head noun – see (9-a). Czech
nouns assign genitive as seen in (9-b). But if the constituent structure I propose is right,
numerals [1,4] embedded under comparative are expected to act as nouns, because the sister
of noun is not only the numeral but the whole [more than Num] constituent. And as noun-like
constituent, we expect the case on its noun complement to be adnominal, namely genitive.
And this expectation is fulfilled – see (9-c) where the noun appears in genitive instead of
accusative. This nice syntactical pattern provides another evidence for the proposed
constituent structure. As for the clausal CN like (3): the same idea can be extended there if
again take the constituent negation as denial marker (speech act of denial in this case) where
we remove the proposition (set of possible worlds) from the generalized quantifier denotation
(over possible worlds) of the respective modal base (desire in case of (3)), the rough
semantics for (3) would be R(w) ⊈ p, where R(w) denotes the accessible desire worlds of the
speaker and p the set of worlds where hearer(s) piss out of windows. In all cases: (2), (3) and
(6) the denial interpretation of negation computes the alternatives in the form and denotation
domains, consequently no usual denotational implicatures arise and as results the scalar
implicatures don’t lead to strengthening (=exhaustive interpretation). That is the reason why
(2) unlike its English translation does have just the interval (non-exhaustified) reading – I
assume that Czech constituent negation unlike English no signals denial and consequently it
doesn’t lead to pragmatic strengthening observable with default (Czech verbal, English
determiner no) negation.
(6) Ne na všech turnajích
byla krásná hra oceněna.
not on all tournaments was beautiful play awarded
‘A beautiful play was awarded not at all tournaments.'
a. *¬ ∀
x[tournament(x)
y[beautiful_play(y) ∧ awarded(x, y)]]
→ ∃
b. ∃
x∃y[tournament(x) ∧ beautiful_play(y) ∧ awarded(x, y)]
c. ∃
x∃y[tournament(x) ∧ beautiful_play(y) ∧ ¬awarded(x, y)]
(7) Přitažlivé je i
luxusní provedení všech částí, kterých se jakkoliv dotýkáte.
appealing is even luxury variant
all
parts which SE however touch
‘Appealing is even luxury carrying out of all parts which you touch in any manner.’
51
(8) *Přitažlivé je i luxusní provedení ne všech částí, kterých se jakkoliv dotýkáte.
‘Appealing is even luxury carrying out of not all parts which you touch in any manner.’
(9) a. Přečetl jsem dvě
knihy.
read-I AUX two-ACC books-ACC
‘I read two books.’
b. kniha [mého přítele]
book my-GEN friend-GEN
‘my friends book’
c. Knih
jsem přečetl [víc než dvě].
Books-GEN AUX read more than two-ACC
‘As for books, I read more than two.’
Summary.
We proposed a denial treatment for Slavic CN: CN with syntactical scope [[¬ OP]
[NP/CP/...]] which is not exhaustified w.r.t. its associated NP/CP/..., because the negation
targets pairs <denotation,form> and reverses the respective relation ([[CN¬][ > ]] → [ ≤ ],
[[CN¬][∀]] → [∃], [[CN¬][ < ]] → [ ≥ ], [[CN¬][ ⊆ ]] → [ ⊈ ], …). Consequently no
exhaustification and contrastive-focus interpretation of the associated NP/CP happens.
References
Borschev, Vladimir, Elena V Paducheva, Barbara H Partee, Yakov G Testelets, and Igor Yanovich. (2006).
Sentential and Constituent Negation in Russian BE-Sentences Revisited. In Formal Approaches to
Slavic Linguistics: The Princeton Meeting 2005 (FASL 14), 50–65. Citeseer.
Büring, Daniel. (2011). “Conditional Exhaustivity Presuppositions in Clefts (and Definites). Ms. Under
Submission.
Büring, Daniel, and Manuel Kriz. (2013). It’s That, and That’s It! Exhaustivity and Homogeneity
Presuppositions in Clefts (and Definites). Semantics and Pragmatics 6(6): 1–29.
Geurts, Bart. (1998). The Mechanisms of Denial. Language 74(2): 274–307.
Groenendijk, Jeroen, and Marin Stokhof. (1984). Studies on the Semantics of Questions and the Pragmatics of
Answers.”PhD thesis, University of Amsterdam.
Jasinskaja, Katja. 2010. Corrective Contrast in Russian, in Contrast. Oslo Studies in Language 2(2): 433–466.
Nouwen, Rick. 2008. Upper-Bounded No More: the Exhaustive Interpretation of Non-Strict Comparison.
Natural Language Semantics 16(4): 271–95. doi:10.1007/s11050-008-9034-2.
Vicente, Luis. 2006. Short Negative Replies in Spanish. Linguistics in the Netherlands 23(1): 199–211.
52
Experimental evidence for neg-raising in Slavic
Mojmír Dočekala, Jakub Dotlačilb
a
Masaryk university, bUniversity of Groningen
a
[email protected],
b
[email protected]
Introduction
Neg-raising (NR) is a linguistic inference which accompanies some classes of
sentence-embedding predicates (e.g. believe, want, …) such that the negation of NR
predicates implies the negation of their propositional argument (¬NR[P] → NR[¬P]). Unlike
in English, where 5 Horn’s classes (see Horn 1989) of NR predicates seems to behave mostly
uniformly (see Gajewski 2005, 2007, Homer 2011 and Horn 1989), Slavic languages (SL)
present a more nuanced picture: Boškovič & Gajewski (2009) – B&G – claim that NRs do not
exist in SL, while Dočekal (2014) defends the existence of NR in SL at least for some NR
predicates. In this talk we: (i) bring new experimental evidence supporting the (limited)
existence of NR in SL (in particular, Czech), and (ii) propose a scalar implicature explanation
(building on Romoli 2012, 2013) of the variation between English and SL.
Experiment
The experiment consisted of two parts: an acceptability judgment task and an inference
task. In the first part, participants had to judge the acceptability of sentences with strict NPIs,
ani ‘not even’, using the 5-point Likert scale (5=best, 1=worst). This served as a test of NRhood since strict NPIs can be licensed (i) in negative clauses, or (ii) in clauses embedded
under negated NRs (Horn, 1989). We tested the acceptability of strict NPIs in 5 environments,
shown in (1) with the crucial experimental manipulation highlighted: (A) a positive sentence,
(B) a negative sentence (both in (1a)), (C) a clause embedded under negated NR predicates of
intention and judgment/obligation (e.g. want, advise), (1b), (D) a clause embedded under
negated NR predicates of opinion (believe), (E) non-NR predicates, (1c).
(1) a. Ztratila/neztratila se ani
jedna ovce.
lost/neg-lost
SE not-even one sheep
‘Not a single sheep is missing/A single sheep is missing.’
b. Nový bača
v Tatrách nechce, aby se ztratila ani
jedna ovce.
new shepherd in Tatras neg-wants C SE lost
not-even one sheep
53
A/B
C
c. Nový bača
v Tatrách si nemyslí/neříká, že se ztratila ani
jedna ovce. D/E
new shepherd in Tatras SI neg-think/neg-say C SE lost
not-even one sheep
‘New shepherd in the Tatra mountains doesn’t think/say that a single sheep is missing.’
In the second part of the experiment, participants had to judge the following
inferences: (I) whether neg-raising is intuitively valid (¬NR[P] → NR[¬P]) (II) whether cyclic
neg-raising is valid (¬NR1[NR2[P]] → NR1[NR2[¬P]]), (III) whether existential wide scope is
valid (¬ ∀x NR1[NR2[P]] → ∃xNR1[NR2[¬P]]). Only NRs of intention were used in this part
of the experiment. There were 40 exp. items in part 1 and 20 exp. items in part 2. Each part
also included 30 fillers. 60 Czech native speakers took part in the experiment. The experiment
was run online in Ibex.
Results
Participants had no problems to pass control fillers, which consisted of
uncontroversially grammatical/ungrammatical sentences: the score of grammatical sentences
was on average higher than the score of ungrammatical sentences for each participant, and the
difference was at least 1.39 points. We therefore kept all the subjects for subsequent analyses.
Counts of responses in the acceptability task is shown in the figure below. We analyzed the
data using mixed-effects ordered probit regression, with one fixed effect, condition (Condition
C was the reference level), and slope+intercept random effects for subjects and items.
54
We found that negated sentences, Condition B, were judged as better than NRs
(β=1.84, z=23, p<.001). In turn, positive sentences, Condition A, were judged as worse than
NRs (β=-1.1, z=-15, p<.001), and so were sentences with negated non-NR predicates,
Condition E (β=-0.65, z=-9, p<.001). There was no difference between Condition C and D
(two types of NRs). The fact that positive sentences were strongly degraded and negative
sentences were acceptable supports our starting assumption that ani ‘not even’ is treated as an
NPI. We take the difference between Condition C and E, on one hand, and the lack of
difference between C and D, on the other hand, to provide evidence (i) for treating ani ‘not
even’ as a strict NPI (otherwise, Condition E should be acceptable, contrary to the facts), and
(ii) that Czech has a class of NR verbs. The analysis of the inference task was carried out
using mixed-effects logistic regression (with 1=inference follows, 0=inference does not
follow), one fixed factor, Condition (I being the reference level), and intercept+slope subject
and item random effects. Condition I was significantly higher than a chance, prob=0.5 (β=0.9,
z=3.3, p=.001), i.e., “inference follows” was preferred. In contrast to that, Condition II and III
had a significantly smaller preference for 1, in fact, they did not significantly differ from
prob=0.5. The difference is unexpected in previous accounts of NRs (according to which all
three inference tasks should be possible) but we believe the difference is likely a consequence
of the higher complexity of Condition II and III (i.e., more clauses, more complex sentences)
and should thus not play a role in the development of NR theories. Finally, since the (first
twenty) items in Condition C of the acceptability task and Condition I of the inference task
were similar to each other (using the same NR verbs), we checked whether the acceptability
of the strict NPI ani correlated with the acceptability of the inference NR[P] → NR[P].
Indeed, we found a strong positive correlation in item random effects between the two tasks
(r=0.7, p<.001). This suggests that the acceptance of strict NPIs and NR interpretation are
related, strengthening our conclusions and arguing against B & G, who claim that NR
inference is independent of NPI licensing in SL.
The analysis
We follow the scalar approach to NR (esp. Romoli 2012, 2013) where NR predicates
(beside the assertion) contribute the excluded middle (EM) to the semantic composition:
Alt(NR) = {λpλx.□x[p], λpλx.[□x[p] ∨ □x[¬p]]}. When a sentence like (1-b) contains NR in the
scope of negation, the exhaustification of the EM alternative leads to a schematical
interpretation like ¬wantsp ∧ ¬¬[wantsp ∨ wants¬p] resulting in a deductively valid NR
55
inference wants¬p. We analyze the limited NR inferences ability of SL as a result of NR
inferences suspension: for English the suspension was observed by Gajewski 2007 (in case
the auxiliary carrrying the negation is stressed, the NR inference dissapears); for the
suspension in SL we assume that it takes place when speakers interpret the verbal negation
(systematically ambiguous between the propositional and constituent scope) with constituent
scope. Such interpretation then marks the sentence as an answer to a different QUD and the
EM alternative becomes irrelevant (→ not being exhaustified).
References
Bošković, Željko, Gajewski, Jon. (2009). Semantic Correlates of the NP/DP Parameter. In Proceedings of NELS
39.
Dočekal, Mojmír. (2014). Intervention Effects: Negation, Universal Quantifiers, and Czech. In Complex Visibles
Out There, edited by Markéta, Ludmila; Janebová Veselovská. Olomouc: Palacký university. 369–87.
Gajewski, Jon Robert. (2005). Neg-Raising: Polarity and Presupposition. Massachusetts Institute of Technology.
http://gajewski.uconn.edu/papers/thesis.pdf.
———. (2007). Neg-Raising and Polarity. Linguistics and Philosophy 30 (3): 289–328.
Homer, Vincent. 2011. Polarity and Modality. University of California Los Angeles.
http://www.linguistics.ucla.edu/general/Dissertations/HomerDissertationUCLA2011.pdf.
Horn, Laurence. (1989). A Natural History of Negation. Chicago: University Of Chicago Press.
Romoli, Jacopo. (2013). A Scalar Implicature-Based Approach to Neg-Raising. Linguistics and Philosophy 36
(4): 291–353.
56
The Relationship between (In)Alienable Possession
and Deverbal Nominalizers in Hungarian
Judit Farkasa and Gábor Albertib
a
Hungarian Academy of Sciences, Research Institute for Linguistics; bUniversity of Pécs,
Faculty of Humanities, Department of Linguistics
[email protected], [email protected]
1. This paper is primarily concerned with such Hungarian [(synchronic) stem + -j- +A]
possessive forms relative to which the same stem has an alternative, morphologically
“shorter”, possessive form. Such possessive forms are claimed to express alienable possession
(cf. den Dikken 2015). We argue that Hungarian deverbal nominals―and especially the
groups of T-nouns, defined in III below―are relevant to this claim, and corroborate it in a
certain sense.
2. In Hungarian there are some nouns which have two different stems in the possessive
paradigm: gyapja ‘wool.Poss.3Sg’, for instance, is an inflected version of gyapjú ‘wool’, and
it has an inalienable interpretation since the wool belongs to a sheep. If we emphasize that the
wool belongs to someone else, for instance, a shepherd, we use an alternative inflected form:
gyapjúja ‘wool.Poss.3Sg’; this reading relies on the so-called alienable interpretation (Kiefer
1985, Laczkó 2009, den Dikken 2015). In a similar fashion, the noun ablak ‘window’ also has
an inalienable and an alienable possessive form: ablak-a (that of a house) and ablak-j-a (that
of a distributor).
3. There is a parallelism between the results of a potential eventive T-nominalization
(producing ‘TEV-nouns’) and the complex-event denoting
ÁS-nominalization
(1a), on the one
hand, and there is a “complementary distribution” between those of a potential Theme
denoting T-nominalization (producing such ‘TTH-nouns’ as the one presented in (1c)) and the
“active key participant” denoting Ó-nominalization (1b), on the other. The possibility for the
latter relationship can be raised on the basis of the following analogy. The suffix -Ó is
primarily known as the present (or continuous / simultaneous / active) participial derivational
suffix in Hungarian grammar but it also functions as an “immediate” deverbal nominalizer
(Laczkó 2000). The suffix -(Vt)t can be regarded as its supplement on the basis of its
functioning as the past (or perfect / anterior / passive) participial derivational suffix in
Hungarian. It is not surprising, thus, that -(Vt)t is capable of deriving such noun phrases as the
57
one demonstrated in (1c) below, which denotes the “passive key participant”, namely the
Theme, of the input complex event.
(1) A potential system of ÁS-, Ó-, T- and HATNÉK-nominalization in Hungarian
a. Amerika (?)fel-fedez-t-é-vel/fel-fedez-és-é-vel
új korszak kezdődött.
America up-cover-T-POSS.3SG.INS/up-cover-ÁS-POSS.3SG.INS new age
begin.PAST.3SG
‘With America discovered, a new age has begun.’
b. Péter volt
Dóri fel-fedez-ő-je.
Péter be.PAST.3SG Dóri up-cover-Ó-POSS.3SG
‘Péter was the person who discovered Dóri.’
c. Dóri Péter fel-fedez-ett-je
volt.
Dóri Péter up-cover-T-POSS.3SG be.PAST.3SG
‘It was Dóri who was discovered by Péter.’
d. Ásítoz-hatnék-ja
van.
gape-HATNÉK-POSS.3PL be.3SG
‘(S)he has the urge to gape.’
The suffix -(Vt)t, however, is discussed neither as a Theme denoting nor as a complexevent denoting “immediate” deverbal nominalizer in the mainstream literature (Kiefer 2000).
This is not surprising, either, in the light of the numerous problems we will also take into
account in our talk. What is at stake here is whether the suffix -(Vt)t can be regarded as a
productive deverbal nominalizer (in present-day Hungarian) in either or both of its
aforementioned functions (or the existing
T-noun
constructions must be regarded as
lexicalized fossils). This depends on whether there are precisely definable (pragmaticosemantic and/or morphophonological) domains, however small they are (see Kiefer and
Ladányi 2000a:149, for instance), over which -(Vt)t performs as a “total function” in a
mathematical sense (i.e., in which it can be shown to be productive).
4. Let us now return to den Dikken’s (2015:138) claim on the potential morphemic
status of -j- within the possessedness suffix -(j)A in Hungarian nouns which (happen to) have
two possessed forms: “...Hungarian makes a morphosemantic distinction between alienable
and inalienable possession—a distinction that fits in with the linguistic universal ... [proposed
by Haspelmath (2008)], which says that, in languages that distinguish morphologically
between the two, alienably possessed DPs are morphologically richer than inalienably
possessed ones...”
We observed (see (1a,c) above) that if the third-person possessed form of a TEV-noun is
different from that of its
TTH-noun
counterpart, the additional -j- appears with the
TTH-noun,
where the possessor inevitably corresponds to the input Agent, and never with the
TEV-noun,
58
where the possessor primarily corresponds to the input Theme (or, “at most”, to a nonprototypical Agent who can be regarded as [+affected] in Dowty’s (1991) spirit in the absence
of an input object). The Agent is held to stand in a non-intrinsic relationship with the verb
(that corresponds to the output possessed noun), in contrast to the intrinsic relationship
between verbs and their Themes (Marantz 1984, Kratzer 1996). The appearance of the
additional -j-, thus, is associated with the less intrinsic semantic relationship, in harmony with
den Dikken’s (2015) thesis.
5. As for other types of deverbal nominalization, it is worth noting that all
ÓAG-noun
forms with a third-person possessor (1b) contain the -jA allomorph of the possessedness suffix
-(j)A in spite of the fact that this possessor corresponds to the input Theme. This fact, at first
glance, seems to serve as a counterexample to den Dikken’s (2015) thesis. That is not the
case, however. In these nouns, the appearance of -j- is motivated exclusively phonetically:
only the -jA allomorph of the possessedness suffix -(j)A can be attached to nouns ending in a
vowel, and Ó-nouns per definitionem end in a vowel. The -j- in question, thus, in the absence
of alternatives without a -j-, does not have the aforementioned “morphemic status”
responsible for the alienable interpretation according to den Dikken (2015). HATNÉK-nouns,
illustrated in (1d), however, can be characterized by the appearance of -j-, in total harmony
with the fact that the possessor of a
HATNÉK-noun
corresponds to the Agent of the verbal
derivational basis (and not the Theme).
References
den Dikken, Marcel. (2015). The morphosyntax of (in)alienably possessed noun phrases: The Hungarian
contribution. In AtoH 14. Amsterdam: Benjamins. 121–145.
Haspelmath, Martin. (2008). Syntactic universals and usage frequency, 3: Alienable vs. inalienable possessive
constructions. Handout, Leipzig Spring School on Linguistic Diversity.
Kiefer, Ferenc. (1985). Natural morphology. Acta Linguistica Hungarica 35: 85–105.
Kiefer Ferenc. (2000). Strukturális magyar nyelvtan 3, Morfológia. Bp.: Akadémiai.
Kiefer Ferenc, Ladányi Mária. (2000). A szóképzés. In: Kiefer (2000). 137–164.
Kratzer, Angelika. 1996. Severing the External Argument from its Verb. In: J. Rooryck, L. Zaring (eds.): Phrase
Structure and the Lexicon. Dordrecht: Kluwer. 109–137.
Laczkó Tibor. (2000). Az ige argumentumszerkezetét megőrző főnévképzés. In: Kiefer (2000). 293–407.
Laczkó, Tibor. (2009). Relational Nouns and Argument Structure – Evidence from Hungarian. In: Proc’s of the
LFG09 Conference. CSLI Publications, Stanford. 399–419.
Marantz, Alec. (1984). On the Nature of Grammatical Relations. Cambridge, MA: MIT Press.
59
Reversed Coordinated Multiple Wh-Questions in
Japanese
Yasuyuki Fukutomi
Fukushima University
[email protected]
This paper deals with a less-examined multiple wh-question in Japanese, which I refer
as a “Reversed Coordinated Multiple Wh-Question” and argues that it is derived from a
“concealed cleft” construction. It certainly constitutes a kind of multiple wh-question, since it
requires a wh-question as the first conjunct as in (1):
(1) a. nani-o
Mary-ni ageta no sosite dare-ga?
what-ACC Mary-DAT gave Q and who-NOM
Lit. ‘What gave to Mary and who?’
b. *ringo-o
Mary-ni ageta yo sosite dare-ga?
apple-ACC Mary-DAT gave PRT and who-NOM
Lit. ‘Gave an apple to Mary and who?’
Mary-ni ageta no sosite dare-ga?
c. *ringo-o
apple-ACC Mary-DAT gave Q and who-NOM
Lit. ‘Did give an apple to Mary and who?’
While (1a) includes a wh-question as the first conjunct, (1b) and (1c) include a
declarative sentence and a yes/no question respectively, resulting in unacceptable sentences.
Ishii (2014) argues convincingly that Japanese has a coordinated multiple wh-question (CWH)
with two wh-arguments, although it is a non-multiple-wh-language like English. He attributes
the difference between Japanese and English to the existence of a scrambling operation in
Japanese, which allows backward sluicing to obey the parallelism condition on deletion (Fox
and Lasnik 2003):
(2) a. dare-ga sosite nani-o
Mary-ni ageta no?
who-NOM and what-ACC Mary-DAT gave Q
Lit. ‘Who and what gave to Mary?’
b. *Who and what gave to Mary?
(3) [CP [CP dare-ga1 [TP(Elided Clause) nani-o2 [TP t’1 [vP Mary-ni t2 ageta]]] C1+Q1] [&P sosite [CP
[TP nani-o3 [TP(Antecedent Clause) t1 [vP Mary-ni t3 ageta]]] no]]
60
The required formal parallelism on deletion also accounts for the obligatory
application of scrambling to the second conjunct of CWHs, although we may not be able to
refer the example (4) as CWHs because two wh-arguments are not apparently conjoined:
(4) *dare-ga sosite Mary-ni nani-o
ageta no?
who-NOM and Mary-DAT what-ACC gave Q
Lit. ‘Who and gave what to Mary?’
Interestingly, when the order of the first and second conjuncts in (4) is reversed, the
sentence becomes acceptable:
(5) Mary-ni nani-o
ageta no sosite dare-ga?
Mary-DAT what-ACC gave Q and who-NOM
Lit. ‘What gave to Mary and who?’
The difference in acceptability between (4) and (5) indicates that the sentence with
forward deletion has a different derivation from that with backward sluicing; as is wellknown, the example in (5) could be derived from a “concealed cleft” construction (Saito 2003
and Takahashi 1994, among others):
(6) Mary-ni
nani-o ageta no sosite [CP Opi [TP ti Mary-ni pro ageta] no]-wa darei-ga
Mary-DAT
gave that-TOP who-NOM
In fact, a pronominal and a copular can appear in reversed CWHs (7), but not in
“ordinary” CWHs as in (8):
(7) Mary-ni nani-o
ageta no soiste (sore-wa) dare-ga da?
Mary-DAT what-ACC gave Q and
it-TOP who-NOM is
Lit. ‘What gave to Mary and is it who?’
(8) *(sore-wa) dare-ga da sosite nani-o Mary-ni
ageta no?
it-TOP who-NOM is and what-ACC Mary-DAT gave Q
Lit. ‘Is it who and what gave to Mary?’
In addition, the example in (9) is ungrammatical because the corresponding cleft
sentence is ruled out by Subjacency (10):
61
(9) *kinoo [NP nani-ga nusumareta to yuu nyuusu]-o kiita no sosite doko-kara?
yesterday what-NOM stolen-was that
news –ACC heard Q and where-from
Lit. ‘Yesterday what did you heae the news that was stolen and from where?’
(10) *[CP Opi [TP kinoo [NP ti genkin-ga nusumareta to yuu nyuusu]-o kiita] no]-wa Tokyo Ginkoo karai desyoo.
-TOP
from must-be
yesterday cash-NOM stolen-was that news-ACC heard
‘It must be from the Bank of Tokyo that you heard the news yesterday that cash was stolen.’
In contrast to the reversed counterparts, CWHs do not exhibit Subjacency effects,
because, as is well known, sluicing remedies Subjacency violations and makes the sentence
grammatical. The contrast between (9) and (11) corroborates the proposal that CWHs and
their reversed counterparts have different derivations.
(11)
doko-kara sosite nani-ga kinoo
nusumareta to yuu nyuusu-o kiita no?
where-from and what-NOM yesterday stolen-was that news-ACC heard Q
Lit. ‘From where and what did you hear the news that was stolen?’
The fact that CWHs involves a full wh-movement (cf. Takahashi 1993) and backward
sluicing is confirmed by their sensitivity to the superiority effect, as pointed out in Ishii
(2014):
(12)
?*
nani-o
sosite dare-ga Mary-ni ageta no?
what-ACC and who-NOM Mary-DAT gave Q
Lit. ‘What and who gave to Mary?’
(13)
dare-ga sosite nani-o Mary-ni ageta no?
who-NOM and what-ACC Mary-DAT gave Q
Lit. ‘Who and what gave to Mary?’
Compare (12) with (2a), repeated here as (13), in which dare-ga (who-NOM) sideward
moves to the SPEC of CP in the first conjunct, and it blocks the movement of nani-o (whatACC), resulting in the superiority effect. Here again, reversing the first and second conjuncts
ameliorates the acceptability of (12), since the corresponding reversed CWHs involve
concealed clefts as shown in (15):
(14)
a. dare-ga Mary-ni ageta no sosite nani-o
who-NOM Mary-DAT gave Q and what-ACC
Lit. ‘Who gave to Mary and what?’
b. nani-o
Mary-ni ageta no sosite dare-ga
what-ACC Mary-DAT gave Q and who-NOM
Lit. ‘What gave to Mary and who?’
62
(15) a. …sosite [CP Opi [TP pro Mary-ni ti ageta] no]-wa nanii-o
b. …sosite [CP Opi [TP ti Mary-ni pro ageta]no]-wa darei-ga
Furthermore,
the
proposed
analysis
through
concealed
clefts
accounts
straightforwardly for the lack of pair-list interpretation of reversed CWHs.
In summary, our proposal offers an additional evidence for “concealed clefts” analysis
of forward sluicing in Japanese and in tandem with Ishii’s (2014) analysis of Japanese CWHs
through backward sluicing, we argue that there are at least two kinds of deletion process in
Japanese syntax.
References
Fox, Danny and Lasnik, Howard. (2003). Successive-Cyclic Movement and Island Repair: The Difference
between Sluicing and VP-Elipsis. Linguistic Inquiry 34: 143–154.
Ishii, Toru. (2014). Coordinated Multiple Wh-Questions: A Crosslinguistic Perspective. Proceedings of 16th
SICOGG. 141–156.
Saito, Mamoru. (2004). Ellipsis and pronominal reference in Japanese clefts. Nanzan Linguistics 1: 21-50.
Takahashi, Daiko. (1993). Movement of wh-phrases in Japanese. NLLT 11(4): 655– 678.
Takahashi, Daiko. (1994). Sluicing in Japanese. Journal of East Asian Linguistics 3: 241–265.
63
On answers to negative questions
Hana Gruet-Skrabalova
Université Blaise Pascal, Clermont-Ferrand
[email protected]
Aim
In this contribution, we deal with answers to negative yes-no questions, focusing on
data from Czech. We show that there are two sets of answers to negative yes-no questions and
argue that their distribution depends on the interpretation of negation in the question. In
particular, we argue that negation in questions has either expletive or true interpretation (cf.
Brown & Franks 1995, Abels 2002), each of them triggering a different set of answers.
Data
A yes-no question is a question that asks to identify whether P or not P is true
(Hamblin 1973). It can be minimally answered by an answering particle, see (1), or, in some
languages (Basque, Irish, Finnish...), by using the finite verb of the question (Holmberg
2001), see Czech in (2). In all examples, yes and a positive verb indicate a positive answer (=
A1), no and a negative verb indicate a negative answer (= A2).
(1) a. Q: Did you send the letter to Paul?
b. Q: ¿Enviaste la carta a Paul?
c. Q: As-tu envoyé la lettre à Paul?
A1: Yes. // A2: No.
A1: Sì. // A2: No.
A1: Oui. // A2: Non.
(Spanish)
(French)
(2) Q: Poslal
jsi
Pavlovi ten dopis? A1: Ano./Poslal. // A2: Ne./Neposlal. (Czech)
sent-M.SG aux-2.SG Paul-DAT the letter-ACC yes sent-M.SG no neg-sent-M.SG
The same particles are used in answers to negative yes-no questions, see Spanish in
(3), although some languages have a specific particle for positive answers, like French in (4),
and some “reinforce” the positive particle (Pope 1972), like English in (5), where yes
combines with a positive tag, or Czech in (6), where yes combines with the conjunction ale
(‘but’).
(3) Q: ¿No enviaste la carta a Paul?
(4) Q: N’as-tu pas envoyé la lettre à Paul?
(5) Q: Didn’t you send the letter to Paul?
(6) Q: Neposlal jsi Pavlovi ten dopis?
A2: No. // A1: Sì.
A2: Non. // A1: *Oui./Si.
A2: No. // A1: Yes, I did.
A2: Ne (neposlal). / Ale ano (poslal).
64
(Spanish)
(French)
(Czech)
What we are interested here are negative yes-no questions as in (7), that can be
answered in two ways, which we call type A and type B answers. In type A answers,
corresponding to answers in (3) to (5) above, ne expresses a negative answer, while (ale) ano
expresses a positive answer. In contrast, in type B answers, ne expresses a positive answer,
while ano expresses a negative answer. (These two types of answers can also be found in
Spanish though to less extent.)
(7) Q: Rodiče nejsou doma?
parents neg-are home
‘Aren’t the parents at home?’
A2: Ne (nejsou). // A1: Ale ano (jsou). (Czech)
no neg-are
but yes are
B2: Ano (nejsou). // B1: Ne (jsou).
yes neg-are
no are
(they are not at home) (they are at home)
Type B answers in Czech can be both positive and negative, and can be used as replies
to Qs in which sentential negation licenses NPIs, see (8). We claim therefore that they differ
from negative questions in English that can be “confirmed” by yes and that are analysed by
Holmberg (2001, 2013) as positive yes-no questions containing VP-negation, see (9).
(8) Q: Jan *(ne)pozval ani jednoho?
John (neg)-invited no one
(9) Q: Is John not coming?
B2: Ano (ani jednoho). // B1: Ne (jednoho pozval).
yes no one
no one (he) invited
B2: Yes (he is not).
The puzzle we are dealing with here is not only why both particles can express positive
and negative answers to questions as in (7), but also why type B answers are not always
available, as for instance in (6) above:
(6) Q: Neposlal jsi Pavlovi ten dopis?
B2: *Ano (neposlal). // B1: *Ne (poslal).
Analysis.
To explain the distribution of type B answers, we first deal with negative questions.
We show that there are two major types of negative yes-no questions (with several semantic
subtypes, cf. Brown & Franks 1995): negative interrogative clauses and negative declarative
clauses with rising intonation.
From a syntactic point of view, negative interrogatives typically have a VSO order and
do not allow presence of negative words (though Czech is a language with negative concord),
while negative declaratives clauses have a declarative word order (SVO, SOV) and allow
presence of negative words, see (10) and (11) respectively.
65
(10) Q: Nepotřebuje někdo/*nikdo pomoct?
neg-needs somebody / nobody help
(11) Q: Nikdo/Někdo nepotřebuje pomoct?
nobody/somebody neg-needs help
From a semantic point of view, negative interrogative clauses are used when the
speaker is supposed to be neutral with respect to the proposition conveyed by the question. In
contrast, negative declarative clauses cannot be used in contexts where the speaker is
supposed to be impartial or ignorant, because they convey negative presupposition (cf. rising
declaratives in Gunlogson 2001). Importantly, these two types of questions correlate with
possible answers.
Negative interrogatives receive only type A answers, behaving
therefore like positive interrogatives. We argue that this is due to the expletive interpretation
of the negation (Espinal 1992, Brown & Franks 1995, Abels 2002). We propose that
sentential negation “becomes expletive” when the verb moves to the CP domain, because its
negative force is “neutralised” by an abstract interrogative operator. This operator also blocks
the reconstruction of the negation, preventing it from licensing negative phrases like nikdo
(‘nobody’) at LF.
Negative declaratives can receive type B answers and behave therefore like negative
assertions containing true negation. They are used to elicit (dis)agreement with the negative
presupposition they convey. We propose that particles in type B answers (dis)agree with the
polarity of the question, so express not absolute but relative polarity (cf. Farkas 2010). When
negative declaratives receive type A answers, we claim that the verb moves to the CP at LF,
triggering again “expletive” interpretation of the negation.
References
Abels, K. (2002). Expletive (!) Negation. In J. Toman (ed.), Proceedings of FASL 10: 1–20.
Brown, S. & S. Franks. (1995). Asymmetries in the scope of Russian negation. Journal of Slavic Linguistics
3(2): 239–287.
Espinal, M. T. (1992). Expletive Negation and Logical Absorption. Linguistic Review 9: 333–358.
Farkas, D. (2010). The grammar of polarity particles in Romanian. In A. Di Sciullo & V. Hill (eds.) Edges,
Heads, and Projection: Interface properties, Amsterdam: Benjamins. 81–124.
Gunlogson, C. (2001). True to form: Rising and Falling Declaratives as Questions in English. PhD diss. Univ. of
California at Santa Cruz.
Hamblin, C. L. (1973). Questions in Montague English. Foundations of Language 10: 41–53.
Holmberg, A. (2001). The syntax of yes and no in Finnish. Studia linguistica 55: 141–174.
Holmberg, A. (2013). The syntax of negative questions and their answers. In N. Gato et al. (eds), Proceedings of
GLOW in Asia 9: 1-18.
Pope, E. (1972). Questions and answers in English. PhD. Diss. MIT.
66
Resumption in Slavic relative clauses
Marko Hladnik
Utrecht University
[email protected]
Introduction
When we consider the four possible relative clause configurations varying in the
choice of the relative element (either a pronoun or a complementizer) and the presence or
absence of resumption, we see that all of them are in fact attested in Slavic languages. Such a
situation may give an initial impression of arbitrariness in the use of resumption and
contribute to certain inaccurate characterizations of the constructions in the literature. I tease
apart the properties of the different constructions, establish under what conditions they are
available, and provide a derivation analysis for each of them.
I argue that i) the two (well-known) superficially different relativizing strategies in
Slavic (see (1) and (2)) should receive the same underlying syntactic analysis, ii) that an
apparent optionality in the use of resumption is not due to processing differences but rather
due to differences in the underlying syntax, and iii) that another (hitherto unidentified) type of
resumption that arises as a result of processing, non-syntactic constraints should be
distinguished.
Two common strategies
Two common ways of constructing RCs in Slavic are here illustrated by Slovene
examples: RCs can be either introduced by a pronoun (1), or by a complementizer (2), in
which case a resumptive clitic pronoun is present as well.
(1) Poznam človeka, katerega so
iskali.
Know-1SG man-ACC which-ACC AUX-3.PL searched
(Slovene)
(2) Poznam človeka, ki so
*(ga)
iskali.
Know-1SG man-ACC C AUX-3.PL he-ACC.CL searched
Both: ‘I know the man they were looking for.’
I argue that resumption in Slovene relative clauses cannot be explained by stipulating
the presence of a syntactic island, and that a different mechanism is at work (see below). This
follows from the observations that i) resumption in RCs functions as a primary strategy, not
associated with any reduced grammaticality, and ii) resumption in Slovene in general cannot
67
be used as a last resort to repair illicit extractions. This argument is strengthened by the fact
that both types of relative clauses become ungrammatical when we try to relativize out of
known island configurations; the presence of resumption in (3) cannot obviate the movement
restriction. The same sensitivity is demonstrated for Bosnian/Croatian/Serbian (B/C/S) and
Polish as well (cf. Bošković 2009, LaTerza 2013).
(3) *človek, ki je
Janez jezen, ker
ga
je
Peter odpustil
man C AUX-3SG Janez angry because he-ACC.CL AUX-3SG Peter fired
Intended: ‘the man that John is angry because Peter fired’
This sensitivity to islands is taken as an indication that both (1) and (2) involve
wh-movement. I argue that in fact the two strategies of relativization share one and the same
underlying syntax, and are formed by the relative pronoun moving to Spec,CP. The
configuration can, however, be spelled out in two different ways, depending on the
(essentially free) choice between either the pronoun or the complementizer being overt.
Resumption arises at spell-out to ensure the recoverability of the case information pertaining
to the relativized position. Case morphology is spelled out on the relative pronoun when
overt, giving rise to (1). However, when the complementizer is overt, the pronunciation of the
relative pronoun is suppressed (since only one of the two elements can be overt at the same
time). In that situation, case is expressed by the resumptive pronoun as a partial spell-out of a
lower copy of the relative pronoun instead, resulting in (2).
Apparent optionality of resumption
I adopt a syntactic structure where relative clauses are complements to an external D
head, with the RC head located within the relative clause in the highest specifier of a split CP
(cf. Kayne 1994, Bianchi 1999, Aoun & Li 2003, Boef 2013, a.o.). The derivation itself can
proceed in one of two ways. The first one involves base-generating the RC head in its
specifier position, while the dependency is created by moving a relative pronoun to a lower
Spec,CP. This derivation underlies the examples in (1-2) discussed above. The alternative
derivation employs RC head raising from the relativized position to the highest Spec,CP. The
availability of two different derivations explains the variation in complementizer RCs
exemplified by (4a-b), where resumption seems to be optional. I argue that while they appear
superficially similar, the two examples are derived each in its own way.
(4) a. Ovo je auto
što sam
ga
kupio.
this is car-NOM C AUX.-1SG he-ACC.CL bought
68
(B/C/S)
b. Ovo je auto
što sam
kupio.
this is car-NOM C AUX-1SG bought
Both: ‘This is the car that I bought.’
The derivation of (4a) involves pronoun movement. The pronoun remains silent, which
triggers resumption, as above. Example (4b), on the other hand, is derived by RC head
raising. Since no relative pronoun is part of the derivation, the RC in (4b) contains no
resumption. Another difference between the two derivations is that only the raising derivation
features a representation of the RC head at the relativization site, predicting the presence of
reconstruction effects in examples such as (4b) but not in examples such as (4a) or in pronoun
RCs, which is borne out.
The type of construction in (4b) is only available under two necessary conditions (cf.
Gračanin-Yuksek 2010). Firstly, the case morphology of the RC head must match both the
case assigned by the matrix predicate as well as the case assigned at the relativization site.
Secondly, the case assigned at the relativization site must not be inherent. I show how these
conditions follow from the details of the adopted raising analysis, which involves the
incorporation of an internal determiner (as part of the RC head) into the external D (cf.
Bianchi 2000). The analysis, interacting with the specific morphosyntactic properties of the
language, extends to explain RC data from Bulgarian (cf. Krapova 2010) as well.
Processing-driven resumption
Thus far we have only discussed uses of resumption in relative clauses that are a result
of morphosyntactic requirements, and as such absolute in nature. However, resumption driven
by processing considerations can be identified as well, its properties contrasting starkly with
syntax-driven resumption. It is attested in long distance relative clauses introduced by a
pronoun, as exemplified in (5).
(5) Poznam človeka, katerega mislim, da (ga)
iščejo.
Know-1SG man-ACC which-ACC think-1.SG that he-ACC.CL search-3.PL
‘I know a man who I think they are looking for.’
Based on i) the non-obligatory nature of such resumption, ii) the fact that it more
readily appears in spoken varieties, and iii) the fact that is sensitive to the complexity and
length of the dependency, I conclude that this type of resumption (involving a partial spell-out
of a lower copy) is driven by processing considerations and facilitates the formation of the
long-distance relative dependency by marking the relativization site explicitly by overt
material. The prediction that the acceptability of such resumption should therefore increase
69
with the number of embedded clauses intervening between the relative pronoun and the
relativization site (and differ from the acceptability patterns of the morphosyntactic
resumption type, which is obligatory and not sensitive to dependency length) has been
confirmed by a survey involving speakers of Slovene, Polish, and B/C/S.
Conclusion
This paper connects the dots of disparate and partial empirical observations in the
literature, and together with original findings ties them into a coherent account of the rather
complex situation concerning resumption patterns in Slavic relative clauses. It teases apart the
syntax-driven use of resumption from the processing-driven one, and demonstrates the
necessity of maintaining a distinction between the different types.
References
Aoun, J., and Y.A. Li. (2003). Essays on the Representational and Derivational Nature of Grammar. The
Diversity of Wh-constructions. Linguistic Inquiry Monographs 40. Cambridge, MA: MIT Press.
Bianchi, V. (1999). Consequences of Antisymmetry. Berlin: Mouton de Gruyter.
Bianchi, V. (2000). The Raising Analysis of Relative Clauses: A Reply to Borsley. Linguistic Inquiry 31: 123–
140.
Boef, E. (2013). Doubling in Relative Clauses: Aspects of Morphosyntactic Microvariation in Dutch. Doctoral
Dissertation, Utrecht University.
Bošković, Ž. (2009). On relativization strategies and resumptive pronouns. FDSL 7 Proceedings. 1-13.
Gračanin-Yuksek, M. (2010). On a Matching Effect in Headed Relative Clauses. In Wayles, B., Cooper, A.,
Fisher, A., Kesici, E., Predolac, N., and Zec, D. (eds.) Formal Approaches to Slavic Linguistics 18,
Michigan Slavic Publications, Ann Arbor. 193–209.
Kayne, R.S. (1994). The Antisymmetry of Syntax. Linguistic Inquiry Monographs 25. Cambridge, MA: MIT
Press.
Krapova, I. (2010). Bulgarian Relative and Factive Clauses with an Invariant Complementizer. Lingua 120:
1240–1272.
LaTerza, I. (2013). The Case of Resumption in Serbian Relative Clauses. Ms., Stony Brook University.
70
Tense and Modality in the Nominal Domain
Negin Ilkhanipour
Department of Linguistics, University of Tehran
[email protected], [email protected]
It is well discussed in the literature that epistemic modals are base-generated higher
than non-epistemic/root modals in the clausal structure (Cinque 1999, 2013; Butler 2003;
Portner 2009; Hacquard 2010; among many others). More precisely, it has been argued that
epistemic modals (Modepis) are higher than Tense (T), while root modals (Modroot) are lower
than T, and that high modals are evaluated in the context of the speech event (i.e. with regard
to the speaker at the speech time), whereas low modals are evaluated in the context of the VP
event (with regard to an argument at the event time). This is illustrated in (1).
(1) [AssertP [ModepisP [TP [ModrootP [VP ...
Speech event
VP event
In this study, looking with favour upon Lecarme’s (2004) and Nordlinger & Sadler’s
(2003, 2004) claim for tense and modal functional projections in the nominal domain, and
following Cinque’s (1994, 2002, 2010) and Scott’s (2002) proposal that adjective phrases are
base-generated in the specifiers of distinct functional projections, I argue that, similar to the
structure of CP in (1), epistemic and non-epistemic modal elements have different positions in
DP; epistemic adjectives appear in the specifier of ModN.episP above nominal tense (TNP),
while non-epistemic adjectives appear in the specifier of ModN.rootP below TNP, where
nominal tense is the time of the existence or occurrence of the modified noun. This is shown
in (2).
(2) [AssertNP [ModN.episP [TNP [ModN.rootP [NP ...
Speech event
NP event
With this aim in view, firstly, I show that the ambiguity of (3) in Persian is due to the
two possible distinct positions of the adjective qæbli ‘previous’: the specifier of TNP in
context A, as in (4a), and the specifier of ordinalP in context B, as in (4b).
71
(3) bærænde-ye qæbli
winner-EZ previous
‘the previous winner’
Context A
The host of a quiz show, talking about the winner of the previous round, says: “bærænde-ye
qæbli tehrani bud” ‘The previous winner was from Tehran’.
Context B
In a quiz show with two winners, the host introduces first winner 1 and then winner 2. While
introducing winner 2, the host says: “Ɂin bærænde mesl-e bærænde-ye qæbli tehrani-e” ‘This
winner, like the previous winner, is from Tehran’.
(4a)
qæbli
TNP
(4b)
TN’
TN
NP
bærænde
ordinalP
qæbli
ordinal
ordinal’
TNP
bærænde
Next, I explain that this structural ambiguity of qæbli ‘previous’ is observed when it
co-occurs with Persian non-epistemic adjectives, such as qabel-e-ɁeɁtemad ‘reliable’, as in
(5a) where both contexts A and B are possible (Importantly, the DP is acceptable in context
A). This suggests that the position of the non-epistemic adjective is lower than TN, where it is
interpreted with respect to the modified noun bærænde ‘winner’ at the time of the quiz show
(i.e. NP event). This is illustrated in (6a). It is also discussed that in (5b) the modal adjective
has moved to the specifier of FocusP that brings about the marked reading, as in (6b).
(5a) bærænde-ye qabel-e-ɁeɁtemad-e qæbli
winner-EZ reliable-EZ
previous
the previous reliable winner’
(5b) ?bærænde-ye qæbli-ye
qabel-e-ɁeɁtemad
winner-EZ previous-EZ reliable
‘?the reliable previous winner’
72
(6a)
ordinalP
(6b)
FocusP
[qæbli] ordinal’
qabel-e-ɁeɁtemad Focus’
ordinal
TNP
Focus
ordinalP
[qæbli] TN’
[qæbli]
ordinal’
TN
ModN.rootP
ordinal
TNP
qabel-e-ɁeɁtemad
ModN.root’
[qæbli] TN’
ModN.root
NP
TN
ModN.rootP
bærænde
(qabel-e-ɁeɁtemad)
ModN.root’
ModN.root NP
bærænde
With epistemic adjectives (e.g., Ɂehtemali ‘probable’), as in (7a), the adjective qæbli
‘previous’ is not ambiguous, though; that is, it can be interpreted only as in context B and can
be base-generated only in Spec,ordinalP. On the other hand, merging Ɂehtemali ‘probable’
below TN brings about ungrammaticality, as in (7b). This implies that the epistemic modal is
higher than TN, where it is evaluated with regard to the speaker’s knowledge at the speech
time. This is illustrated in (8).
(7a) bærænde-ye Ɂehtemali-ye qæbli
winner-EZ probable-EZ previous
‘the previous probable winner’
(8)
(7b) *bærænde-ye qæbli-ye Ɂehtemali
winner-EZ previous-EZ probable
intended: ‘the previous probable winner’
ordinalP
qæbli
ordinal’
ordinal
ModN.episP
Ɂehtemali
ModN.epis’
ModN.epis
TNP
bærænde
Thus, we see that the interaction of temporal and modal adjectives in DP provides
evidence for the structural hierarchy proposed in (2), which is parallel to its counterpart at the
clausal level.
References
Butler, J. (2003). A minimalist treatment of modality. Lingua 113: 967–996.
Cinque, G.�(1994)- Evidence for partial N-movement in the Romance DP. In G. Cinque, J. Koster, J. Pollock, L.
Rizzi, and R. Zanuttini (eds.) Paths towards universal grammar: Studies in honor of Richard S.
Kayne. Washington, D.C.: Georgetown University Press. 85–110.
Cinque, G. (1999). Adverbs and functional heads: A crosslinguistic perspective. New York and Oxford: OUP.
Cinque, G- (2002). Mapping functional structute: A project. In G. Cinque (ed.) Functional structure in DP and
IP: The cartography of syntactic structures. Oxford: OUP. 3–11.
Cinque, G. (2010). The syntax of adjectives: A comparative study. Cambridge, MA: MIT Press.
73
Cinque, G. (2013). Cognition, universal grammar, and typological generalizations. Lingua 130: 50–65.
Hacquard, V. (2010). On the event relativity of modal auxiliaries. Natural Language Semantics 18(1): 79–114.
Lecarme, J. (2004). Tense in nominals. In J. Guéron and J. Lecarme (eds.) The Syntax of Time. Cambridge, MA:
MIT Press. 440–475.
Nordlinger, R., & L. Sadler. 2003. The syntax and semantics of tensed nominals. In M. Butt and T. Holloway
King (eds.) Proceedings of LFG03 Conference. State University of New York at Albany. 328–346.
Nordlinger, R., and L. Sadler. (2004). Tense beyond the verb: Encoding clausal tense/aspect/mood on nominal
dependents. Natural Language and Linguistic Theory 22: 597-641.
Portner, P. (2009). Modality. Oxford: OUP.
Scott, G. (2002). Stacked adjectival modification and the structure of nominal phrases. In G. Cinque (ed.)
Functional structure in DP and IP: The cartography of syntactic structures. vol. 1. Oxford: OUP.
91–20.
74
Agreement-by-Correspondence with spreading
Peter Jurgec
University of Toronto
[email protected]
75
76
77
The connection between metaphony and unstressed
vowel reduction in Italian dialects
Martin Krämer
University of Tromsø
[email protected]
In the most pervasive type of metaphony in Italian dialects, a post-tonic high vowel
causes raising of the stressed vowel in a word (Maiden 1991, Krämer 2009, Calabrese 2011).
Dialects differ in the behaviour of intervening vowels, in which stressed vowels allow raising,
and whether they only raise or diphthongize as well. Dialects differ as well with respect to the
fate of the triggering vowel. It may be realized faithfully, reduced to schwa or deleted. A
chain shift can be observed in the targeted vowels in two ways. In the most extensive system,
that of the Ischia dialect (Maiden 1991), stressed /a/ raises to [ɛ], /ɛ/ to [e], and /e/ raises to [i],
preceding an unstressed high vowel. Regarding the targets of raising, Maiden (1991)
identified an implicational hierarchy: if a lower vowel raises the higher vowels raise as well,
but not vice versa.
(1)
Metaphonic chain shift typology (ignoring diphthongisation)
/ˈe/ → [ˈi]/__C(CVC)i,u#
Veneto
Southern
/ˈɛ/ → [ˈe]/__ C(CVC)i,u#
Umbro
/ˈa/ → [ˈɛ]/__ C(CVC)i,u#
Ischia
It is to date by no means clear which phonological process underlies the raising
pattern, change by spreading or copying and how to analyse the chain shift nature of the
process.
In this paper I argue that the implicational nature of the typology, that is, that lax mid
vowels only raise where tense mid vowels do, and low vowels only raise where lax mid
vowels do, is partially conditioned by constraint rankings that are necessary for the
neutralisation of the tenseness contrast in mid vowels in unstressed position (mid vowel
tensing) which is also active in these varieties (illustrated with Italian examples in 2).
78
(2)
Unstressed vowel neutralisation in Standard Italian
a. ortopeˈdiːa
‘orthopaedics’
b. ˈlɔdʒika
ortoˈpɛdiko
‘orthopaedist/
lodʒikaˈmente
orthopaedic’
apˈpɛndo
‘hang (1sg)’
ˈkɔːpri
appenˈdjaːmo ‘hang (1pl)’
koˈpriːte
ˈbɛːne
‘well’
ˈpɔrto
beneˈfiːtʃo
‘benefit’
porˈtjaːmo
‘logics’
‘logically’
‘cover (2sg)’
‘cover (2pl)’
‘carry (1sg)’
‘carry (1pl)’
Walker (2011) categorizes vowel patterns as direct licensing, indirect licensing or licensing by identity. In direct licensing patterns, certain contrasts are only permitted in certain
prominent positions (e.g., Italian unstressed mid vowel tensing). Indirect licensing is the case
if a feature spreads to a prominent position to be realised in its weak position of origin.
Identity licensing does not involve spreading but rather the establishment of a correspondence
relation (McCarthy & Prince 1995) between feature bearers in the surface representation
(Krämer 2003, Walker 2011). It is not immediately clear which of the two latter mechanisms
is at work in the above metaphony patterns, spreading or correspondence. Consideration of
the additional pattern of unstressed mid vowel tensing provides the arguments to decide this
issue.
The assumption that mid vowel tensing is markedness reduction and that reduction is
achieved by feature delinking (i.e., the lax mid vowels bear a feature specification that tense
mid vowels lack, and which is removed in unstressed position to decrease markedness) leads
to the conclusion that metaphony is not feature spreading from the final high to the stressed
vowel (e.g., to achieve some kind of licensing in a prominent position). Tense mid vowels are
derived from lax mid vowels both in metaphony and in unstressed vowel reduction. However,
in metaphony we expect the mechanism to be spreading of the marked feature while in
reduction it is assumed to be delinking of the marked feature. These two operations can’t be
used to derive the same surface segments, i.e., [e, o], from the same underlying segments, i.e.,
/ɛ, ɔ/.
(3)
/ɛ/ → [e]
a. Unstressed syllable reduction
x
=
[F]
b. Metaphony
x
y
[F]
Thus, metaphony has to be an effect of surface correspondence between the stressed
and the post-stress vowels, rather than feature spreading.
79
Lastly, from the vantage point of Optimality Theory the question arises why not all
vowels simply raise to high. These apparently opaque patterns, or cases of Derived Environment Blocking, will be accounted for here in a parallelist fashion, combining Syntagmatic Correspondence (Krämer 1998, 2001, 2003) with Local Conjunctions (henceforth
LC; Prince & Smolensky 1993/2004, Smolensky 1993, Kirchner 1996, Crowhurst & Hewitt
1997, Moreton & Smolensky 2002, Lubowicz 2002, Krämer 2003). While earlier approaches
use Local Conjunctions in an opportunistic way (usually some constraint C1 is conjoined with
some other constraint C2 into {C1&C2} to achieve some desired effect, regardless of the
nature of the constraints), in this paper we will explore a more principled approach to LC,
based on Kirchner’s (1996) account of one-step vowel raising as LCs of Faithfulness
constraints (F&F LCs). We will consider all LCs of all faithfulness constraints on height and
a.
 b.
 c.
 d.
e.
 f.
 g.
h.
i.
/e-i/ – e-i
/e-i/ – i-i
/ɛ-i/ – ɛ-i
/ɛ-i/ – e-i
/ɛ-i/ – i-i
/a-i/ – a-i
/a-i/ – ɛ-i
/a-i/ – e-i
/a-i/ – i-i
ID[hi]
ID[RTR]
ID[lo]
S-ID
[h/l/R]
ID[lo]&
ID[RTR]
ID[hi]&
ID[lo]
(4)
ID[hi]&
ID[RTR]
their effect, as shown in the tableau in (4).
*
*
**
*
*
*
*
*
*
***
**
*
*
*
*
*
*
*
*
*
*
In conclusion, examining one phenomenon in isolation, such as metaphony, is not
particularly insightful. Taking into consideration additional properties of the system, such as
unstressed mid vowel reduction, leads to further insights into the originally explored pattern.
References
Andrea Calabrese. (2011). Metaphony in Romance. In C. Ewen, M. van Oostendorp, & B. Hume, (eds.), The
Blackwell Companion to Phonology. Wiley-Blackwell.
Crowhurst, Megan, and Marc Hewitt. (1997). Boolean operations and constraint interactions in Optimality
Theory. Ms., University of North Carolina at Chapel Hill and Brandeis University.
Kirchner, Robert. (1996). Synchronic chainshifts in Optimality Theory. Linguistic Inquiry 27: 341–350.
Krämer, Martin. (1998). A Correspondence Approach to Vowel Harmony and Disharmony. SFB 282 Working
Paper Nr. 107. Heinrich-Heine-Universität Düsseldorf.
— (2001). Yucatec Maya Vowel Alternations – Harmony as Syntagmatic Identity. Zeitschrift für
Sprachwissenschaft 20(2): 175–217.
— (2003). Vowel Harmony and Correspondence Theory. Studies in Generative Grammar 66. Berlin, New York:
Mouton de Gruyter.
80
— (2009). The Phonology of Italian. Oxford University Press.
Łubowicz, Anna. (2002). Derived Environment Effects in Optimality Theory. Lingua 112: 243–280.
Maiden, Martin. (1991). Interactive Morphonology. Metaphony in Italy. London: Routledge.
McCarthy, John J. & Alan S. Prince. (1995). Faithfulness and reduplicative identity, UMOP 18: 249-384.
Moreton, Elliot and Paul Smolensky. (2002). Typological Consequences of Local Constraint Conjunction.
WCCFL 21.
Prince, Alan S. and Paul Smolensky (1993/2004). Optimality Theory: Constraint interaction in generative
grammar. Technical Report of the Rutgers Center of Cognitive Science, RUCCS-TR 2, Rutgers
University, New Brunswick and Technical Report of the University of Colorado Computer Science
Department, CU-CS-696-95, University of Colorado, Boulder. Publication in 2004: Malden (Mass.),
Oxford: Blackwell Publishers.
Smolensky, Paul. (1993). Harmony, markedness, and phonological activity. Paper presented at Rutgers
Optimality Workshop-1, Rutgers University, New Brunswick, NJ. [Rutgers Optimality Archive,
http://ruccs.rutgers.edu/roa.html.]
Walker, Rachel. (2011). Vowel Patterns in Language. Cambridge University Press.
81
Alignment and locality in the typology of language
games
Martin Krämer a, Barbara Vogt b
a
University of Tromsø, bUniversità di Trieste
a
[email protected] ,[email protected]
In secret languages, word games, language games or ludlings (Laycock 1972)
(henceforth LGs) existing words of a given language are concealed by inserting, exchanging,
switching or repeating phonological material (see the examples in 1 to 3). LGs are often used
in linguistic analysis as they are considered to use specifically linguistic, not "metalinguistic", tools and units. (Compare e. g. Bertinetto 1987 on a LG in Italian, Piñeros 1998 on
LGs in Spanish as well as Davis 1993, Bagemihl 1995, Botne & Davis 2000, Vaux 2011,
Borowsky 2012 for overviews). The hypothesis is that LGs “extend, modify, or exaggerate
attested natural language processes” (Bagemihl 1995) which is demonstrated among other
things by the fact that not all conceivable language disguises are found. There are clear limits
in variation and these limits are shaped by constraints relying on specifically linguistic
representations and processes: For example, LG normally do not require the breaking of the
subsyllabic constituents onset, coda or nucleus, there are no “rules” based on a count of single
sounds or of syllables. On the other hand, infixed dummy affixes and reduplicants seem to be
anchored to subsyllabic constituents, splitting all syllables in a word between onset and
nucleus or between nucleus and coda (1a,b) or surfacing before or after each syllable (2a,b) or
next to the stressed vowel (3).
(1a)
(1b)
(2a)
(2b)
(3)
German:
Dutch:
French:
Spanish:
English:
Karten
en mannetje
Je suis jeune
adios
missed the bastard
→ kabarteben
→ epen mepannepetjepe
→ feje fuisuis feujeufene
→ akamadikimioskomo
→ milfíssed the balfástard
‘cards’
‘a man(dim)’
‘I am young.’
‘good-bye’
This led researchers to posit language game-specific alignment constraints (e.g., Yu 2008).
In this talk we will provide a typology of LGs and an optimality-theoretic analysis in
which the landing sites of LG dummy affixes and reduplicants are explained as a side effect
of general, independently motivated, phonotactic markedness constraints (such as ONSET,
*CODA, *COMPLEX etc.). We only make use of three very general Alignment/Anchoring
82
constraints, which require base material to be at the left edge, and the affix/reduplicant to be
at the left/right edge of the LG form, respectively.
Our data are taken from a corpus currently containing 87 LGs covering 42 structurally
different types of LGs, from 12 languages (also non-Indo-European), extracted from the
linguistic literature and our own research, which were tagged according to syllable size and
structure, processes and segmental make-up. We will concentrate on the processes involved
(reduplication, affixation and truncation) and analyze possible locations for inserted and/or
reduplicated strings in the derived (manipulated) form. Transpositions are not considered
here.
In our analysis we follow Alber’s (2010) and Alber & Arndt-Lappe’s (2012) work on
anchoring in truncation: In their analysis ANCHOR-L, the constraint responsible for
anchoring a truncation morpheme to the left edge of the base, is best defined as an alignment
constraint while ANCHORSTRESS, the constraint responsible for anchoring the truncation
morpheme to the stressed syllable of the base, should be defined as a faithfulness constraint.
In our analysis, the constraints placing dummy affixes and reduplicants are of the Alignment
type and the constraint placing the base or its initial material at the left edge is an Anchoring
constraint (see McCarthy & Prince 1994, 1995 for definitions of the two families). In LGs, the
Faithfulness constraint locating the dummy/reduplicant near base stress is BR-MAX-V́ (‘A
vowel that is stressed in the Base has to be present in the Reduplicant’). The analysis of the
data in language games by and large confirms this approach to anchoring: at the left edge
typical alignment effects are visible, while in stress anchored LGs the manipulations are
analyzable as a form of stress-preservation.
In the latter case, BR-MAX-V́ conspires with BR-LOCALITY (‘Assign one * for every
segment between correspondent segments in Base and Reduplicant.’ Based on Riggle 2003,
2006) to place the LG affix near the stressed base vowel and to exclude reduplication of other
material from the base. This is illustrated in (4).
(4)
/bá1sta2rd –REDlf/
BR-MAX-ˈV
LOCALITY
BR-MAX-V
*!
**
*
*
a.
ba1sta2rda2lf
b.
ba1sta2rda1lf
***!**
b.
ba1lfa1sta2rda2lf
**,*!*
 c.
ba1lfa1stard
**
83
*
Iterativity is also an epiphenomenon of Locality. The more general BR-MAX-V and
LOCALITY outrank Markedness (which constrains the range of segments in the reduplicant),
forcing infixation of the reduplicative LG affix as near as possible to every vowel in the base.
The quality of these reduplicated vowels is also often restricted, leading to emergent fixed
segmentism (1b).
As in grammatical reduplication (Alderete et al 1999), fixed segmentism has two
origins, either the above Emergence of The Unmarked (TETU) effect or lexical specification
of phonologically marked segment(s), e.g., the /p/ in (1b). The tableau in (5) summarizes the
analyses of iterativity, TETU and lexically fixed segmentism.
(5)
/mannetje –REDp/
ONSET
ALIGN
LRED
LOCALI
**
**
*
TY
*MAR
KV
BRIDENT(F)
a.
epmannetje
b.
mepannetje
*
*!*
*
*
c.
mepannetjepe
*
*!
**
*
d.
mapannepetjepe
*
***
 e.
mepannepetjepe
*
***
m-ennetjep-annetje
*
20
f.
*!
BRMAX-V
*!
*
In conclusion we account for the following characteristics of LGs, infixation,
iterativity, fixed segmentism, and affix placement without recourse to any LG-specific
constraints and show that the typology of LGs emerges from a creative use of independently
motivated constraints.
References
Alber, B. (2010). An Exploration of Truncation in Italian. Rutgers Working Papers in Linguistics vol. 3: 1–30.
Alber, B. and S. Arndt-Lappe. (2012). Templatic and Subtractive Truncation. In J. Trommer (ed.). Phonology
and Morphology of Exponence - the State of the Art. OUP.
Alderete, J., J. Beckman, L. Benua, A. Gnanadesikan, J. McCarthy and S. Urbanczyk. (1999). Reduplication
with Fixed Segmentism. Linguistic Inquiry 30: 327–364.
Bagemihl, B. (1995). Language Games and Related Areas. In J. Goldsmith (ed.): The Handbook of Phonological
Theory. Oxford: Blackwell. 697–713.
Bertinetto, P. M. (1987). Lingue segrete, e segreti delle lingue. Alcuni probleme di fonologia italiana studiati
attraverso un gioco linguistico, in Annali della Scuola Normale Superiore di Pisa, serie III, 17: 889–
920.
Borowski, T. (2012). Language Disguise in OT. Reversing and Truncating. Indigenous Language and Social
Identity, in Pacific Linguistics 626, 367-385
Botne, R. and S. Davis. (2000). Language games, segment imposition, and the syllable. Studies in Language 24:
319-344.
Davis, S. (1993) Language games. In The Encyclopedia of Language and Linguistics. Oxford and New York:
Pergamon Press. 1980-1985.
Laycock, D. (1972). Towards a typology of ludlings or play languages. Linguistic Communications 6: 61-113
84
McCarthy, J. and A. Prince. (1994). Generalized Alignment. In G. Booij & J. van Marle (eds.). Yearbook of
Morphology 1993, Dordrecht: Kluwer. 79–153.
— and — (1995). Faithfulness and Reduplicative Identity. UMOP 18: 249-384.
Nelson, N. (2003). Asymmetric anchoring. Ph.D. dissertation. Rutgers State University of New Jersey.
Piñeros, C. (1998). Prosodic morphology in Spanish: Constraint interaction in word-formation, Ph.D. diss.,
Ohio State University, Columbus.
Riggle, J. (2003). Nonlocal reduplication, in K. Moulton & M. Wolf (eds.). Proceedings of NELS 34, Amherst,
MA: GLSA. 485–496.
— (2006). Infixing reduplication in Pima and its theoretical consequences. Natural Language & Linguistic
Theory 24: 857–891
Vaux, B. (2011). Language Games. In J. Goldsmith, J. Riggle & A. C. L. Yu (eds). The Handbook of
Phonological Theory, 2nd Edition, Oxford, UK: Wiley-Blackwell. 722-750.
Vogt, B. (2012). Prosodische Morphologie in deutschen Geheimsprachen, Roma, Aracne.
Yu, A. C.L. (2008). On iterative infixation. In Ch. B. Chang & H. J. Haynie (eds.). Proceedings of tWCCFL 26,
Ithaca NY: Cascadilla Press, 516-524.
— (2007). A Natural History of Infixation. Oxford: OUP.
85
Floating Universal Quantifier as a Base-Generated
Functional Head
Dong-yi Lin
Ghent University
[email protected]
Like English all and French tous, the universal quantifier niz ‘all’ in Kavalan, an
Austronesian language in eastern Taiwan, can immediately precede a DP (1a) or appear in a
quantifier-floating construction where it is not adjacent to its DP associate (1b).
(1) a. m-lizaq tu wasu [ya m-niz sunis].
AV-like OBL dog ABS AV-all child
‘All the children like dogs.’
b. m-niz m-lizaq tu wasu [ya sunis].
AV-all
AV-like OBL dog ABS child
‘The children all like dogs.’
According to Sportiche (1988), Shlonsky (1991), and Bošković (2004), quantifier
floating results from the stranding of a universal quantifier in an intermediate position where
its DP associate passes. Baltin (1995) and Doetjes (1997) however argue that a floating
quantifier is base-generated in the left periphery of VP. The present paper argues for the basegeneration approach to floating niz ‘all’, which exhibits semantic and morphosyntactic
differences from non-floating niz.
The first difference between floating and non-floating niz concerns scope interaction
with negation. (2a) shows that either non-floating niz or the negation marker mai can take
wider scope over the other. In contrast, floating niz exhibits scope-freezing effects, as
illustrated by the unambiguity of (2b) or (2c). The ambiguity of (2a) can be attributed to
reconstruction. The absolutive DP in (2a) occupies Spec, TopP (Lin 2013), which is higher
than mai, whereas its base position is VP-internal and lower than mai. Reconstruction cannot
take place in (2b) or (2c), as floating niz is base-generated above the lexical verb, either
higher than mai (2b) or lower than mai (2c), and thus there is no lower copy that licenses
reconstruction.
86
(2) a. mai qibasi-an-na ni imuy [ya m-niz qudus].
NEG wash-PV-3ERG ERG Imuy ABS AV-all clothes
‘Imuy didn’t wash all the clothes.’ (NEG > all; all > NEG)
b. m-niz mai qibasi-an-na
ni imuy [ya qudus].
AV-all NEG wash-PV-3ERG ERG Imuy ABS clothes
‘Imuy didn’t wash all the clothes.’ (*NEG > all ; all > NEG)
c. mai m-niz qibasi-an-na ni imuy [ya
qudus].
NEG AV-all wash-PV-3ERG ERG Imuy ABS
clothes
‘Imuy didn’t wash all the clothes.’ (NEG > all ; *all > NEG)
Secondly, while non-floating niz is a nominal modifier, floating niz should be analyzed
as a full-fledged verb. The contrast between (3a) and (3b) shows that floating niz, but not its
non-floating counterpart, can take the imperative suffix. Floating niz, but not non-floating niz,
can also take the causative prefix (4). As illustrated in (5), floating niz can be affixed with the
patient voice (PV) marker -an, whereas non-floating niz cannot. Moreover, the voice markers
on floating niz are verb-defining v, which can determine the argument structure of a sentence.
While PV -an by itself can assign an external argument and an affected theme, AV m- cannot
(Lin 2013, to appear). PV-marked niz in (6a) is thus grammatical even without a lexical verb,
but this is not true of AV-marked niz in (6b).
(3) a. niz-ika
m-liyam [ya sudad].
AV-IMP.PV AV-read ABS book
‘Read all the books!’
b. *m-liyam [ya niz-ika
sudad].
AV-read
ABS all-IMP.PV book
(4) pa-niz=iku
[tu sunis] pa-taqsi.
CAUS-all=1SG.ABS OBL child CAUS-study
‘I make all the children study.’
(5) a. niz-an-na=ti
ni
abas q<m>an [ya byabas].
all-PV-3ERG=PFV ERG Abas <AV>eat ABS guava
‘Abas ate all the guavas.’
b. *qan-an-na=ti
ni abas [ya niz-an byabas].
eat-PV-3ERG=PFV ERG Abas ABS all-PV guava
(6) a. niz-an-na=ti
ni abas ya byabas.
all-PV-3ERG=PFV ERG Abas ABS guava
‘Abas ate/used up all the guavas.’
87
b. *m-niz ti-abas
tu
byabas.
AV-all NCM-Abas OBL guava
These morphosyntactic facts suggest that floating niz should be analyzed as the main
verb of a sentence. This empirical generalization is incompatible with the stranding analysis,
which predicts that the stranded quantifier should be embedded inside a specifier position and
cannot undergo head movement to v. Instead, as low adverbials in the VP periphery are all
realized as a verb in Kavalan (7) (Chang 2006), the facts shown in (3) – (6) corroborate the
analysis of floating niz as a base-generated head in the VP periphery below v.
(7) a. paqanas-an-ku
m-liyam ya sudad.
slowly-PV-1SG.ERG AV-read ABS book
‘I read the book slowly.’
b. m-duna
qan-an-ku
ya ’may.
AV-always eat-PV-1SG.ERG ABS rice
‘I always eat (this kind of) rice.’
Thirdly, if floating niz and non-floating niz are not derivationally related, as claimed
by the base-generation approach, it is predicted that they should be able to co-occur in a
sentence. As shown in (8), this prediction is borne out.
(8) niz-an-na ni abas q<m>an [ya m-niz byabas].
all-PV-3ERG ERG Abas <AV>eat ABS AV-all guava
‘Abas ate up all the guavas.’
Finally, floating niz can receive interpretations that are not available to non-floating
niz. Floating niz can quantify over either entities/individuals or predicates (9). Moreover, like
Chinese dou in (10a), which can quantify over a set of propositions denoted by an embedded
question (i.e., their possible answers) (Cheng 1995), Kavalan floating niz can also be
interpreted as a quantifier over a set of propositions, as illustrated in (10b).
(9) m-niz tengen ya qudus-ku.
AV-all black ABS clothes-1SG.GEN
‘My clothes are completely black.’ or ‘All my clothes are black.’
(10) a. ni guyong shei, wo dou hui bang ta.
you hire
who I all will help him
‘No matter who you hire, I will help him.’
b. t<m>ayta tu ti-tiana wasu zau nani, niz-an-na Raytunguz
<AV>see OBL NCM-who dog this DM all-PV-3ERG bark
‘No matter who this dog sees, it barks (at him).’
88
(Chinese)
In conclusion, floating niz is not derived from non-floating niz as a result of stranding.
The two differ both semantically and syntactically. Non-floating niz is a nominal modifier,
whereas floating niz is base-generated in a head position in the VP periphery below v and
exhibits properties of a full-fledged verb just like other adverbial verbs in Kavalan.
References
Baltin, Mark R. (1995). Floating quantifiers, PRO, and predication. Linguistic Inquiry 26: 199–248.
Bošković, Željko. (2004). Be careful where you float your quantifiers. NLLT 22: 681–742.
Chang, Yung-li. (2006). The guest playing host: Adverbial modifiers as matrix verbs in Kavalan. In Clause
Structure and Adjuncts in Austronesian Languages, eds. by Gärtner Hans-Martin, Paul Law, and
Joachim Sabel. Berlin: Mouton de Gruyter. 43–82.
Cheng, L. L.-S. (1995). On dou-quantification. Journal of East Asian Linguistics 4: 197–234.
Doetjes, Jenny Sandra. (1997). Quantifiers and selection: On the distribution of quantifying expressions in
French, Dutch and English. PhD Dissertation. Leiden University, The Hague.
Lin, Dong-yi. To appear. The syntactic derivations of interrogative verbs in Amis and Kavalan. In New advances
in Formosan linguistics, eds. by Elizabeth Zeitoun, Stacy F. Teng, and Joy J. Wu. Canberra: AsiaPacific Linguistics.
Lin, Dong-yi. (2013). Interrogative constructions in Kavalan and Amis. PhD dissertation. University of Florida,
Gainesville.
Shlonsky, Ur. (1991). Quantifiers as functional heads: A study of Quantifier Float in Hebrew. Lingua 84: 159–
180.
Sportiche, Dominique. (1988). A theory of floating quantifiers and its corollaries for constituent structure.
Linguistic Inquiry 19: 425–449.
89
In the South Slavonic Garden: Landscaping the
Landscape of Arguments andNon-Arguments
Marijana Marelj
UiL OTS, Utrecht University
[email protected]
Mysterious Mysteriousness
Cognate object construction (COC) raises intriguing puzzles for different modules of
grammar. Firstly, the construction raises the issue of the scarcity of syntactic space. Though
canonical COC examples like (1) from Serbo-Croatian (SC) and their English counterparts (2)
are based on unergatives (see Kayser&Roeper 1984; Hale& Kayser 2002), one finds COCs of
canonical unaccusatives (3) as well:
(1) a. Sanja (lep) san.
b. Živi udoban život.
c. Pleše ples.
(2) a. He dreams a (nice) dream.
b. He lives a life.
c. He dances a dance.
(3) a. He died a natural death.
b. Umro
je
prirodnom smrću.
death-INSTR
die-PRT.M.SG AUX-3.SG natural
(SC)
The fact that the unaccusatives have internal arguments leaves no syntactic space for
the cognate object (CO) natural death in (3a):
(4) He [VP [V’ died he]
unaccusatives
Though the syntax of unergatives might look unproblematic, if they are treated as
denominal verbs, where bare nouns (DREAM, LIFE, DANCE etc.) are generated in the
complement position of the semantically empty DO & into which they subsequently
incorporates (see Hale and Kayser (1993 et seq) (5), the puzzling scarcity of syntactic space
arises again!
90
(5)
V′/VP
V
do
N
dance
INCORPORATION/CONFLATION
Though COs in (1) – (3) would seem to indicate that the CO is always etymological
related to the verb (hence, cognatus), in actual fact, it is only in a subset of the cases that some
these nominal phrases are obligatorily morphologically related to the relevant verbs:
(6)
My clairvoyant dreamt the most unusual thing
(Massam 1990: 3)
The question that necessarily arises here is what the presence/lack of morphological
cognatehood of the relevant nominals tells us. Further, though they did not receive any
analysis in the literature so far, note, interestingly, that Sometimes COs positions are not
occupied by nominals at all (7).
(7)
Luka dreamt [CP that he was an eagle]
Its other puzzling properties seem to indicate that COC is an interface phenomenon.
Note, for instance, that death in (3) – quite unlike life, dance, and dream in (1) & (2) – cannot
occur unmodified. Importantly, this requirement for modification cannot be relegated to the
lexicon (as an idiosyncrasy). It is neither restricted to DIE, nor to these two languages. Note,
further, that there is a well-know mystery of the status of the cognate object, across different
languages as well as within the same language. there is a well-know mystery about the status
of the nominal phrase. There are researchers that claim that the cognate object are arguments
- Massam, 1990; Hale i Kayser, 1993 for English; Macfarland, 1995 for French and English;
Matsumoto, 1996 for English and Japanese; Pham, 1990 for Vietnamese a.o. Another group
of researchers claims these are non-arguments, adjuncts (Zubizarreta, 1987; Jones, 1988 for
English;; Moltmann, 1989 for English and German; Lefebvre, 1994 Fongbe; Mittwoch, 1998
for English and Hebrew; Stewart, 1998 for Èdó. Finally, there is a third group of researchers
that adopt a “hybrid” approach to cognate objects, either through the prism of different
languages as in Mittwoch, 1998, who analyses English cognate objects as arguments and
those in Hebrew as adjuncts) or within a single language Pereltsvaig, 1999 for Russian) .
91
Finally, there is even a disagreement among researchers with respect to the status of
the data; (8) is ungrammatical in Moltmann (1989) but grammatical in Massam (1990).
(8)
Tom sneezed every sneeze that we heard that day.
Proposal
This presentation deals with the morphological, syntactic, semantic, and pragmatic
aspects of COC with a particularly reference to SC and Slovene. I will demonstrate that an
examination of such morphologically robust languages facilitates understanding of some of
the puzzling properties of COC cross-linguistically and also offers a way of explaining the
noted disagreements in judgments found in the literature on Germanic languages like English
(8)
Based on a barrage of tests, I argue that there are two types of COs: ACC-COs and
INSTR-COs in SC. Slovene data pattern with the SC data in all relevant respects. The upshot
of my analysis is that whereas the ACC-COs in SC are arguments, on a par in their function
and status with complements of “regular” transitive verbs, INSTR-COs are modifiers.
Adopting the treatment of modifiers within the Davidsonian tradition (see Davidson 1967,
Carlson 1984; Parsons 1990), I analyze them as first order predicates. My approach to
INSTR-COs allows me to capture several of their core properties, among which is the
“notorious” property of obligatory modification in (3). Specifically, I argue that this property
is rooted in the semantics of INSTR-COs, but “regulated” by pragmatic concerns (9) (hence,
the appropriate “evaluation” of (3) is as #, NOT “*”, as typically found in the literature). The
analyses further predicts that the modification is optional in cases like (1) and (2). This
prediction is borne out. Whereas the modifier/INSTR-cognate object is a predicate that
directly modifies the event variable, an argument in the neo-Davidsonian tradition modifies
this variable indirectly, via the thematic role. And the fact that this relation is mediated
through a thematic role makes it informative enough, regardless of whether this element is
cognate or not. Hence, no modification of the argument is required.
Contra standard views in the literature on Slavonic, I will argue that that though this
NPs are non-verbal predicates, the fact that they are marked with the instrumental case is
nothing more than a quirk of morphology, with no deeper meaning or relevance. Namely,
though it is typically argued that non-verbal predicates bear instrumental as predicate case
(see Matushansky 2010 for discussion and references on Russian) there is no correlation to be
established between any “special” case and non-verbal predicates in South Slavonic. Namely,
92
SC instances of both primary and secondary non-verbal predicates in nominative case are far
more common that those marked with the morphological instrumental case. Moreover, data
like (9) from Slovene clearly corroborate that there is no correlation as the COCs analyzed
here as non-verbal predicates always appear in genitive case in Slovene. Needless to say, in
both of these languages clear-cut cases of arguments in instrumental case are found in
abundance.
(9) Umrl
je
naravne smrti.
die-PRT.M.SG AUX-3.SG natural death-GEN
He died a natural death.
(Slo)
(10) a. #Tristram je umro smrću.
a'. ∃e [ die (e, T) & death(e)]
b. Tristram je umro užasnom smrću.
b'. ∃e [ die (e, T) & horrible death (e)]
My analysis further explains why verbs that take predicative COs never allow
sentential complements, while the ACC-taking ones can (6). The semantic parallelism
between the canonical adverbial modifiers and INSTR-COs extends to the syntax as well.
Treating INSTR-COs as syntactic adjuncts resolves the problem of syntactic space (4a)
directly. What about the syntax of unergatives? I will show that no incorporation/conflation
type of analysis doesn’t account for the data, but unlike Pereltsvaig (1999) & Nakayama
(2006) I will also argue that unergatives are a primitive category.
References
Blom, C., P.de Groote, Y.Winter, & J., Zwarts. (2011). Implicit Arguments: Event Modification or Option Type
Categories. In M. Aloni et al. (Eds.): Amsterdam Colloquium 2011, LNCS 7218, pp. 240–250, 2012.
Hale, K. & S.-J. Kayser. (2002). Prolegomenon to a theory of argument structure. Cambridge, MA.: MIT Press.
Marelj, M. and O. Matushansky. (2015). Mistaking for: Testing the theory of Mediated Predication. Linguistic
Inqury 46/1.
Parsons, T. (1990). Events in the Semantics of English: A study in Sub- Atomic Semantics. Current Studies in
Linguistics Series 21. Cambridge, Mass.: MIT Press.
Pereltsvaig, Asya. (1999). Cognate Objects in Russian: Is the notion “Cognate”relevant for Syntax? Canadian
Journal of Linguistics 44(3): 267–291.
93
Wh-in-situ in a multiple wh-fronting language
Petra Mišmaš
University of Nova Gorica
[email protected]
Multiple wh-fronting has been described as obligatory in the past (e.g. Bošković 2002),
but some authors have noted that not all wh-phrases need to move to the clause initial
position. For example, Citko (2010) shows that Polish exhibits short wh-movement (whphrase is located after the subject), Šimík (2010) explores wh-in-situ in Czech, and Mišmaš
(2014) shows that multiple wh-fronting can be optional in Slovenian. This talk explores
instances of wh-in situ in a multiple wh-fronting language. The main focus of the talk is on
Slovenian, but the phenomena can also be observed in Serbo-Croatian. The goal of the talk is
to provide an account of wh-in-situ in multiple wh-questions in a multiple wh-fronting
language.
Slovenian is a multiple wh-fronting language (Golden 1997). However, if we take a
multiple wh-question with two wh-phrases in Slovenian, we can observe three types of
questions. Both wh-phrases can move to a clause initial position, (1), or only one wh-phrase
moves to a clause initial position and the second wh-phrase undergoes short movement, (2), or
stays in situ, (3). Example (4) shows that wh-phrases can also stay in situ in Serbo-Croatian.
(1)
a. Kaj
je komu Miha
kupil?
what-ACC AUX who-DAT Miha-NOM buy
…..‘What did Miha buy for whom?’
b. Komu je kaj Miha kupil?
(2)
a. Kaj
je Miha
komu
kupil?
What-ACC AUX Miha-NOM who-DAT buy
‘What did Miha buy for whom?’
(3)
a. Kaj
je Miha
kupil komu?
What-ACC AUX Miha-NOM buy who-DAT
‘What did Miha buy for whom?’
b. Komu je Miha kupil kaj?
94
Slovenian
(4)
a. Ko
je kupio šta?
who-NOM AUX buy what-ACC
‘Who bought what?’
Serbo-Croatian
b. Ko je šta/Šta je ko kupio?
Crucially, in Slovenian a wh-phrase can stay in situ in a multiple wh-question only if at
least one wh-phrase fronts to the clause initial position (but as (5) shows for a question with
three wh-phrases, more than one wh-phrase can stay in situ, and more than one wh-phrase can
move if a wh-phrase stays in situ).
(5)
a. Kdo
je izpostavil koga
čemu?
who-NOM AUX expose who.-ACC what-DAT
‘Who exposed whom to what?’
Slovenian
b. Kdo je koga izpostavil čemu?
c. Kdo je čemu izpostavil koga?
d. Kdo je koga čemu/čemu koga izpostavil?
e. Čemu je kdo koga/koga kdo izpostavil?
f. Koga je kdo čemu/čemu kdo izpostavil?
e. Je kdo izpostavil koga čemu?
‘Did somebody expose someone to something?’/# ‘Who exposed whom to what?’
Note also that (1), (2), and (3) all receive the same interpretation. However, if all whphrases stay in situ, the wh-question is unacceptable under a true question reading, but can be
interpreted either as an echo question (if the wh-phrases are emphasized) or a yes/no-question
if the question has a yes/no-question marker or yes/no-question intonation, (5e). Crucially, in
yes/no-questions, (5e), the wh-pronoun is interpreted as an indefinite pronoun. I take whpronouns in (5e) to be bare wh-pronouns (as they have no prefix in Slovenian). In addition to
these, Slovenian also has indefinite pronouns, specified with the prefix ne- (e.g. nekdo
‘somebody’) which can appear in any context, and interrogative wh-pronouns, which I take to
be specified with a wh-feature and which receive an interrogative reading and undergo whmovement (as in (1)). Bare wh-pronouns, however, can appear in specific contexts (i.e.
polarity contexts in Cheng (1991), such as yes/no-questions, (5e)) and need to be licensed.
When they appear in polarity context, they receive an indefinite reading, as already proposed
in Cheng (1991), who observes a similar division of pronouns in other multiple wh-fronting
languages. However, I propose that bare wh-pronouns can also appear in situ in wh-questions,
(3), in which they get the interrogative reading. That is, wh-phrases in situ in Slovenian are
bare wh-pronouns which are underspecified for wh-feature and need to be licensed.
95
Slovenian wh-in-situ is therefore similar to wh-pronouns in wh-in-situ languages in
that in both a wh-pronoun needs to be licensed by some other element in the sentence. I
follow Cheng’s (1991) analysis of polarity items in multiple wh-fronting languages and
assume that when bare wh-pronouns have an indefinite (i.e. existential) reading this is
achieved by Existential Closure which introduces an (null) existential quantifier that binds the
pronoun and gives it existential force. The bare wh-pronoun is interpreted as interrogative in a
similar way, however, the proposal here hinges on the analysis of wh-movement.
Despite the lack of Superiority in Slovenian (cf. (1), (5d–f)), I do not adopt a focus
analysis of wh-movement (e.g. Bošković 2002), but rather assume that wh-phrases are whmoved to the left periphery. I follow Soare (2007) and propose that Slovenian forms a
complex unvalued interpretable Q+wh-feature (responsible for Clause Typing) that appears on
the Interrogative head in the left periphery of wh-questions (and not Focus0, cf. Soare (2007))
and comes with an EPP-subfeature. Also, the left periphery hosts Wh-Projections which come
with an unvalued interpretable wh-feature and an EPP-subfeature, while interrogative whphrases carry valued uninterpretable wh-features. This means that one wh-phrase has to move
to the Interrogative Projection (which precedes all other projections in the left periphery of
wh-questions), while the remaining interrogative wh-phrases move to Wh-projections.
Two properties of wh-in-situ hold in Slovenian. (i) The wh-in-situ is a bare whpronoun that has no wh-feature and hence does not agree with the Q+wh-feature on Inter0 and
does not move (nor can it agree with the interpretable wh-feature on Wh0). (ii) Because it is a
bare wh-pronoun it needs to be licensed. In multiple wh-questions the licensing operator is the
interpretable Q+wh-feature on the Inter0. If the underspecified wh-phrase is in the scope of the
Q+wh-feature, the wh-phrase in situ is always interpreted as an interrogative wh-phrase (just
as moved wh-phrases in (1), hence the same interpretation of (1) and (3)). This means that a
wh-phrase which has a wh-feature agrees with the complex Q+wh-feature on Inter0 (or the whfeature on Wh0) and moves because of the EPP-subfeature on the Q+wh-feature (or on the whfeature on Wh0). However, the interpretable complex feature on Inter0 is also responsible for
licensing the interrogative reading of the wh-in-situ, which explains why bare wh-pronouns
only receive an interrogative reading in questions with clause initial wh-phrases.
And finally, this proposal implies that wh-movement in Slovenian is not optional. That
is, if the wh-phrases are specified with a wh-feature, they need to move. However, when they
are underspecified bare wh-pronouns (and have no motivation to move), they stay in situ.
96
References
Bošković, Željko. (2002). On multiple wh-fronting. Linguistic Inquiry 33(3): 351−383.
Cheng, Lisa Lai‐Shen. (1991). On the typology of wh-questions. Cambridge: MIT. (Ph.D. dissertation)
Citko, Barbara. (2010). On the (a)symmetric nature of movement: A view from Polish wh- and passive
movement. In Browne, Wayles et al. (eds.), Formal Approaches to Slavic Linguistics 18: The Cornell
Meeting 2009. Ann Arbor: Michigan Slavic Publications. 38–57.
Golden, Marija. (1997). Multiple wh-questions in Slovene. In Browne, Wayles et al. (eds.), Formal Approaches
to Slavic Linguistics: The Cornell Meeting, 1995. Ann Arbor: Michigan Slavic Publications.
240−266.
Mišmaš, Petra. (2014). Multiple wh-fronting can be optional too. In Chapman, Cassandra et al. (eds.), Formal
Approaches to Slavic Linguistics 22: The McMaster Meeting. Ann Arbor: Michigan Slavic
Publications. 225–244.
Soare, Gabriela. (2007). A cross-linguistic typology of question formation and the antisymmetry hypothesis.
Generative grammar in Geneva 5: 107–133.
Šimík, Radek. (2010). Interpretation of multiple interrogatives: An information structure sensitive account. In
Browne, Wayles et al. (eds.), Formal Approaches to Slavic Linguistics 18: The Cornell Meeting
2009. Ann Arbor: Michigan Slavic Publications. 486–501.
97
Complementizer doubling and clausal topics in ItaloRomance
Nicola Munaro
Università Ca’ Foscari Venezia
[email protected]
Building on some previous studies on complementizer doubling in (early) Romance,
with this paper I intend to contribute to a deeper understanding of the functional articulation
of the Topic field within the left periphery of embedded clauses, examining at the same time
the implications of the observed developments for a formal theory of downward reanalysis.
The phenomenon of complementizer doubling has received considerable attention in
recent years: it consists in the double phonetic realization of the subordinating
complementizer, which appears preceding and following a preposed constituent of the
embedded clause; the second occurrence of the complementizer has been repeatedly proposed
to lexicalize the head Topic° of a TopicP projection hosting the topicalized constituent in its
specifier.
For example, Paoli (2005), discussing some cases of complementizer doubling in early
Romance - like the one exemplified in (1) from 13th century Tuscan - takes the second
occurrence of che to head the TopicP projection (which hosts the topicalized if-clause in its
specifier); the overt realization of the complementizer in Topic° is taken to reflect a spec-head
agreement relation between Topic° and the clausal constituent occupying Spec,TopicP:
(1) …si pensò ir re Pelleus che, se elli potesse tanto fare che Giason suo nepote volesse
andare in quella isola per lo tosone conquistare, che mai non tornerebbe, e in tal manera
si diliverebbe di lui.
‘…king Pelleus thought that, if he could do so that his nephew Giason wanted to go to
that island to take that …, that he would never come back, and so he could get rid of
him’.
In the same vein, Ledgeway (2005) - in examples like (2) from an early Southern
Italian variety - interprets the first occurrence of che as the lexicalization of Force° and the
second one as the phonetically realized trace left in Topic° by the complementizer raising up
to Force°:
98
(2)…le aveva ditto che se sua maistà voleva lo stato suo che se llo venesse a ppigliare co la
spata in mano.
‘he had told him that, if his majesty wanted his state, that he should come and take it with
his sword in hand’.
Hence, in early Italo-Romance varieties if-clauses – and clausal adjuncts in general –
were among the most plausible candidates to fill the position sandwiched between the two
occurrences of the complementizer (cf. also Vincent (2006)).
Nowadays, within the Italo-Romance domain, complementizer doubling is exclusively
attested in some North-Western Italian varieties like Ligurian and Turinese, as exemplified in
(3) and (4) from Poletto (2000) and Paoli (2005) respectively:
(3) Sperem che Gianni ch’u lese questu libru
‘We hope that John reads this book’.
(4) Majo a pensa che Franchin ch’as n’ ancorza.
‘Mario thinks that Frank will realize it’.
The possibility for a clausal adjunct to appear between the two instances of che has
been lost with time in Italo-Romance and, as is clear from (3) and (4), only phrasal
constituents – most frequently the subject – can now intervene between the higher and the
lower complementizer.
On the other hand, the hypothesis that the second complementizer lexicalizes a Topic°
head has been recently reproposed for the cases of complementizer doubling attested in
modern Ibero-Romance (cf. Mascarenhas (2007) on European Portuguese, Villa-Garcìa
(2012) on Spanish and Gonzàlez i Planas (2013) on Catalan), where the possibility for an ifclause to intervene between two instances of que is still attested.
If we accept the correctness of this approach, then the fact that preposed protases can
easily enter the complementizer doubling construction can be seen as an empirical argument
in favour of the hypothesis that they belong to the Topic field of the associated embedded
clause. Moreover, the data presented above suggest that the Topic field should be split – along
the lines of Benincà & Poletto (2004) – into at least two distinct Topic subfields, the higher of
which dedicated to host topicalized clausal adjuncts, the lower one hosting topicalized phrasal
constituents. Adopting this view, the impossibility in modern Italo-Romance to sandwich a
clausal adjunct between two complementizers can be interpreted as a consequence of a
diachronic process of downward reanalysis to the effect that the second occurrence of che in
(1) and (2), originally lexicalizing a high head of the Topic field, has been reanalyzed by the
speakers – due to the structural ambiguity of the linear string – as the lexicalization of a lower
99
Topic° head, namely the one associated with the specifier position hosting preposed phrasal
constituents.
Interestingly, the reanalysis can apply recursively downwards and the complementizer
heading the lower Topic° head can be further reanalyzed as a lexicalization of Fin°, the lowest
functional head of the left periphery (according to Rizzi’s (1997) seminal proposal); this
hypothesis receives empirical support from the fact that in examples like (3) and (4) the
presence of the second complementizer correlates with the use of the subjunctive in the
embedded clause, as pointed out by Paoli (2005).
If the diachronic path sketched above is on the right track, it challenges some aspects
of the notion of downward reanalysis as defined in Roberts & Roussou (2003), in so far as the
reanalysis process investigated here involves no category change and no interface effects, but
at the same time seems to be cyclic and to affect only some members of the relevant lexical
class.
References
Benincà, P. & C. Poletto. (2004). Topic, Focus and V2 – Defining the CP sublayers. In L. Rizzi (ed.) The
structure of IP and CP. Oxford/New York: Oxford University Press. 52–75.
Gonzàlez i Planas, F. (2013). On quotative recomplementation: Between pragmatics and morphosyntax.
Manuscript. University of Girona.
Ledgeway, A. (2005). Moving through the left periphery: The dual complementizer system in the dialects of
Southern Italy. Transactions of the Philological Society 103(3): 339–396.
Mascarenhas, S. (2007). Complementizer doubling in European Portuguese. Manuscript, Amsterdam ILLC/
NYU.
Paoli, S. (2005). COMP: a multi-talented category: Evidence from Romance. In L. Brugè et alii (eds.)
Contributions to the 30th Incontro di Grammatica Generativa. Venice: Cafoscarina. 185–202.
Poletto, C. (2000). The higher functional field – Evidence from Northern Italian dialects. Oxford/ New York:
Oxford University Press.
Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (ed.) Elements of grammar. Dordrecht:
Kluwer. 281–337.
Roberts, I. & A. Roussou. (2003). Syntactic change: a minimalist approach to grammaticalization. Cambridge:
Cambridge University Press.
Villa-Garcìa, J. (2012). Recomplementation and locality of movement in Spanish. Probus 24(2): 257–314.
Vincent, N. (2006). Il problema del doppio complementatore nei primi volgari d’Italia. Talk delivered at the
University of Padua. April 2006.
100
Locative expressions in Slovenian Sign Language
(SZJ)
Matic Pavlič
Università Ca' Foscari, Venice
[email protected]
Introduction
In sign language literature, it is commonly assumed that sign languages lack spatial
adpositions and encode locative information in spatially agreeing predicates instead. Spatially
agreeing predicates are defined in relation to other types of verbs (agreeing and non-agreeing
verbs) 1 and predicates (classifier and non-classifier predicates). 2 In almost all SVO sign
languages for which locative constructions have been studied in detail, it has been discovered
that they feature non-basic word order compared to the word order in transitive sentences.
Usually, a Ground is signed first, then the located participant (Figure) is introduced, and lastly
the locative relation between the two is established via classifier predicate – as reported for
American Sign Language (Liddell 1980: ASL), Russian Sign Language (Kimmelman 2012:
RSL), Croatian Sign Language (Milković et al. 2007: HZJ), The sign language of
Neatherlands (Coerts 1994: NGT), Australian, Flemish and Irish Sign Language (Johnston et
al. 2007: AUSLAN, VGT and IrSL). Locative constructions in different unrelated sign
languages look very much alike, because signers productively use mental representations of
space as their underlying visual metaphors. Strikingly, a similar pattern is reflected even in
the pantomime of non-signers, elicited and analyzed by (Laudanna and Volterra 1991). Their
results indicate that the word order used in locative constructions does not follow (only)
linguistic rules – but might also be bound to the principles of the human thought system,
perception and visual modality. Nevertheless, the initial conceptual trigger for specific word
order in locative constructions might indeed be due to the visual modality, but the technical
(re)ordering of the elements is still expected to be carried out by linguistic processes. Indeed,
the tendency to introduce locative referents before the information about them is predicated
1 Non-agreeing verbs are phonologically specified for a place of articulation near or on the body and are
incapable of adapting their form to the location of the participants by manipulating their movement
subcomponent. Agreeing verbs, on the other hand, adapt the starting and ending point of their movement
subcomponent to the places where two of their arguments are articulated (in order to agree with them).
2 Classifiers in sign languages are meaningful hand configurations that denote a salient characteristic of their
referent and form classifier predicates.
101
can be found in both spoken and signed languages (Perniss 2007). On the other hand, there
are also some remarkable differences between the locative constructions in sign languages
versus locative constructions in spoken languages: the presence of overt spatial adpositions
and copula, for example. In this paper, I discuss the word order of locative construction and
the structure of the locative predicate in example (1). The sentence was produced by a native
Slovenian Sign Language (SZJ) signer. To elicit it, I used a picture description task (PDT;
Volterra et al. 1984), more precisely a stimulus picture in (2).
The word order of locative constructions in SZJ
In (1), a Ground is established first (TABLE). It is signed in r(eferential)-locus 'a' with
two b-hands moving away from each other facing downwards. Its non-dominant handshape is
held in place, while a dominant hand simultaneously produces a Figure in r-locus 'b'. The
Figure is a complex DP with numeral THREE quantifying an NP BOOK. Finally, the predicate is
signed three times progressively further along r-locus 'a'. In sign language literature, such
reduplication of a predicate is formally captured in terms of a distributivity morpheme that
according to Benedicto and Brentari (2004) attaches only to the predicates that license an
internal argument. The same seems to be true for SZJ (Pavlič, in preparation). The possibility
of a SZJ locative predicate to reduplicate confirms the Figure as an internal argument of SZJ
locative predicates.
Now, examine the movement of the predicate. The predicate sign is modulated such
that its movement starts in the location 'a' at which the Figure has been articulated and ends in
the location 'b' at which the Ground has been articulated. Therefore, the internal argument
represented by Figure functions as a subject while the Ground seems to be an oblique
argument. The Ground is non-manually marked with raised eye-brows and ends with a
prosodic marker, an eye-blink. Raised eye-brows often mark topicalised constituents in sign
languages as shown for various unrelated sign languages, among others for ASL (Aarons
1994), Finnish SL (Jantunen 2007), Hong Kong SL Sze (2008), Israeli SL (Rosenstein 2001),
and NGT (Coerts 1992) and SZJ (Pavlič, in preparation).
In the literature on the word order, Figures are usually linked to Subjects while
Grounds are linked to Objects. This way, a Figure-Ground word order can be compared to the
SO pattern and a Ground-Figure word order can be compared to the OS pattern. If we apply it
to the SZJ data presented above, we can conclude, that the canonical word order in SZJ
locative constructions appears to be OSV (due to topicalisation) and not the basic SVO
102
(Pavlič 2015). Note, however, that the Figure and the locative predicate in (1) also do not take
the expected positions in the sentence as I show in the next section.
(SZJ; m44a) 3
'There are three books on a/the table.'
The structure of locative predicates in SZJ
The predicate in (1) qualifies as a classifier predicate because it is a complex sign in which the
movement subcomponent represents the root, while the dominant handshape is a bound
morpheme that refers back to the Figure. Crucially, SZJ classifier predicate triggers a nonbasic SOV word order also in non-locative constructions (Pavlič, in preparation). This effect
is also seen in locative example (3a), to which, in addition to the classifier predicate, a copular
verb
HAVE
is inserted. I assume, that its covert version is also present in (1). The covert (3b)
and overt (3c) version of the copular verb may be negated while the classifier predicate
cannot be (3d) – (3e). What is the category of a locative classifier predicate, then?
(3)
a. TABLEA HAVE THREE BOOKB ABE-LOCATED:BOOK(B-DOWN)+++B
'There are three books on a/the table.'
(SZJ; m44b)
b. TABLEA NOT THREE BOOKB ABE-LOCATED:BOOK(B-DOWN)+++B
'There are not three books on a/the table.'
(SZJ; m44c)
c. TABLEA NOT-HAVE THREE BOOKB ABE-LOCATED:BOOK(B-DOWN)+++B
'There are not three books on a/the table.'
(SZJ; m44d)
d. *TABLEA THREE BOOKB NOT ABE-LOCATED:BOOK(B-DOWN)+++B
'There are not three books on a/the table.'
(SZJ; m44e)
e. *TABLEA NOT-HAVE THREE BOOKB NOT ABE-LOCATED:BOOK(B-DOWN)+++B
'There are not three books on a/the table.'
(SZJ; m44f)
Since SZJ locative classifier predicate cannot be negated, the whole structure cannot be
analyzed as bi-clausal but as mono-clausal. I assume that SZJ locative constructions with a
classifier predicate are structures projected by the overt or covert copular verb
HAVE
that
selects a small clause as its complement. I understand small clause as a predicative projection
called PredP that is projected by a head Pred0. Pred0 is represented by a movement
subcomponent that starts in Figure's r-locus and ends in Ground's r-locus (see above). It
licenses the Figure (internal argument) in its specifier position and in complemented by a
3 I adopt a common notational convention in the sign language literature: signs are glossed by literal English
translation in SMALL CAPS with non-manuals marked with a line extending over the signs they co-occur with.
103
prepositional phrase PP. P0 is as silent head that selects Ground as it complement and an
axial-part phrase as its specifier. What is the axial part phrase? When articulating a predicate
sign in (1), the non-dominant hand resumes Ground's hand configuration. Note, that during
the Figure's articulation the non-dominant hand was oriented downwards while during the
predicate articulation it is oriented upwards and does not refer to the Ground as a whole
anymore – but to the part of the Ground (namely to its axial part; see Svenonius 2006), that is
relevant for defining Ground's spatial configuration with respect to the Figure. Pfau and Aboh
(2012) compare the Ground's axial part in NGT to the function of oral language prepositional
domain in expressions such as 'A cat is on the surface of the roof'. With reference to SZJ
examples presented above, I further elaborate their proposal.
Conclusion
In this paper, I examine syntactic functions of the Figure and Ground constituents in
SZJ locative construction. This enables me to determine their base generated positions and
discuss the processes that lead to the surface word orders in examples of SZJ locative
constructions described above. I also look at the non-dominant hand that is frequently
preserved in space during the production of SZJ locative constructions. I explore the relation
among the Ground DP and the axial part NP (see also Svenonius 2006).
References
Aarons, D. (1994). Aspects of the Syntax of American Sign Language. Ph. D. thesis.
Aboh, E. (2010). The p route. In G. Cinque and L. Rizzi (Eds.), Mapping Spatial PPs Mapping Spatial PPs,
Volume 6 of The Cartography of Syntactic Structures. Oxford: Oxford University Press. 225–261.
Benedicto, E. and D. Brentari. (2004). Where did all the arguments go? Argument-changing properties of
classifiers in ASL. Natural Language and Linguistic Theory 22(4): 743–810.
Coerts, J. (1992). Nonmanual grammatical markers: An analysis of interrogatives, negations and topicalisations
in sign language of the Netherlands. University of Amsterdam.
Coerts, J. (1994). Constituent order in sign language of the Netherlands and the functions of orientations. In B.
Inger, A. Bergman, and M. Brennan (Eds.), Perspectives on Sign Language Structure: Papers from
the Fifth International Symposium on Sign Language Research, Volume 1. Durham. 69–88.
Jantunen, T. (2007). On topic in Finnish sign language.
Johnston, T., M. Vermeerbergen, A. Schembri, and L. Leeson (2007). Real data are messy: Considering crosslinguistic analysis of constituent ordering in AUSLAN, VGT, and ISL. In P. Perniss, R. Pfau, and M.
Steinbach (Eds.), Visible variation. Berlin: Mouton de Gruyter. 163–205.
Kimmelman, V. (2012). Word order in Russian Sign Language. In Linguistics in Amsterdam. 1–56.
Liddell, S. K. (1980). American Sign Language Syntax. Hague: Mouton Publishers.
Milković, M., S. Bradarić-Jončić, and R. Wilbur (2007). Information status and word order in Croatian
sign language. Clinical Linguistics and Phonetics 21(11/12), 1007–1017.
Pfau, R. and E. O. Aboh (2012). On the syntax of spatial adpositions in sign languages. In MIT Working
Papers in Linguistics, Volume 68. 83–104.
Rizzi, L. (1997). The ne structure of the left periphery. In L. Haegeman (Ed.), Elements of Grammar. Dordrecht:
Kluwer. 281–337.
Rosenstein, O. (2001). ISL as a topic prominent language. Master's thesis.
Sze, F. Y. B. (2008). Topic constructions in Hong Kong Sign Language. Ph. D. thesis.
Svenonius, P. (2006). The Emergence of Axial Parts. In P. Svenonius and M. Pantcheva (Ed.), Adpositions:
Special issue of Nordlyd: Tromsø Working Papers in Linguistics 33(1): 49–77.
104
Volterra, V., A. Laudanna, S. Corazza, E. Radutzky, and F. Natale (1984). LIS: the order of elements in the
declarative sentence. In F. Loncke, P. Boyes-Braem, and Y. Lebrun (Eds.), Recent research on Eu
SL. 19–48.
105
Symmetry in Syntactic Theory:English and Polish
NPN forms
Wiktor Pskit
University of Lodz, Poland
[email protected]
The generativist theorizing has long been concerned with the notions of asymmetry
and antisymmetry. These concepts are assumed to account for the asymmetric aspects of
syntactic structure.
The term asymmetry (or antisymmetry) is employed in a number of ways (e.g. Boeckx
2009). Among others, it refers to the organization of clause structure, to differences between
types of constituents with respect to the internal structure of such constituents or the range of
syntactic operations that apply to the relevant constituents. The crucial role of asymmetry is
emphasized in Kayne (1994), where the asymmetric c-command relation underlies linear
order, and in Moro (2000), where movement is motivated by the need to eliminate symmetric
relations that arise in the course of derivation.
On the other hand, symmetry can be viewed as a theory-internal notion since much of
generativist research has been guided by the idea of symmetry between structure of the verbal
domain and the nominal domain. Thus, in the era of the Government and Binding theory, the
discovery of rich functional layer within the traditional VP inspired the investigation of a
similar array of functional projection within the traditionally conceived NP.
More recently, Citko (2011) explores the notion of symmetry in the context of the
Minimalist Program, concluding that, while asymmetry is basic to grammar, symmetry is
found in the operations Merge and Move as well as in the labelling of products of these
operations.
This paper focuses on the structure of NPN forms in English (e.g. day by day) and
Polish (e.g. dzień za dniem). English NPNs are discussed by linguists representing various
theoretical persuasions, among others Pi (1995), Travis (2001, 2003), Beck and von Stechow
(2007), Jackendoff (2008), Haïk (2013), and Zwarts (2013). Fewer publications are devoted to
their Polish counterparts (Daszczyńska (1997), Dobaczewski (2009), Rosalska (2011), Pskit
(2015)).
106
Most of the authors point to a number of idiosyncratic properties of such structures: a
limited set of prepositions occurring in NPNs, constraints on the type and form of the nouns,
problematic categorial status of NPNs (NP, PP or still something else?). While Polish NPN
forms share a number of similarities with their English counterparts, they also exhibit a
number of differences. For instance, one of the striking features of English NPNs is the
absence of determiners on the nouns (including singular countable nouns). Polish NPNs do
not appear to be idiosyncratic in this respect since Polish is an articleless language and the
structure of NPN forms does not depart from the standard structure of NPs with embedded
PPs: nouns without determiners are expected anyway.
One of the most interesting aspects of NPNs is the relation between a sort of doubling
in form (in the case of NPNs with identical nouns) and a sort of iteration (of entities or events)
in meaning. The doubling in form can be viewed as surface symmetry with a preposition
being preceded and followed by two occurrences of the same noun. Indeed, Haïk (2013)
refers to such NPNs as symmetric structures. In Minimalist terms, this apparent surface
symmetry can be accounted for in terms of the mechanism of syntactic (iterative)
reduplication, which is responsible for the derivation of NPNs, as suggested in Travis (2001,
2003). Syntactic reduplication, triggered by a reduplicative head Q, can be regarded as a
special type of movement (Internal Merge), where copying does not entail deletion.
While stressing the advantages of syntactic reduplication as a way of deriving NPNs in
both English and Polish, this paper also points out a number of problems, including the fact
that the relevant operation only accounts for NPNs with two identical nouns or the
complication of the derivational apparatus with the postulated operation of iterative
reduplication.
References
Beck, S. and A. von Stechow. (2007). Pluractional adverbials. Journal of Semantics 24: 215–254
Boeckx, C. (2009). On the locus of asymmetry in UG. Catalan Journal of Linguistics 8: 41–53.
Citko, B. (2011). Symmetry in Syntax. Merge, Move, and Labels. Cambridge: Cambridge University Press.
Daszczyńska, Izabela. (1997). Reduplikacja jako jeden ze sposobów pierwotnej derywacji frazeologicznej. In:
M. Blicharski and H. Fontanski (eds.) Współczesne tendencje rozwoju języków słowiańskich. Vol. 2.
Katowice: Wydawnictwo Uniwersytetu Slaskiego. 54–62.
Dobaczewski, Adam. (2009). Operacje iterujące w języku polskim (wprowadzenie do opisu). Poradnik Językowy
9: 26–36.
Haïk, I. (2013). Symmetric structures. Corela 11:1, accessed September 08, 2013, http://corela.edel.univpoitiers.fr/index.php?id=2875.
Jackendoff, R. (2008). Construction after construction and its theoretical challenges. Language 84: 8–28.
Kayne, R. (1994). The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press.
Moro, A. (2000). Dynamic Antisymmetry. Cambridge, Mass.: MIT Press.
Pi, C.T. (1995). The structure of English iteratives. Canadian Linguistics Association Proceedings (Toronto
Working Papers in Linguistics). 434–445.
107
Pskit, W. (2015). The categorial status and internal structure of NPN forms in English and Polish. In: A.
Bondaruk and A. Prazmowska (eds.) Within Language. Beyond Theories. Vol. 1. Studies in
Theoretical Linguistics. Newcastle upon Tyne: Cambridge Scholars Publishing. 27–42.
Rosalska, P. (2011). Krok po kroku, krok za krokiem, krok w krok – konstrukcje składniowe czy jednostki
języka?. Linguistica Copernicana 2(6): 149–62.
Travis, L. (2001). The syntax of reduplication. Proceedings of NELS 31: 455–69.
Travis, Lisa. (2003). Reduplication feeding syntactic movement. In: S. Burelle and S. Somesfalean (eds.)
Proceedings of the 2003 Annual Conference of the Canadian Linguistic Association. Université du
Québec à Montréal, Département de linguistique et de didactique des langues. 236–247.
Zwarts, J. (2013). From N to N. The anatomy of the construction. Linguistics and Philosophy 36: 65–90.
108
Types and degrees of vowel neutrality
Péter Rebrusa, Miklós Törkenczyb
a
Research Institute for Linguistics (Hungarian Academy of Sciences), bELTE University
[email protected], [email protected]
Traditionally, the distinction between neutral and harmonic behaviour in vowel
harmony has been considered a categorical property of (sets of) vowels (e.g. van der Hulst –
van de Weijer 1995). In this paper, (where we focus on the ways in which neutrality is
realised in suffixed forms and do not consider monomorphemic “mixed roots”), we argue that
neutrality is not categorical. The graduality of neutrality has been discussed or suggested in
the literature (e.g. Anderson 1980) but has never been given a general explicit
characterisation. We argue here that neutrality can manifest itself in different ways in
different harmony systems since it derives from the ability to occur in more than one pattern
of neutrality and a given pattern (a) may or may not be present in a system, (b) may or may
not combine with the other patterns and (c) may be subject to variation and the occurrence of
variation in a given pattern influences the degree of neutrality. Based on this, we will set up a
scale of neutrality and offer a tentative quantification. We will also show that in languages
with more than one neutral vowel, the different neutral vowels may be of different types and
degrees of neutrality.
Patterns of neutrality
In a harmonic system, vowels are biased to co-occur in such a way that they agree in
the value of a designated feature (the harmonic feature) within a morphologically and/or
phonologically circumscribed domain. In such a system, vowel neutrality can be identified as
the lack of such bias: (i) a neutral vowel is unbiased as a ‘target’ in that it can systematically
co-occur with vowels of either harmonic value (pattern: invariant suffixes) and it is unbiased
as a ‘trigger’: (ii) it does not modify the bias imposed by other (harmonic) sources (pattern:
transparency to the propagation of harmonic features) and (iii) it does not impose its own
harmony requirement on other vowels (pattern: anti-harmony requiring suffixation with
disharmonic suffix alternants). This is shown in (1) below for front/back harmony, where
neutral vowels are phonetically front (B/F/N: back/front/neutral vowels, [ ]: stem boundaries).
109
(1) Patterns of neutrality in suffixed forms
i. invariant suffixes (IS)
[[B]N] and [[F]N]
ii. transparency (TP)
[[BN]B] and [[FN]F]
iii. anti-harmony (AH)
[[N]B]
The patterns (1i-ii) above involve both harmonic values, i.e. invariant suffixes
necessarily occur after both front and back stems, and transparency typically consists in the
neutral vowel’s lack of bias while passing on both the front and the back bias of the preceding
harmonic vowel. Anti-harmony (1iii), however, only involves a harmonic value in the suffix
which is the opposite of the phonetic value of the (neutral vowel-)stem, i.e. [[N]B], but not
necessarily [[N]F].
Typology
Each of the patterns in (1) may be present or absent in a given harmony system (cf.
Kiparsky – Pajusalu 2003). Thus neutrality can manifest itself in more than one way and the
different patterns may or may not co-occur in a given system. Therefore, it is possible to
define degrees of neutrality and set up a classification of systems according to the types of
neutrality they permit. In (2o-iii) below we show four language types, which exhibit different
types of neutrality of N. Since we defined neutrality as the lack of bias, the different
properties of the patterns (1i-ii) vs. (1iii) have a consequence on the degree of neutrality. The
presence of IS and/or TP truly means the lack of bias since N can occur after both harmonic
values (IS) and both harmonic values can occur after N (TP). By contrast, invariable AH is a
kind of bias because only B can occur after N in a suffix. Therefore, none of the language
types (2o-iii) in (2) have maximal neutrality, because neither the ones without AH (2o-ii), and
the ones with AH (2iii) have 'ideally' neutral properties, i.e. a complete lack of bias in this
respect.
(2) Some language types on the basis of neutrality patterns (1i-iii)
language types
o. no neutral vowel – Turkish
i. only IS, no TP/AH – Eastern Khanty /i/
ii. IS & TP, no AH – Finnish /i/, /e/
iii. IS & TP & AH – Uyghur /i/
iv. IS & TP & variable AH – Hungarian /iː/
(i) IS
−
+
+
+
+
110
(ii) TP (iii) AH
−
−
−
−
+
−
+
+
+
+/−
neutrality
no
minimal
intermediate
intermediate
maximal
Variation
Patterns (1i-iii) can occur in variation too: (i) when a given vowel occurs in both
invariant and harmonically alternating suffixes in a system, or (ii) in the case of variable
transparency–opacity [[BN]B/F] or (iii) in the case of variable anti-harmony [[N]B/F]. This
yields further language types, e.g. (2iv) above, where maximal neutrality occurs because
when AH is variable (indicated by +/−), the vowel of the all-neutral root imposes no bias at all
on the suffix (i.e. both B and F can occur). This is the case in Hungarian for the vowel /iː/,
where IS and TP are invariable and AH is variable.
Graduality
Neutrality may be non-homogeneous in a given harmonic system. Different neutral
vowels can show different degrees of neutrality because they are involved in the patterns (1iiii) to a different degree. This is referred to as the height effect in Hungarian (e.g. Hayes –
Cziráky Londe 2006): the more high a front unrounded vowel is, the more neutrally it
behaves. This is illustrated in (3) below, where rows contain the different front vowels and
columns show the patterns of neutrality. The numbers in square brackets in the text and in (4)
below indicate the approximate degree of neutrality on a 5-point scale from 0 to 4 (0=nonneutral, 4=totally neutral, 1,2,3=variably neutral). (i) IS: long /iː/ only occurs in invariant
suffixes [4], short /i/ mostly occurs in invariant ones [3], while /eː/ is frequent in both
invariant and alternating suffixes [2], and /ɛ/ occurs in the former only in special diminutive
constructions [1]; (ii) TP: /i, i:/ are always transparent [4], /eː/ is usually transparent [3], while
/ɛ/ is highly variable [2]; (iii) AH: /i, iː/ frequently occur in anti-harmonic stems [4], /eː/
rarely does [2] and /ɛ/ practically does not [0].
(3) Rough estimation of the degree of neutrality for front vowels in H. (5-point scale: 0,...,4)
high iː
front vowels
(= totally neutral)
high i
high-mid eː
low-mid ɛ
round y yː ø øː (= non-neutral)
(i) IS
4
3
2
1
0
(ii) TP (iii) AH
N-score
4
4
4.0 (100%)
4
3
2
0
4
2
0
0
3.7 (92 %)
2.3 (58 %)
1.0 (25 %)
0.0 (0 %)
A more refined way of quantifying vowel neutrality is based on frequency ratios of
items realising these patterns. The paper presents preliminary results pursuing this approach.
111
References
Anderson, L. (1980). Using asymmetrical and gradient data in the study of vowel harmony. In R. Vago (ed.)
Issues in Vowel Harmony. Amsterdam: John Benjamins. 271–340.
Hayes, Bruce, Zsuzsa Cziráky Londe. (2006). Stochastic phonological knowledge: The case of Hungarian vowel
harmony. Phonology 23: 59–104.
Hulst, Harry van der, Jeroen van de Weijer. (1995). Vowel harmony. In John A. Goldsmith (ed.) 1995. The
handbook of phonological theory. Cambridge, MA & Oxford: Blackwell. 495–534.
Kiparsky, Paul and K. Pajusalu. (2003). Towards a typology of disharmony. The Linguistic Review 20: 217–241.
112
Possessives and the NP or DP? Wrangle
Branimir Stanković
University of Niš, Serbia
[email protected]
In this paper I present empirical arguments concerning possessives in article-less
Serbo-Croatian (SC), Slovenian (S) and Chinese language (C) that support a DP analysis of
their nominal expressions: SC, S and C pre-cardinal possessives induce an exhaustivity
presupposition, C possessives can be modified by another possessive (or adjective), alike
English ones, and, most importantly, SC, S and C possessives can be used to mark
definiteness and the interlocutor that introduced the NP referent to the discourse, when in
initial position of phrases such as (taj) tvoj Marijin mamin sin ‘your Mary’s mama’s boy / that
Mary’s mama’s boy of yours’, while blocking left branch and adjunct extractions at the same
time. I will argue that SC possessives are not inherently definite (Progovac 1998, Kuna 2003,
Caruso 2012), but that they can trigger definite interpretation after a movement to the left
periphery, from post-cardinal to pre-cardinal position, where they induce exhaustivity
presupposition to the entire nominal/cardinal expression.
Two approaches to article-less SC, S and C nominal expressions are present in
literature. The DP hypothesis assumes that all languages project determiner phrases above
their NPs, but DPs can be phonologically null in languages lacking (definite) articles (for SC;
Progovac 1988, Leko 1999, Aljović 2000, 2002, Kuna 2003, Rutkowsky & Progovac 2005,
Caruso 2011, 2012, 2013, Arsenijević 2014, Stanković 2014). In the opposing view of the DP
parameter, articles-less languages do not project DPs, which as a consequence brings a list of
fundamental differences in the syntactic behavior between DP and no-DP languages (Corver
1992, Bošković 2008, 2012, Bošković & Gajewski 2009, Zlatić 1997, 1998, Trenkić 2004,
Despić 2011, 2013, Runić 2013). One essential difference between the two language groups
concerns the availability of left branch (LBE) and adjunct extractions (AE) from the rest of
the expression, a feature present only in article-less languages, such as SC (1) (cf. (2)), which
is analyzed as an indication that SC lacks DPs that would serve as a barrier for LBE and AE.
(1) Ovui/lepuj
sam pronašla [ti /tj knjigu.] (SC)
book-A.F.SG
this-A.F.SG/nice-A.F.SG AUX found
Lit. ‘*This/nice I found book’
113
(2) *Thisi /nicej I found ti /tj book.
Bošković (2008) points out that SC possessives can be found in the predicate position,
(3-a), but they cannot be modified by a possessive or adjective, (4-a), while in English the
situation is converse, (3-b) and (4-b).
(3) a. Ova knjiga je moja. (SC)
this book is my/mine
‘this book is mine’
(4) a. *moj/bogati
My/rich
b. *This book is my.
susjedov konj
neighbor’s horse
b. my/rich neighbor’s horse
Partee (2006) contrasts English expressions such as (5-a), which bring the
presupposition that Zhangsan has exactly three sweaters, with C ones (5-b), which do not
bring such presupposition. Including many other languages in perspective, Bošković (2008)
postulates the generalization that possessors may induce an exhaustivity presupposition only
in DP languages.
(5) a. Zhangsan’s three sweaters
b. Zhangsan de
[san jian maoxianyi]
Zhangsan de-POSS three CL sweaters
‘Zhangsan’s three sweaters’
(C)
Some linguists prefer the DP hypothesis, taking other facts in consideration. Recalling
Kuna’s (2003) remark that SC possessive suffixes -ov, -ev, -in are strong definiteness
markers, Caruso (2012) assumes that possessive elements move out of the PossP to Def0 in
order to check their [+def] feature, which is supported by the definite adjectival inflection,
where the successive adjectival modifiers have to be marked definite when they co-occur with
possessive elements:
(6) a. Marijin-ø
star-i
rođak
Mary-POS.NOM old-DEF. cousin
b. *Marijin-ø
star-ø
rođak
Mary-POS.NOM old-INDEF. cousin
(SC)
Despite the presented in (3), there are languages with articles and with predicative
possessives: for example, possessives in Italian and Macedonian can be found in predicate
position (7)-(8).
(7) a. Il libro è mio.
this book is my/mine
b. mio libro
my book
114
(Italian)
(8) a. Ovaa kniga e moja.
this book is my/mine
b. moja kniga
my book
(Macedonian)
In Macedonian, possessives even bear the postpositional definite article in certain
contexts, marking definiteness in Heim’s (1983) sense, or uniqueness in Russell’s (1905)
sense:
(9) a. moja-ta
kniga(definite, unique)
my-DEF.ART book
‘my book’
b. moja-ø kniga(indefinite, non-unique) (Maced.)
my
book
‘a book of mine’
SC possessives can be combined with both indefinite (10-a) and definite ‘determiners’
(10-b):
b. ona tvoja prijateljica(definite)
that your friend
‘that friend of yours’
(10) a. jedna/neka tvoja prijateljica(indefinite)
one/some
your friend
.....‘a / some friend of yours’
(SC)
Also, Aljović (2000, 2001), Marušič & Žaucer (2014) and Stanković (2014) show that
SC and S possessives obtain exhaustivity presupposition in pre-cardinal position:
(11) a. Marijine tri sestre (exhausted)
Mary’s three sisters
‘Mary’s three sisters’
b. tri
Marijine sestre (non-exhausted) (SC)
three Mary’s sisters
‘three of Mary’s sisters’
Moreover, C possessives can be modified by other possessives, as exemplified in (12):
(12) wǒ de
línjū
de
mǎ
I
de-POSS neighbor de-POSS horse
‘my neighbor’s horse’
(Mandarin C)
Finally, SC, S and C possessives, alike English ones (cf. English translation in (13)),
have an atypical colloquial use as a (definite and) discourse marker that the co-locutor
contributed to the discourse by introducing the NP. In (13) the possessive tvoj ‘your’ indicates
that it is the hearer who introduced the NP Marijin mamin sin ‘Mary’s mama’s boy’ to the
discourse.
(13) A: Marijin verenik je razmažen(-i) mamin sin koji ne ume da se spakuje sam.
alone
Mary’s fiancé is spoiled-(DEF.) mama’s son who NEG know to REFL pack
.....‘Mary’s fiancé is a spoiled mama’s boy who doesn’t even know how to pack himself.’
B: [DIS (Taj) Tvoj [Marijin [mamin sin]]] me uopšte ne zanima!
that your Mary’s mama’s son me at all not interests
‘Your Mary’s mama’s boy/That Mary’s mama’s boy of yours doesn’t interest me at all.’
115
In its discourse-related reading, 2nd person possessive precedes all other adjectives,
including other possessives, but it is not predicative (14), and, most importantly, it blocks
LBE (15):
(14) a. *Lepi
tvoj-DIS mamin sin b. *Marijin tvoj-DIS mamin sin c. Verenik je ??taj tvoj.
Handsome your mama’s son
Mary’s your mama’s son fiancé
is that yours
(15) *Lepog / *Marijinog sam video (tog) tvo-DIS maminog sina.
handsome Mary’s
AUX seen that your
mama’s son
Based on the facts presented in (7)-(15), I argue that SC possessives are not inherently
definite, but that they can trigger this interpretation after a movement to some functional
projection of the split-DP, where they induce an exhaustivity presupposition. The initial, 2nd
person possessive in the pseudo-paradoxical phrase (taj) tvoj Marijin mamin sin (‘your
Mary’s mama’s boy’ / ‘that Mary’s mama’s boy of yours’) marks definiteness and the fact
that it is the addressee who introduced the NP to the discourse, while blocking LBE at the
same time, which supports the hypothesis that a d-related functional projection should be
projected on top of SC, S and C NPs.
116
Pragmatic abilities in bilinguals: the case of scalar
implicatures*
Penka Statevaa, Sara Andreettaa, Ludivine Dupuyb, Anne Cheylusb, Viviane
Déprezb, Jean-Baptiste van der Henstb, Jacques Jayezb, Arthur Stepanova, Pia
Živecc, Anne Reboulb
a
Center for Cognitive Science of Language, University of Nova Gorica, Slovenia, bCNRS L2C2, France, cFaculty of Teology, University of Ljubljana, Slovenia
[email protected]
Introduction
Scalar implicatures have been extensively investigated in the recent experimental
literature, but almost exclusively in monolingual speakers. Very little research has yet been
conducted on the pragmatic abilities of multilingual populations, from early bilinguals to L2
learners, a gap that the current study aims to remedy.
Experimental approaches to scalar implicatures
Many of the experimental protocols exploited the fact that utterances with scalar implicature
triggers give rise to two distinct interpretations. (1a) is consistent with:
(1) a. The pianist played some Mozart sonatas.
b. The pianist played some but not all Mozart sonatas.
c. The pianist played some and maybe all Mozart sonatas.
(Pragmatic intepretation)
(Semantic interpretation)
Most studies have used essentially similar paradigms. Participants were presented with
an utterance that can be considered under-informative relative either to a context or
encyclopedic knowledge. Participants were then asked to judge these utterances, in a truth
value judgment task choosing between true/false, agree/disagree; or (less frequently) well
said/not well said. While both answers are correct, the participant favors either the semantic
lower-bounded interpretation (true/agree/well said) or the pragmatic upper-bounded one
(false/disagree/not well said) depending on her answer choice. This type of truth-value
judgment task has formed the staple of experimental work on scalar implicatures in children
(Noveck 2011, Chierchia et al. 2001, Gualmini et al. 2001, Papafragou & Musolino 2003,
*
This research is funded by the European Union’s Seventh Framework Programme for research,
technological development and demonstration under grant agreement no. 613465 (ATHEME).
117
Guasti et al. 2005, Pouscoulous et al. 2007, among others) and adults. Convergent results of
developmental studies have shown that there appear to be a developmental trajectory
regarding children pragmatic abilities, with 5 year-olds making less pragmatic interpretations
than 7 year-olds, who in turn make less pragmatic interpretations than 9 year-olds who still
have not reached adult level competence. However, these developmental results could also be
given another interpretation according to which pragmatic interpretations depend on learning
which items are relevant in each case for constructing alternatives for evaluation, or learning
pragmatic constraints on informativity (Foppolo et al 2012, Barner et al 2011).
The main study on multilingual speakers’ scalar implicatures
was conducted by
Slabakova (2010) on L2 learners. Slabakova asked whether there are pragmatic differences
between native and L2 speakers and if so, whether these difference can be found to persist
even when L2 speakers that have near native competence. The results of our study show that
L2 learners produce more pragmatic interpretations than control monolingual speakers and
therefore, can be assumed to manifest a bilingual advantage in pragmatic abilities. On the
assumption that L2 learner have diminished processing capacities in their L2, these results are
interpreted by Slabakova to provide support for the neo-Gricean theory of implicatures which
gives the pragmatic interpretations (rather than the semantic one) the status of default
interpretation which is easily accessible. These results, however, are not entirely reliable both
because of the choice of experimental material and because of relevant control: the L2
learners were not tested in their L1 language, hence the assumption regarding processing
capacity in L2 cannot be substantiated.
A handful of existing studies assessing the development of pragmatic abilities of early
bilingual speakers in comparison to their monolingual peers (Siegal et al 2007, Siegal et al
2009, Siegal et al 2010, Antoniou et al 2013) sought to address the question of whether
bilingualism could provide certain overall cognitive development advantages, and whether the
development of general pragmatic abililies as a potential advantage (i.e. sensitivity to
pragmatic deviation and comprehension of different kinds of implicatures), could be
correlated with other advantages in the domain of metalinguistic awareness (Cromdal 1999)
and executive control abilities (cf. Bialystok 1999, 2010, 2011, Bialystok 2009 et al. among
others) already discussed in the literature. However, some studies that more specifically
focused on scalar implicature processing have lead to two diverging claims:
1. Early bilinguals significantly outperform monolingually developing children in the
production of scalar implicatures. This advantage could be attributed to a nonlinguistic consequence of bilingualism (Siegal et al 2007).
118
2. There is no robust evidence that suggests a multilingual advantage in the
comprehension of scalar implicatures (Antoniou at al 2013).
Experiment 1
Experiment 1 aimed at replicating Slabakova’s results, avoiding her methodological
problems by testing our multilingual populations in both of their languages. It also
comparatively assessed the impact of late second language acquisition (SLA) as compared to
early bilingualism on the computation of SI. We tested 90 adult participants, including 30
French monolinguals, 30 French learners of English and 30 French learners of Spanish. The
L2 learners were tested in both their L1 (French) and L2 (English or Spanish). Rather
surprisingly, the analysis revealed a significant difference between French monolinguals and
French L2 learners (p < .0001 for both the L2 English group and the L2 Spanish group) with
regard to the proportion of pragmatic answers they gave in French.These differences are due
to the fact that the two multilingual learner groups give considerably more pragmatic
responses in French than the French monolingual speakers when judging the same sentences.
Furthermore, the L2 learners give approximatively the same amount of pragmatic answers in
their respective L2 languages (p = .771 for L2 Spanish in Spanish versus French and p = .330
for L2 English in English versus French). We observe no difference between the performance
of the two groups of L2 learners either with respect to the rate of pragmatic answers in their
L2 (English versus. Spanish, p = .543) or in their L1 (p = .814). Hence Slabakova’s results are
here replicated: in our study, adult L2 learners also gave significantly more pragmatic
interpretations than monolingual adults did. Besides confirming her results, we can further
attest of their crosslinguistic generalization. Indeed given that Slabakova’s study tested
Korean learners of English while ours tested French learners of English and Spanish, it would
appear that the languages involved make no difference. The increase of pragmatic answers
for L2 learners thus seems to be a solid crosslinguistic phenomenon. However, the
explanation Slabakova provides for this increase in pragmatic answers in L2 learners cannot
be sustained. Indeed, as noted, her neo-Gricean lexicalist account which attributes the
observed L2 pragmatic increase to a tendency to favor ‘easiness’ as a consequence of the
augmented pressure of the L2 processing load (recall that for Neo-Gricean, pragmatic answers
are easier) cannot explain why the L2 learners in our study had an identical proportion of
pragmatic answers in both of their languages, i.e in their L1 as well as in their L2. Rather, it
predicts a significant increase in pragmatic interpretation in the L2 only, hence a presumed
significant within-subject difference between the L2 and L1 languages, which we did not find.
119
We will discuss the question of why L2 learners should manifest a comparative increase of
pragmatic interpretation in their L1 as well.
Experiment 2
Experiment 2 served two goals: (i) against the background of Siegal et al (2007) and
Antonoiu et al (2013), we wanted to contribute to the topic of a potential bilingual pragmatic
advantage; (ii) we also aimed at studying the question of whether balanced early bilingualism
affects pragmatic abilities. We tested 31 monolingual Slovenian children and 33 balanced
bilingual Slovenian-Italian children aged 10 – 10.11. Slovenian monolingual subjects saw 20
items (8 target infelicitous some, 4 felicitous some, 4 false all, 4 true all) in Slovenian, while
bilingual subjects were randomly assigned to two groups each of which saw half of all 20
items in Italian, and the other half in Slovenian and vice versa (counterbalanced for language
(see section 6.2). The Slovenian monolingual participants produced a comparable amount of
negative answers to the unfelicitous some condition as the bilingual participants did. The
bilingual children gave slightly more negative answers in Italian than in Slovenian, which,
however, does not quite reach significance (p=0.076). This suggests that the amount of scalar
implicatures generated by children does not seem to be affected by bilingualism, nor does
there seem to be a robust preference in generating more implicatures in any of the two
languages within the bilingual group. These results support Antonious et al (2013) who also
report a lack of significant advantage in the development of pragmatic abilities that can be
linked to (early) bilingualism, and goes against the results of Siegal et al (2007). More
specifically, we found no evidence that early bilingual development affects the development
of pragmatic abilities. In this respect, it is also important to emphasize that the two groups in
our study were carefully constructed so that each participant could come as close as possible
to the profile of a balanced early bilingual Italian-Slovenian speaker and its monolingual
Slovenian peer, respectively. In that sense, we can claim that the empirical basis for our
conclusion is reasonably solid.
References
Antoniou, K., Grohmann, K., Kambanaros, M., and Katsos, N. (2013). Does multilingualism confer an
advantage for pragmatic abilities?. In Online Proceedings Supplement of the 37th Boston University
Conference on Language Development, eds. S. Baiz, N. Goldman, and R. Hawkes.
Bialystok, E. (1999). Cognitive complexity and attentional control in the bilingual mind. Child Development 70:
636-644.
Bialystok, E. (2011). Reshaping the mind: The benefits of bilingualism. Canadian Journal of Experimental
Psychology 65: 229-235.
Foppolo, F., and Guasti, M.T. (2012). Scalar implicatures in child language: give children a chance. Language
120
Learning and Development 8: 365–394.
Grice, P. (1989). Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Guasti, M.T., Chierchia, G., Crain, S., Foppolo, F., Gualmini, A., Meroni, L. (2005). Why children and adults
sometimes (but not always) compute implicatures. Language and Cognitive Processes 20(5): 667–
696.
Noveck, I.A. (2001). When children are more logical than adults: Experimental investigations of scalar
implicatures. Cognition 78: 165–188.
Papafragou, A., Musolino, J. (2003). Scalar implicatures: Experiments at the semantics-pragmatics interface.
Cognition 86: 253–282.
Pouscoulous, N., Noveck, I.A., Politzer, G., Bastide, A. (2007). A developmental investigation of processing
costs in implicature production. Language Acquisition 14(4): 347-375.
Siegal, M., Matsuo, A., Pond, C., and Otsu, Y. (2007). Bilingualism and cognitive development: Evidence from
scalar implicatures, in Proceedings from the Eight Tokyo Conference on Psycholinguistics, ed. Y.
Otsu. Tokyo: Hituzi Syobo.
Siegal, M., Surian, L., Matsuo, A., Geraci, A., Iozzi, L., Okumura, Y., and Itakura, S. (2010). Bilingualism
accentuates children’s conversational understanding. PloS One 5(2): e9004.
Slabakova, R. (2010) Scalar implicatures in second language acquisition. Lingua 120: 2444-2462.
121
Stating the Obvious: Unifying the restrictions on the
subjects of imperatives and directive subjunctives
Adrian Stegovec
University of Connecticut
[email protected]
Imperatives were traditionally seen as impossible in embedded environments (Sadock
and Zwicky 1985; Han 2000, a.o.), but over the past 10–15 years, many counterexamples
were presented, resulting in the current view that embedded imperatives actually do exist.
This talk deals with obligatory obviation effects observed in embedded imperatives and
directive subjunctives in Slovenian, and proposes an account which not only treats the two on
par, but also extends to the restrictions on the subjects of matrix imperatives and subjunctives.
Central issue
Slovenian imperatives can be embedded in speech reports (1) (they can also be
embedded in relative clauses); cf. Stegovec and Kaufmann (to appear) (S&K) for evidence of
true syntactic embedding. But there is a restriction against co-reference between the matrix
and the embedded subject in (2) that S&K do not account for. The same obviation effect also
occurs in directive subjunctive clauses with the particle naj (3), thus mirroring the obligatory
subject obviation of Romance subjunctives (cf. Picallo 1984). In both imperatives and
subjunctives the restriction only applies to subjects; all other referential elements are exempt.
(1)
Rekel
je
da si
pomagaj
sam!
Said-SG.M are-3P.SG that self-DAT help-2P.SG.IMP alone-SG.M
‘He said that you must help yourself on your own!’
(2) * Rekel
si
da si
pomagaj
sam!
said.SG.M are.2P.SG that self.DAT help.2P.SG.IMP alone.SG.M
‘Youi said that youi must help yourself on your own!’
(3)
Rekel
jei
da naj si
pomaga*i,k
sam!
said-SG.M is-3P.SG that let self-DAT help-3P.SG.IND alone-SG.M
‘Hei said that he*i,k must help himself on his own!’
Similarly, there is also a crosslinguistically much more common ban on exclusive 1P
matrix imperative subjects (where I take inclusive 1P to result from 1P+2P features) and 3P
122
matrix imperative subjects, as illustrated for Slovenian in table (4). The table in (4) also
illustrates the ban on matrix subjunctives with 1P and 2P subjects (cf. Schlenker 2005, for the
same ban in French). Matrix imperatives and subjunctives are thus in complementary
distribution, however both constructions are impossible with singular 1P and all exclusive 1P
subjects.
(4)
pomagati (to help)
st
SINGULAR
DUAL
PLURAL
1P (1 person) IMP.
***
pomaga-j-va *(+2P)
pomaga-j-mo *(+2P)
2P (2nd person) IMP.
pomaga-j
pomaga-j-ta
pomaga-j-te
naj pomaga
naj pomagata
naj pomagate
rd
3P (3 person) SUBJ.
Proposal
In embedded contexts, imperatives (5a,b) and subjunctives (5c,d) show the obviation
effect for the two subjects. But note that if the referential information of the “director” (in
embedded contexts the subject of the matrix verb) is syntactically encoded in the directive
clause, the restriction against 1P (singular or exclusive) matrix subjects can be represented the
same way, as shown in (6), as a matrix director is always 1P (= the speaker).
(5)
(6)
 [CP1 proi said [CP2 that prok VIMP ]]
a.
b.
c.
d.
 [CP1 proi said [CP2 that naj prok V ]]
 [CP1 proi said [CP2 that naj pro*i V ]]
a.
b.
c.
d.
 < pro1P
 < pro1P
 < pro1P
 < pro1P
 [CP1 proi said [CP2 that pro*i VIMP ]]
says >
says >
says >
says >
[CP1
[CP1
[CP1
[CP1
pro2P VIMP ]]
pro*1P VIMP ]]
naj pro3P V ]]
naj pro*1P V ]]
I propose that, realized as either imperatives or directive subjunctives, Slovenian
directive clauses always have the syntactic structure in (7). I stipulate that the dir element is
treated as an indexical w.r.t. the computation of binding condition violations, and carries the
referential information of the speaker in matrix directive and the matrix subject in embedded
directives.
(7)
[CP (that) [ diri [TP T [vP SU*i,k v [ApplP (IOi,*k) Appl [VP V (DOi,*k) ]]]]]]
I assume CP and vP are phases (Chomsky 2000), and that a phase is the domain for
Condition B effects. As the subject (SU) is introduced in SpecvP (the phase edge), any coreferring internal argument in vP will cause a Condition B violation. Conversely, a dir co123
referential with the subject, accessible for binding at the phase edge, will cause a Condition B
violation, while, crucially, a dir co-referential only with an inaccessible internal argument will
not.
I show that the “indexical nature of dir” is compatible with a modified version of
Kaufmann's (2012) performative modal analysis of imperatives, needed independently for
Slovenian. Kaufmann proposes that a modal operator OPIMP (corresponding to the dir head
proposed above) is the source of imperative meaning. OPIMP is essentially the necessity modal
‘must’, as analyzed by Kratzer (2012), which can also be used performatively, but differs in
that OPIMP can only be used felicitously in contexts resulting in performative use of the
modal, which is implemented in terms of specific presuppositions triggered by the imperative.
The exact details are left out for reasons of space. S&K modify this to analyze Slovenian
embedded imperatives, where the presuppositions triggered by OPIMP depend on two contexts:
(i) the requirement that the action P described by the imperative constitutes a possible action
for the addressee has to hold in the actual context, and (ii) the aspect that bans “speaker
distancing” from the imperative presupposition holds for the original context. The latter is
illustrated with (8,9), where distancing in the original (reported) context (8) is infelicitous,
while distancing by the speaker of the actual context (9) is felicitous.
(8) # Peter je rekel da me
poslušaj
ampak da tega noče.
Paul is said that me-ACC listen-2P.SG.IMP but
that this neg-want-3P.SG.IND
‘Paul said that you should listen to me but he doesn’t want that.’
(9)
Peter je rekel da me
poslušaj
ampak (jaz) tega nočem.
Paul is said that me.ACC listen.2P.SG.IMP but
I
this not want.1P.SG.IND
‘Paul said that you should listen to me but I don’t want that.’
S&K capture the distancing asymmetry in (8,9) by suggesting OPIMP is semantically
bound by the matrix attitude verb (which is actually very similar to Kempchinsky's (1986;
2009) treatment of subject obviation in Romance subjunctives), which leads to the operator
being interpreted w.r.t. both contexts. I extend this approach further by arguing that through
semantic binding, the subject of the matrix attitude verb becomes co-referential with a
variable within OPIMP, thus encoding the director in OPIMP itself also in matrix contexts.
The analysis extends to matrix directive clauses, where the variable in OPIMP must
always refer to the speaker of the actual context, correctly predicting a binding violation with
1P subjects, cf. (6a,d). The obviation analysis of matrix directives also allows for a new
analysis of questions formed from directive clauses. The ban against 1P subjects in
subjunctives (10a) is lifted in corresponding questions (10b), while matrix imperative
124
questions are impossible (11b). I assume this asymmetry follows from questions being
“addressee oriented”, resulting in the variable in OPIMP referring to the addressee, now
causing a binding violation with 2P (11b) but not 1P (10b) subjects. Such speaker to addressee
shifts in questions also also found in so called conjunct-disjunct agreement systems (see
Person 2012; Wechsler 2012).
(10)
a. * Naj si
pomagam
sam!
let self-DAT help-1P.SG.IND alone
‘I must help myself on my own!’
(11) a. Pomagaj
si
sam!
Help-2P.SG.IMP self-DAT alone
‘Help yourself on your own!’
b. Naj si
pomagam
sam?
let self-DAT help-1P.SG.IND alone
‘Must I help myself on my own?’
b.* Pomagaj
si
sam?
help-2P.SG.IMP self-DAT alone
‘Must you help yourself on your own?’
I further argue that the difference between imperatives and directive subjunctives is
entirely morphological (contra Schlenker 2005; details in the talk), and that the lack of 3P
imperatives (12a,c) and 2P subjunctives (12b,d) does not follow from binding, but from
morphological blocking in a Distributed Morphology approach. I propose that imperative
morphology in Slovenian is the exponent of OPIMP in the environment of 2P complements
(13a), namely when a 2P subject agrees with T0, as TP is the complement of OPIMP.
Subjunctive naj is the elsewhere exponent of OPIMP (13b), and hence occurs with 1P/3P
subjects.
(12) a. 
b. 
c. 
d. 
< pro1P says > [CP1
< pro1P says > [CP1
[CP1 pro1P said [CP2
[CP1 pro3P said [CP2
(13) a.
b.
OPIMP <=> {-j-} / [ ___ [TP 2P]]
OPIMP <=> naj / elsewhere
pro3P VIMP ]]
naj pro2P V ]]
that pro3P VIMP ]]
that naj pro2P V ]]
(= imperative morphology)
(= subjunctive morphology)
Recall also that the subjunctive differs from the imperative in that clitics can intervene
between the verb and naj (cf. the position of the clitic “si” in (1) vs. (3)), and that the main
verb surfaces with indicative morphology. I take this to indicate that not only do 2P features in
T trigger the realization of OPIMP as imperative morphology, but also that OPIMP has a
diacritic which attracts the {V + T} complex to head-move to OPIMP only when T is 2P. In
the subjunctive the OPIMP realizes as naj, and the verbal complex realizes as indicative due to
the OPIMP not being part of the verbal complex. The contrast is illustrated with (14a) versus
(14b).
125
(14)
a. [AGR [OPimp [T [v HELP v ] T{2P,PL} ] OPIMP ] AGR ]
[
[
[
pomaga-
]
-j-
(= imperative)
] -te ]
b. [ OPIMP ] … [AGR [T [v HELP v ] T{3P,PL} ] AGR ]
[ naj ] … [
[
pomaga-] -jo
]
(= subjunctive)
Conclusion
The proposed analysis derives the obviation effect in embedded imperatives and
subjunctives and the subject restrictions in matrix imperatives and subjunctives as a unified
phenomenon, combining insights from the semantics, syntax, and morphology of imperatives.
References:
Han, C.-h. (2000). The structure and interpretation of imperatives. New York: Garland.
Kaufmann, M. (2012). Interpreting Imperatives. Berlin: Springer.
Kempchinsky, P. (1986). Romance Subjunctive Clauses and Logical Form. Ph.D. thesis, UCLA.
Kempchinsky, P. (2009). What can the subjunctive disjoint reference effect tell us about the subjunctive? Lingua
119(12): 1788–1810.
Kratzer, A. (2012). Modals and Conditionals. Oxford: Oxford University Press.
Pearson, H. (2012). The Sense of Self: Topics in the Semantics of De Se Expressions. Ph.D. thesis, Harvard.
Picallo, M. C. (1984). The INFL Node and the Null Subject Parameter. Linguistic Inquiry 15(1): 75–101.
Sadock, J. M. and A. M. Zwicky (1985). Speech act distinctions in syntax. In Language Typology and Syntactic
Description, Volume I, 155–196. Cambridge: Cambridge University Press.
Schlenker, P. (2005). The Lazy Frenchman’s Approach to the Subjunctive. In Romance Languages and
Linguistic Theory 2003. Amsterdam: John Benjamins. 269–310.
Stegovec, A. and M. Kaufmann (to appear). Slovenian Imperatives: You Can’t Always Embed What You Want!
To appear in Proceedings of SuB19.
Wechsler, S. (2012). Self-ascription in conjunct-disjunct systems. Ms. University of Texas at Austin.
126
Two (non-) islands in Slovenian: A study in
experimental syntax
Artur Stepanov, Manca Mušič and Penka Stateva
University of Nova Gorica
[email protected]
Problem
Golden (1996a,b) reports that wh-extraction out of wh-islands and subject islands does
not lead to degradation in acceptability in Slovenian (while other island structures do):
(1) a. Kaj se je Peter spraševal, [kako je Špela popravila _ ]
what REFL is Peter wondered how is Špela fixed
‘??What did Peter wonder how Špela fixed?’
b. Čigavim predlogom se
mu je [ugovarjati _ na oddelčnih sestankih] zdelo nesmiselno?
at departmental meetings seem pointless
whose proposals REFL him is discuss
‘*Whose proposals did to discuss _ at the departmental meetings seem pointless to him?’
At the face of it, this state of affairs is surprising in light of the standard theories of
island effects that predict that movement across an island structure is blocked by a
Subjacency-type constraint (but see, e.g. Rizzi 1982, Stepanov 2007). However, Slovenian
speakers often do not concur with the empirical judgments of sentences such as (1) and/or feel
uncertain or ambivalent with regard to their status. Since there is currently no consensus on
this empirical issue, we set out to elucidate it in an experimental study employing the factorial
design (Sprouse et al 2012, Sprouse et al 2015), which proved an effective tool for detecting
basic island effects, and focusing on wh-islands and subject islands.
Method
The factorial design recognizes two (processing) factors that enter into computation of
the island sentences: their LENGTH, that is, distance between the moved wh-phrase and the
gap (short/matrix, long/embedded), and the island STRUCTURE itself (yes, no). Each of
these factors independently contributes to the lower acceptability of the sentence. A factorial
design is thus a 2x2 design where these two factors are crossed:
127
(2) a. Kdo je omenil,
da je Maja brala revijo?
who is mentioned that is Maja read a journal
‘Who mentioned that Maja read a journal?’
[NON-ISLAND | MATRIX]
b. Kaj je Jure omenil,
da je Maja brala?
what is Jure mentioned that is Maja read
‘What did Jure mention that Maja read?’
[NON-ISLAND | EMBEDDED]
c. Kdo _ se je čudil,
kdaj je Maja brala revijo?
who REFL is wondered when is Maja read a journal
‘Who wondered when Maja read a journal?’
[ISLAND | MATRIX]
d. Kaj se je Jure čudil,
kdaj je Maja brala?
what REFL is Jure wondered when is Maja read
‘??What did Jure wonder when Maja read?’
[ISLAND | EMBEDDED]
By hypothesis, if there is no true island effect, then the overall degradation of an island
sentence is a simple additive effect of the length and structure factors ((2a-2d)=(2a-2b)+(2a2c)). A true island effect in this design is a super-additive effect, whereby the overall measure
of degradation of the critical sentence is greater than the sum of the length and island structure
factors ((2a-2d) > (2a-2b)+(2a-2c); see Sprouse et al 2012).
Predictions
We looked for super-additive effects in Slovenian sentences involving whislands/object extraction (cf. 2d), wh-islands/subject extraction (cf. (3)) and left-branch
extraction from NP subjects (cf. (4); see e.g. Boškovič 2009 for discussion of left branch
extraction in Slovenian):
(3)
Kdo se je Janez čudil,
kdaj _ je pokosil travo?
who REFL is Janez wondered when is mowed the grass
‘*Who did Janez wonder when mowed the grass?’
(4)
Kakšen
je tajnica
mislila, da je _ učitelj
kaznoval dijake?
what-kind-of-MASC is secretary-FEM thought that is teacher-MASC punished pupils
‘What-kind-of teacher did the secretary think that _ punished pupils?’
Wh-islands/subject extraction sentences were chosen with a purpose to investigate a
potential ECP effect on top of the Subjacency effect (Lasnik and Saito 1991). In addition, in
the subject island sentences we manipulated the type of embedded verb, with the aim to test
the claim in Polinsky et al (2013) that subject island effects may be ameliorated in sentences
with unaccusative verbs, as opposed to transitives, based on the data from Russian.
128
Procedure
We used four sets of target sentences such as (2) each for wh-island/object and subject
extraction, and eight sets of sentences for subject islands of which half were with
unaccusative verbs, and half with transitive (64 target sentences total, divided into two lists,
each interspersed with 32 fillers; the order of the items was pseudo-randomized for each list).
The experiment was conducted as a pen-and-pencil or online questionnaire study (60
participants after cleaning). The task was magnitude estimation (Bard et al 1996): after a
preliminary training, subjects were asked to evaluate the acceptability of each sentence with
regard to the modulus sentence assigned the value 100. There was no time limit on the task.
Results and Discussion
Prior to the analysis, the raw results were z-score transformed to avoid scale-related
biases. We used linear mixed models with LENGTH and STRUCTURE as fixed factors and
subject and items as random factors, as well as TYPE (unaccusative, transitive) as an
additional fixed factor for subject islands.
Main effects of LENGTH and STRUCTURE in all types of island structure under
question were revealed (see Figure 1 and Table 1). Furthermore, for the wh-island/object
extraction structure, we observed no interaction of LENGTH and STRUCTURE, a typical
additive effect illustrated by the parallel lines on the graph. This suggests that although
Slovenian speakers are sensitive to both factors each of which reduces overall acceptability
(contra Golden 1996a,b), there is no true island effect observed in this structure.
In contrast, the wh-subject/subject extraction structure shows a clear interaction of
both factors, but the pattern of judgments is actually sub-additive, rather that super-additive.
This points to a previously unnoticed reverse island effect, similar to that observed in Sprouse
et al (2011) for multiple wh-questions in English. In the spirit of Sprouse et al’s (2011), we
suggest that this reverse island effect is a result of processing considerations that affect global
judgment: in this case, the pro-drop nature of the language intervenes in the online processing
of the subject gap, creating an illusion (a “good enough” structure) of better grammaticality.
Finally, the subject island structure shows a marginal super-additive effect. Here we
believe the results go in the direction of showing a real island effect, again contrary to Golden
(1996a,b). In addition, there was no main effect of TYPE, suggesting that unaccusativity does
not play a role. We thus found no support of Polinsky’s et al (2013) unaccusativity factor as
relevant for the subject island. Overall, our results contribute to establishing a solid empirical
129
base for further investigations of the island effects in Slovenian as well as in the context of the
general syntactic theory.
Figure 1. Interaction plots for each island type (N = 60).
WH-object
χ2
p
32.03 <.0001
39.08 <.0001
.04
.8382
WH-subject
χ2
p
52.15
<.0001
28.17
<.0001
6.80
.0091
Subject
FULL 2 x 2 MODEL
χ2
p
main effect of LENGTH
13.903
.0009
main effect of STRUCTURE
6.53
.0368
interaction LENGTH x STRUCTURE
3.106
.078
TYPE (unacc., transitive)
0.5001
0.4794
Table 1. χ2 and p-values for the two-way linear mixed-effects models for each island type
References
Bošković, Ž. (2009). The NP/DP analysis and Slovenian. In Proceedings of the University of Novi Sad Workshop
on Generative Syntax, 53–73. Novi Sad.
Golden, M. (1996a). Interrogative wh-movement in Slovene and English. In: Mind & language, Dettelbach: J. H.
Röll. 145 –187.
Golden, M. (1996b). K-premik in skladenjski otoki v slovenski skladnji. Razprave. Classis 2, Razred za filološke
in literarne vede, Philologia et litterae 15: 237 –253.
Lasnik, H. and M. Saito (1991). Move alpha. Cambridge: MIT press.
Polinsky M., Gallo CG, Graff P, Kravtchenko E, Morgan AM, Sturgeon A. (2013). Subject islands are different.
In: Experimental syntax and Island effects, ed. J. Sprouse. Cambridge University Press. 286 –309.
Rizzi, L. (1982). Violations of the wh-island constraint and the subjacency condition. In Issues in Italian syntax,
ed. Luigi Rizzi. Dordrecht: Foris. 49–76.
Sprouse, J., S. Fukuda, H. Ono, & R. Kluender (2011). Reverse island effects and the backward search for a
licensor in multiple wh-questions. Syntax 14(2):179 –203.
Sprouse, J., M. Wagers, & C. Phillips (2012). A test of the relation between working memory capacity and
syntactic island effects. Language 88(1): 82 –123.
Sprouse, J., I. Caponigro, C. Greco, and C. Cecchetto (2015). Experimental syntax and the variation of island
effects in English and Italian. NLLT. Online first. 1–38. Doi: 10.1007/s11049-015-9286-8.
Stepanov, A. (2007) The end of CED? Minimalism and extraction domains. Syntax 10: 80–126.
130
Adverbial Left-Branch Extraction and the Structure
of AP in Slavic
Aida Talić
University of Connecticut
[email protected]
This paper shows that adverb extraction (AdvE) out of traditional adjective phrases
(TAPs) is sensitive to the amount of structure projected within the TAP, which I show follows
from a contextual approach to phases; and that the amount of structure projected in the
predicative and attributive position is different. My arguments are based on a cross-linguistic
survey of a number of Slavic languages regarding this extraction, and on a case-study
exploring
phonological,
morphological,
and
syntactic
properties
of
Bosnian/Croatian/Serbian(BCS) short-form adjectives (SFAs) and long-form adjectives
(LFAs).
Adverb Extraction
A survey of Slavic languages regarding examples like (1) reveals two new
generalizations (2):
(1) a. Terriblyi, he was ti tired.
b. *Extremelyi, they met ti smart students.
(BCS, BG, POL, RUS, SLO)
(BCS(long), BG, POL, RUS, SLO)
(2) a. Slavic languages allow AdvE out of predicative position.
b. Slavic languages disallow AdvE out of attributive position.
In many languages the morphology of attributive adjectives differs from that of
predicative adjectives: e.g. long form in BCS and Russian, definite form in Icelandic, agreeing
form in Dutch and German. Bailyn (1994) argues that attributive TAPs quite generally must
have a functional projection above the AP. I show that this suggestion coupled with a
contextual approach to phases straightforwardly captures (2). Parallel to (1), Bošković (2008)
observes a correlation between the (un)availability of adjectival left-branch extraction (LBE)
(3) and the presence of articles, stating a generalization that only article-less languages may
allow LBE, while languages with articles never allow it.
131
(3) (*)Smarti they are ti students.
(4) a
XPAP
b.
PIC 
Anti-locality -
AP
Anti-locality Adv
AP
XAP
AP
Adv
AP
Under a contextual approach to phases, Bošković (2013) argues that every lexical
category projects a phase and that the highest projection in the extended domain of every
lexical head is a phase. Concerning LBE, Bošković argues that the highest projection in the
extended domain of N in all languages is a phase, and that the variation regarding LBE
follows from the presence of the DP layer in languages with articles and the lack thereof in
languages without articles (Corver 1992; Zlatić 1997; Bošković 2008), and an interaction of
locality constraints: (i) the PIC, under which only the edge of a phase can be moved out of it;
and (ii) anti-locality, a ban on movement that is too short which requires movement to cross at
least one full phrase (not just a segment). Assuming adjectives are NP-adjoined (Corver 1992;
Bošković 2008), the DP (phase) blocks adjectival LBE in languages with articles since such
extraction violates either the PIC or anti-locality; LBE is not blocked in languages that lack
DP, given that adjectives originate at the edge of the nominal phase. In sum, the amount of
structure projected within the extended domain of a lexical category correlates with extraction
possibilities of elements contained in it. Regarding (1-2), I suggest that the same interaction of
locality constraints is at work. Parallel to N, A projects a phase in its extended domain.
Assuming intensifying adverbs originate as AP-adjoined, attributive and predicative TAPs
differ regarding the availability of AdvE due to the presence of a functional projection XP
above AP in the attributive position, which is missing in the predicative position. In (1a), the
Adv originates at the edge of the AP (phase), and can be extracted without violating the
PIC/anti-locality (4a). (An account of unavailability of (1a) in English is also provided in the
paper). In (1b), the XP layer is present above the AP to which the Adv is adjoined and
functions as a phase as the highest projection in the domain of A (i.e. Adv is not at the edge of
a phase). To move out of XP, the Adv has to stop in SpecXP (phasal edge), due to the PIC,
but this step of movement is too short; it is ruled out by anti-locality (4b).
Short and Long Adjectives
BCS long form adjectives can be used only attributively. Given that AdvE is
disallowed out of TAPs with LFAs, it follows from above that the functional projection
132
associated with the LFA inflection (XP) is within the TAP. In contrast, previous analyses of
LFAs place XP outside of the TAP, i.e. within the traditional nominal phrase (TNP).
However, I provide evidence that XP is indeed a part of the TAP, rather than TNP, based on
prosodic differences between SFAs and LFAs. Contemporary SFA/LFA distinction is almost
entirely prosodic (cf. 5&6) (see Aljović 2002). Out of forty-two pairs of SFA/LFA forms
resulting from seven Cases and three genders in singular and plural, only in
NOM.SG.M
an
overt inflection [-i] occurs in LFA in addition to the prosodic contrast present in other pairs
(glá:dan – glà:dni ‘hungry’). The prosodic differences between SFAs and LFAs at first glance
do not look systematic, and have not yet received an account in the literature: (i) if SFA has a
rising tone it becomes a falling tone in LFA (5a-6a; 5b-6b); or (ii) if SFA has a rising tone, it
shifts one syllable to the left and remains a rising tone in LFA (5c-6c); or (iii) the accentual
difference is neutralized (5d-6d). Falling and rising accents result from the following rules in
BCS (e.g. Inkelas and Zec 1988): (i) In a word with multiple underlying High(H) tones, the
leftmost H wins; (ii) In the absence of underlying H tones, a default Initial H is inserted; (iii)
A syllable has a rising accent if it precedes a winning H (due to H-spreading); (iv) An initial
H is realized as falling.
(5) SFA: a. plá:vo:j
(6) LFA: a. plà:vo:j
‘blue-DAT.SG.F’
[ ´ ] rising accent;
b. glá:dno:j
b. glà:dno:j
‘hungry-DAT.SG.F’
c. visóko:j
c. vísoko:j
‘tall-DAT.SG.F’
[ ` ] falling accent;
d. làbavo:j
d. làbavo:j
‘loose-DAT.SG.F’
bold = locus of
The messy picture in (5-6), however, reveals what the actual LFA inflection is, which
turns out to be different from the standard view under which LFA inflection is assumed to be
added on top of SFA agreement morphemes, with exponents: [-i] for
NOM.SG.M
and -ø
elsewhere. In particular, I propose that the only LFA inflection is a phonemically null
morpheme with a H tone (i.e. X=øH ). SFAs and LFAs have the following morpheme
sequences, with the underlying H tones indicated by [H]:
(7) SFA: a. plá:v-oH:j
ADJ-DAT.SG.F
b. glá:dn-oH:j
ADJ-DAT.F
(8) LFA: a. plà:v-øH -oH:j
b. glà:dn-øH -oH:j
ADJ- X -DAT.SG.F
ADJ- X -DAT.SG.F
‘blue’
‘hungry’
133
c. visók-oH:j
ADJ-DAT.SG.F
d.làHbav-oH:j
ADJ-DAT.SG.F
c. vísok-øH -oH:j
ADJ- X -DAT.SG.F
‘tall’
d. làHbav-øH -oH:j
ADJ- X -DAT.F
‘loose’
In (7-8), the dative suffix [oH:j] has an underlying H tone, which spreads to the
preceding vowel of the toneless ADJ, giving it a rising accent in (7a-c). In contrast to (7a-c), in
(8a-c) the H tone of the dative suffix [oH:j] is not realized, which is indicated by the fact that
the vowel preceding it does not have a rising accent. Instead, the vowel preceding [oH:j] has
its own H in (8a-c). Given that this H tone is missing from the SFAs in (7a-c), the question is
where this H tone comes from. As suggested above, LFA inflection is a null morpheme with a
H tone. Contrary to the standard analysis where LFA inflection is added on top of agreement
(ADJ-DAT.SG.F- X order), I argue this morpheme is located between
ADJ
and agreement in (8)
(ADJ- X -DAT.SG.F order). Not being underlyingly linked to a vowel, the H tone of the LFA
inflection -øH links to the first vowel preceding it, i.e. the final vowel of
ADJ.
If
ADJ
is
monosyllabic, this results in a falling accent (8a-b). If ADJ is polysyllabic, the H tone spreads
to the vowel preceding it, giving it a rising accent (8c). SFA/LFA distinction is neutralized in
(7d)-(8d) due to the underlying initial H tone of the ADJ, which wins in both SFA and LFA as
the leftmost H in the sequence, regardless of the presence of other H tones, and is realized as
falling. The sequences of morphemes in (7-8) represent what SFAs and LFAs look like in PF.
The remaining question is which of these morphemes correspond to heads that project
syntactic structure and which do not. Assuming syntax provides input to PF and LF, elements
that are present in the syntax are expected to have semantic and/or syntactic reflexes.
Elements that have neither syntactic nor semantic effect can be inserted in PF, as argued for
agreement nodes (Embick and Noyer 2007). I suggest that
ADJ
projects AP in both SFAs and
LFAs (4). The LFA inflection (X=øH ) projects XP above AP (4b). The presence of XP in the
syntax is supported by its blocking effect on AdvE with LFAs. X lowers to ADJ in PF by Mmerger (Marantz 1984; Bobaljik 1995). The
DAT.SG.F
suffix realized as [oH:j], marking
agreement with the noun, is added in PF, hence has no semantic or syntactic effect. Finally,
the ending [-i] that occurs only in NOM.SG.M (in addition to øH ) is not LFA inflection. I argue
there are two vocabulary items realizing agreement in
NOM.SG.M:
(i) [-i] is inserted in the
context of X ([NOM.SG.M][-i]/X__); (ii) [-ø] is inserted everywhere else ([NOM.SG.M][ø]/elsewhere). Their choice is determined by the Elsewhere Principle (Kiparsky 1973). The
analysis of LFA inflection as øH rather than [-i] entirely captures the messy situation in (5-6).
References
Aljović, N. (2002). Long adjectival inflection and specificity in Serbo-Croatian. Recherches linguistiques de
Vincennes 31: 27–42.
Bailyn, John. (1994). The syntax and semantics of Russian long and short adjectives. In Proceedings of Formal
Approaches to Slavic Linguistics (The Ann Arbor Meeting) 2, ed. Jindrich Toman. Ann Arbor, MI:
134
MSP.1 – 30.
Bobaljik, J. (1995). Morphosyntax: The syntax of verbal inflection. Cambridge, MA: MIT dissertation.
Bošković, Ž. (2008). What will you have, DP or NP?. North East Linguistic Society (NELS) 37: 101 – 114.
Bošković, Ž. (2013). Phases beyond clauses. In The nominal constructions in Slavic and beyond, ed. Lilla Schürcks et al. M.
de Gruyter. 75 – 128.
Corver, N. (1992). On deriving certain left branch extraction asymmetries. North East Linguistic Society (NELS) 22: 67 – 84.
Embick, D. & R. Noyer. (2007). Distributed Morphology and the Syntax/Morphology Interface. In Gillian Ramchand &
Charles Reiss (eds.), The Oxford Handbook of Linguistic Interfaces. Oxford: Oxford University Press. 289–324.
Inkelas, S. and D. Zec. (1988). Serbo-Croatian Pitch Accent. Language 64: 227-248.
Kiparsky, P. (1973). "Elsewhere" in Phonology. In S.R. Anderson et al (eds.), Festschrift for Morris Halle. Rinehart and
Winston. 93 – 106.
Marantz, A. (1984). On the nature of grammatical relations. Cambridge MA: The MIT Press.
Zlatić, L. (1997). The structure of the Serbian noun phrase. Austin, TX: University of Texas dissertation.
135
Indexical demonstratives and identificational focus
in Hungarian
Enikő Tóth, Péter Csatár
University of Debrecen
[email protected], [email protected]
This paper presents the results of a rating experiment regarding the use of Hungarian
indexical demonstratives. Crosslinguistically, the traditional view about demonstratives, i.e.
their choice is determined by relative distance from the speaker, has been challenged by
various authors (Piwek et al. 2008, Enfield 2009, Diessel 2012). Recently, Tóth et al. 2014
examined the factors influencing the use of indexical demonstratives in Hungarian, Dutch and
English. They concluded that the pattern of demonstratives changes depending on the nature
of context (neutral vs. contrastive): in neutral contexts distance can explain the choice of
demonstratives, but in contrastive contexts distal demonstratives were preferred by native
speakers when referring to entities that were close to the speaker. This means that the
traditional view cannot account for the choice of demonstratives in contrastive contexts. The
present study compares the use of indexical demonstratives in neutral contexts and in a
subtype of contrastive contexts, where contrastiveness is indicated by Hungarian
identificational focus.
In Hungarian, there are two types of demonstratives, ez/ezek ‘this/these’ are called
proximals, whereas az/azok ‘that/those’ are distals. Regarding their uses, indexical
demonstratives are those that are accompanied by a pointing gesture. Levinson (2004) divides
indexical uses into two subcategories: contrastive and non-contrastive uses, these are
illustrated below:
(1) Az
a
futó
nyert,
és
nem ez.
thatthe runner-NOM winand not thisDEM.DIST.NOM.SG
3P.SG.PAST
DEM.NOM. PROX.SG
‘That runner has won the race and not this one.’
(2) Ez-t
a dinnyé-t
kérem.
this-DEM.PROX.ACC.SG the melon-ACC want-1P.SG.PRESENT
‘I want this melon.’
136
Extending Levinson’s view and relying on the analyses of Chafe (1994), É. Kiss
(1998, 2002) and Kaiser (2011), contrastive contexts are defined along the following features:
- linguistic: contrastiveness is explicitly indicated linguistically, for instance by using a
coordinating conjunction with a contrastive sense, e.g., but, or a sentence containing an
identificational focus;
- epistemic: the entities contrasted are highly accessible and compete to be highlighted in the
joint focus of attention (Tóth et al. 2014).
Contexts that do not satisfy the definition above are labelled as neutral. The results of
Tóth et al. (2014)’s experiment show that the pattern of demonstratives (the number of
proximals and distals) is significantly different in neutral and contrastive contexts when the
entity being referred to is close to the speaker.
The present study tests two hypotheses. On the one hand, it wants to reinforce our
previous findings and tests the traditional distance hypothesis in neutral contexts with a
different method. On the other hand, a subtype of contrastive contexts, where contrastiveness
is explicitly marked by identificational focus, is examined (É. Kiss 1998, 2002: 77-82). More
specifically, it is tested whether utterances with distal demonstratives in identificational focus
receive higher ratings than utterances with proximal demonstratives in focus.
An example of Hungarian identificational focus, where the focussed constituent is a
DP, is shown below:
(3)
Az-t
that-DEM.DIST.ACC.SG
‘I’ll buy that book.’
a
the
könyv-et
book-ACC
veszem
buy-1P.SG.PRES
meg.
PREVERB
Identificational focus in Hungarian is marked by stress (set in bold), and the focussed
constituent moves into a preverbal position. Moreover, if the verb contains a preverb (see the
example above), it will leave its verb and move into a position that is immediately after the
verb.
The function of focus is defined by É. Kiss (1998, 2002) as follows: “The focus
represents a proper subset of the set of contextually or situationally given referents for which
the predicate phrase can potentially hold; it is identified as the exhaustive subset of this set for
which the predicate phrase holds.” (É. Kiss 2002: 78). É. Kiss (1998) also argues that
Hungarian identificational focus can be [± contrastive]. More specifically, it is [+ contrastive]
“if it operates on a closed set of entities whose members are known to the participants of the
discourse. […] In this case, the identification of a subset of the given set also identifies the
137
contrasting complementary subset.” (É. Kiss, 1998: 267) In other words, due to the exhaustive
interpretation of identificational focus, a kind of contrast is created. Moreover, when the verb
contains a preverb and it moves into a postverbal position, the exhaustive interpretation is
obligatory (Balogh, 2009: 139).
The aim of our experiment was to test the two hypotheses mentioned above. As
suggested by the results of various experiments, language users have strong intuitions about
the use of demonstratives (see for instance Maes and De Rooij 2007). In order to ensure
maximum reliability of our test we created rich contexts (Meibauer 2012) by using Lego
DUPLO figures representing a zoo scenario. The task of the participants was to evaluate the
appropriateness of a reply to a question on a Likert scale in a context represented by an
introductory sentence and a picture. A four-point scale was used, i.e. the neutral option from
the middle was removed. This is sometimes labelled as a forced choice method.
36 Hungarian native speakers filled in an online questionnaire. We used a 2x2
between–subjects design, with the factors (± contrastive and ± proximal). The test included 4
items in each condition and 8 filler contexts. The items were presented in a pre-set random
order. For instance, a test item is shown below:
(4)
A két gondozó délután az állatorvoshoz készül a zsiráfokkal.
‘The two zoo tenders are going to take the giraffes to the vet in the afternoon.’
Melyik zsiráfot visszük el délután az állatorvoshoz?
‘Which giraffe are we taking to the vet?’
Azt a zsiráfot (rámutat) visszük el.
‘It’s that giraffe (pointing to the giraffe) that we are taking.’
First we compared (−contrastive, ± proximal) conditions. As expected, there was a
significant difference between the ratings of utterances (sign test, z= –2.00, p < 0.05), i.e.
utterances with proximal demonstratives were preferred when referring to entities that were
close to the speaker. This means that in neutral contexts distance indeed plays a crucial role.
In order to explore the ratings of utterances across the four conditions we used the
Friedman test, which yielded a significant result ((χ2(3) = 21.85, p < 0.01). Post-hoc tests
(Wilcoxon signed rank tests) detected a significant difference between the acceptability
ratings of –proximal sentences in contrastive and neutral contexts, i.e. in contrastive contexts,
which have been marked by identificational focus, distals were judged to be more acceptable
than in neutral contexts. There is no significant difference between the ratings of
138
(±contrastive, −proximal) and (+contrastive, ±proximal) utterances. However, this is not
unexpected, since the entities being referred to were always close to the speaker.
To conclude, the results provide converging evidence and reinforce that distance plays
a crucial role in neutral contexts. Moreover, since utterances with a distal pronoun in
identificational focus received significantly higher ratings on the Likert-scale, the results also
provide evidence in favour of the exhaustive interpretation of identificational focus in
Hungarian under the condition that identificational focus is explicitly indicated by moving the
preverb to a postverbal position.
References
Balogh, K. (2009). Theme with Variations: a Context-based Analysis of Focus. ILLC Dissertation Series.
Amsterdam.
Chafe, W. (1994). Discourse, Consciousness, and Time: The Flow and Displacement of Conscious Experience in
Speaking and Writing. Chicago: University of Chicago Press.
Diessel, H. (2012). Deixis and demonstratives. In: Claudia Maienborn, Klaus von Heusinger and Paul Portner
(eds.) Semantics. An Inter-national Handbook of Natural Language Meaning. Berlin: Mouton de
Gruyter, Vol. 3. 2407–2431.
Enfield, Nick. J. (2009). The Anatomy of Meaning. Speech, Gesture, and Composite Utterances. Cambridge:
Cambridge University Press.
Kiss, É. K. (1998). Informational focus vs. identification focus. Language 74(2): 245–73.
Kiss, É. K. (2002). The Syntax of Hungarian. Cambridge, UK: Cambridge University Press.
Kaiser, E. (2011). Salience and contrast effects in reference resolution: The interpretation of Dutch pronouns and
demonstratives. Language and Cognitive Processes 26(10): 1587–1624.
Levinson, C. S. (2004). Deixis and Pragmatics. In: Laurence R. Horn and Gregory Ward (eds.) The Handbook of
Pragmatics. Oxford: Blackwell. 97–121.
Maes, A. and C. de Rooij. (2007). (How) Do demonstratives code distance? In: António Branco, Tony McEnery,
Ruslan Mitkov and Fátima Silva (eds.) Proceedings of DAARC 2007. Lagos Portugal: Centro de
Linguistica da Universidade de Porto. 83–89.
Meibauer, J. (2012). What is a context? Theoretical and empirical evidence. In: Rita Finkbeiner, Jörg Meibauer
and Petra B. Schumacher (eds.) What is a Context? Linguistic Approaches and Challenges.
Amsterdam: John Benjamins. 9–32.
Piwek, Paul, Robbert-Jan Beun and Anita Cremers. (2008). ‘Proximal’ and ‘distal’ in language and Cognition:
evidence from deictic demonstratives in Dutch. Journal of Pragmatics 40: 694–718.
Tóth E., Péter Csatár and Arina Banga. (2014). Exploring Hungarian and Dutch gestural demonstratives. In: Ludmila
Veselovská and Markéta Janebová (eds.) Complex Visibles Out There. Proceedings of the Olomouc
Linguistics Colloquium 2014: Language Use and Linguistic Structure. Olomouc: Palacký University.
607–625.
139
Emphasis and the syntax of particle verbs
Andreas Trotzke, Stefano Quaglia
University of Konstanz
[email protected], [email protected]
In this talk, we propose a unified analysis for topicalization patterns of German particle
verbs. We first suggest a new classification that not only distinguishes between fully
transparent and fully non-transparent particle verbs, but also identifies intermediate classes.
We show that the picture is even more complicated by demonstrating the existence of the
additional class of expressive particle verbs. On this basis, we propose that if particles are
topicalized to the German left periphery, they must target a position that is related to a special
pragmatic interpretation that cannot be identified with the information-structural notion of
contrast. Instead, we suggest a notion of emphasis that can capture the displacement either of
elements that can be construed contrastively or of elements denoting an upper point on a scale
of remarkability.
Particle topicalization and semantic transparency
One major condition on particle topicalization consists in the possibility of attributing
a contrastive interpretation to the particle (e.g. McIntyre 2001; Müller 2002; Zeller 2001).
This explains the contrast between (1) and (2):
(1) Zu
hat er die Tür gemacht (und nicht auf).
Close-PART has he the door made
and not open
‘He closed the door.’
(2) *Auf
hat Peter mit dem Trinken
up-PART has Peter with the drinking
‘Peter stopped drinking.’
gehört.
heard
(Zeller 2001: 89)
(Zeller 2001: 90)
While the particle topicalized in (1) may enter a relation of paradigmatic semantic
opposition with the particle auf in auf-machen (lit. ‘open-make’, to open), the particle auf in
auf-hören enters no such paradigmatic opposition (cf. #zu-hören, #ab-hören etc.). We refer to
cases such as (1) as ‘transparent’ particle verbs (i.e. the particle can be contrasted and can
occur in a predicative construction, cf. Die Tür ist zu/auf ‘The door is shut/open’). Cases like
(2) are referred to as ‘non-transparent’ cases (i.e. the particle cannot be contrasted and cannot
140
occur in a predicative construction). Combining insights from different approaches to the
idiomaticity of particle verbs, we thus propose a decomposition of semantic transparency in
terms of two features [±predicative] and [±contrast]. However, we present empirical evidence
from an acceptability study showing that not all [–contrast, –predicative] particle verbs
behave equally with respect to topicalization (Trotzke, Quaglia, and Wittenberg in press).
Specifically, cases such as runter-machen (‘to bash sb.’) are more acceptable than cases such
as aufhören (cf. (2) above).
(3) Runter
hat sein Chef ihn vor
allen Kollegen gemacht.
down-PART has his boss him in.front.of all colleagues made
‘His boss bashed him in front of all his colleagues.’
A new class: Expressive particle verbs. The effect we see in (3) is not predicted by the
constraint related to contrastability, according to which all non-transparent particle verbs
should behave alike. An obvious account in the case of runter-machen would be to follow
Stiebels and Wunderlich (1994), who argue that topicalization of resultative or directional
particles (like runter) improves the acceptability of such constructions. We hypothesize,
however, that this is not the whole story. Observe the following minimal pair:
(4) Stell Dir vor! (‘Guess what!’):
a. Raus
hat Costa Rica die Engländer bei der WM gekickt!
the English-PL at the WM kicked
out-PART has C. R.
‘The team of Costa Rica kicked out the English team in the world championship.’
b. * Raus
hat die Band
ihr neues Album gebracht!
out-PART has the band
their new album brought
‘The band published their new album.’
(4) shows that the particle raus can be topicalized in (4a) and not in (4b), although both
rauskicken (4a) and rausbringen (4b) feature particles with a clear directional semantics and
bear the same feature specification, i.e. [–contrast, +predicative]. Accordingly, the difference
must be due to another factor. While rauskicken entails that a team has eliminated a
competitor in a stunning way, rausbringen does not refer to any such remarkability scale that
could serve as basis for expressing evaluation by the speaker: either the band published or did
not publish their album. Applying diagnostics from degree semantics (Morzycki 2012), we
see that while both rauskicken (4a) and runtermachen (3) accept ‘extreme’ degree modifiers
such as regelrecht (‘downright’), particle verbs like rausbringen (4b) are not compatible with
such modification.
141
(5) a. Costa Rica hat die Engländer regelrecht rausgekickt.
C. R.
has the English-PL downright out-PART.kicked
‘The team of C. R. downright kicked out the English team in the world championship.’
b. Sein Chef hat ihn regelrecht runtergemacht.
his boss has him downright down-PART.made
‘His boss downright bashed him.’
c. *Die Band hat ihr neues Album regelrecht rausgebracht.
the band has their new album downright out-PART.brought
‘The band downright published their new album.’
Given what we said concerning the expressive contribution of these modifiers, we thus
see that certain particle verbs are expressive, while others are not.
Emphasis and particle topicalization
We observe that verbal particles cannot be scrambled in the middle field of the
German clause (6), unlike PPs, cf. also Zeller (2003). As a consequence, they cannot be
fronted to the left periphery by Formal Movement (Fanselow 2003) and must target a position
in the left periphery that is, according to Frey (2004, 2010), related to a special pragmatic
interpretation (KontrP in (7)).
(6) a. *(dass) aus
die Männer gestiegen sind.
climbed are
that out-PART the men
‘… that the men stepped out of the car’
b. Aus sind die Männer gestiegen (und nicht ein).
(7) [KontrP AUSi [FinP sindj [TopicP [TP die Männer [VP gestiegen ti tj]]]]]
In contrast to Frey, though, we argue for a composite notion of left peripheral
emphasis that captures the displacement both of elements that can be construed contrastively
and of elements denoting an upper point on a scale of remarkability, as is the case in left
peripheral fronting in (3) and (4a). To implement the insight that a projection associated with
a contrastive interpretation (KontrP) may be too narrow to account for the topicalization
patterns we see in the context of particle verbs, we adopt a distinction from the phonetic
literature (e.g. Niebuhr 2010) between information-structural emphasis (‘emphasis for
contrast’) and non-information-structural emphasis (‘emphasis for intensity’). According to
this distinction, ‘emphasis for intensity’ refers to evaluative implicatures in the pragmatic
domain of speaker attitude (Trotzke and Turco 2015).
142
References
Fanselow, Gisbert. (2003). Surprising specifiers and cyclic spellout. Generative Linguistics in Poland 5: 29 –46.
Frey, Werner. (2004). The grammar-pragmatics interface and the German prefield. Sprache und Pragmatik 52:
1–39.
Frey, Werner. (2010). Ā-Movement and conventional implicatures: About the grammatical encoding of
emphasis in German. Lingua 120: 1416 –1435.
McIntyre, Andrew. (2001). German Double Particles as Preverbs: Morphology and Conceptual Semantics.
Tübingen: Stauffenburg.
Morzycki, Marcin. (2012). Adjectival extremeness: Degree modification and contextually restricted scales.
NLLT 30: 567–609.
Müller, Stefan. (2002). Complex Predicates: Verbal Complexes, Resultative Constructions and Particle Verbs in
German. Stanford: CSLI.
Niebuhr, Oliver. (2010). On the phonetics of intensifying emphasis in German. Phonetica 67:170 –198.
Stiebels, Barbara, and Dieter Wunderlich. (1994). Morphology feeds syntax: The case of particle verbs.
Linguistics 32: 913 –968.
Trotzke, Andreas, Stefano Quaglia, and Eva Wittenberg. in press. Topicalization in German particle verb
constructions: The role of semantic transparency. Linguistische Berichte.
Trotzke, Andreas, and Giuseppina Turco. (2015). The grammatical reflexes of emphasis: New evidence from
German wh-questions. Ms., University of Konstanz & University of Stuttgart.
Zeller, Jochen. (2001). Particle Verbs and Local Domains. Amsterdam & Philadelphia: Benjamins.
Zeller, Jochen. (2003). Moved preverbs in German: Displaced or misplaced? In The Yearbook of Morphology,
eds. Geert Booij, and Ans van Kemenade. Dordrecht: Kluwer. 179 –212.
143
A compositional analysis of French -eur nominalsin
combination with ambiguous adjectives
Lucas Tual
University of Geneva
[email protected]
1 Adjectival ambiguity
This talk investigates the interaction between the semantics of adjectives and their
position within the DP. It concentrates specifically on the French adjective gros in
combination with -eur deverbal nouns.
We will adopt the well-known classification of adjectives into (at least) two categories
(see Kamp and Partee (1995), a.o.) with intersective modifiaers (e.g. red as in “red table”),
and non-intersective modifiers (e.g. skillful as in “skillful surgeon”).
It has been observed that some adjectives are ambiguous between an intersective and a
non-intersective reading (see Larson (1998)). For example, the sentence in (1) is ambiguous
between (i) a reading where Olga is a beautiful person and a dancer (but may dance in an ugly
way), (ii) and a reading where Olga dances beautifully (but may be an ugly person).
(1) Olga is a beautiful dancer.
Larson (1998) proposed a theory in which the noun is analyzed as containing some
event structure (we will denote events with type v), and the ambiguity of the adjective comes
from the fact that it can either modify the individual argument (yielding the reading (i), which
we’ll refer to as the intersective reading) or the event argument (yielding the reading (ii), that
we’ll call the non-intersective reading).
Analyses of this phenomenon have generally started from English data, which only
allows prenominal adjectives. This talk will present data from French, a language in which
adjectives can appear both before and after the noun.
What is striking in French is that, in the case of some adjectives like gros, the
interpretation of the attributive adjective depends on whether it is placed before or after the
noun (this has been observed notably by Bouchard (2002)). As the examples in (2) and (3)
show, there is a strong preference for the non-intersective interpretation when the adjective
appears before the noun (even though an intersective reading seems to be possible). If the
adjective follows the noun, only an intersective reading is possible.
144
(2) a. Un gros fumeur
a heavy smoker
?a fat smoker
b. Un fumeur gros
a smoker fat
*a heavy smoker
(3)a. Un gros dormeur
a heavy sleeper
?a fat sleeper
b. Un dormeur gros
a sleeper fat
*a heavy sleeper
When the adjective gros appears in a predicative position as in (4), or in a position
where the noun has been cliticized by en as in (5) (see Shlonsky (2015) for a cartographic
account of this phenomenon), the adjective can only be interpreted intersectively.
(4) a. Ce fumeur est gros.
This smoker is fat
b. Ce dormeur est gros.
This sleeper is fat
(5) Un fumeur, j’en connais un gros
a smoker, I EN know one fat
Other adjectives like occasionel never have an intersective reading, and they cannot
appear in both position (prenominal or postnominal) when used attributively (see (6)).
(6) a. Un fumeur occasionnel
a smoker occasional
b. ??Un occasionnel fumeur
Generalization: 1. Intersective adjectives can appear in different positions when used
attributively (prenom- inal or postnominal). 2. The position of non-intersective adjectives is
always fixed (either on the left or the right of the NP).
This generalization goes along the lines of previous analyses of adjectives: the “twodomain modification” theory proposed by Larson and Marušicˇ (2004) and the syntactic
analysis of Cinque (2010) (even if Cinque claims that in Romance languages like French,
postnominal adjectives are ambiguous while prenominal ones are not).
2 Analysis
We claim that it is necessary to adopt a decompositional analysis for this type of
modification. Thus, we advocate for an analysis where French -eur nominals have a complex
eventive structure (as it has been proposed by Alexiadou and Schäfer (2010) and Roy and
Soare (2014) notably). This analysis focuses on the compositionality of NPs like fumeur, and
stipulates the presence of a covert operator EUR that is responsible for turning a constituent
containing an unsaturated event argument of type hv,ti into a noun of type he,ti, specifying in
the process the agent of the event.
145
This operator is licensed by the presence of the morpheme -eur, semantically vacuous
but morphologically responsible for the nominalization of nouns like fumeur or dormeur. The
operator EUR has to appear in a position where it has scope over the -eur morpheme, and it
has to have as its sister a constituent of type hv,ti (containing an event argument).
The intersectivity or non-intersectivity of the whole DP depends on whether the
adjective combines with the noun before or after the covert operator. This reasoning is
compatible with an analysis in which the two meanings for the adjective gros actually indicate
that there are two different adjectives – pronounced the same way but with different syntactic
and semantic properties (contra Larson (1998)).
See the structure and composition of the NP “gros fumeur” (with the non-intersective
reading) in (7).
Claims: 1. Intersective adjectives are of type he,ti while non-intersective adjectives are
of more complex types. 2. Intersective adjectives are always merged higher in the structure
(outside the scope of certain opera- tors). 3. Non-intersective adjectives are merged lower in
the structure (within the scope of certain operators).
This analysis is in some ways along the lines of what has been called rebracketing, to
account for the phenomenon of bracketing paradoxes (see Spencer (1988), a.o., and the
examples (8) and (9)), in the sense that it postulates that the adjective gros combines first with
the verbal root fum- before the EUR operator does.
(8) Uneasier: [un- [easi- -er]]⇒ [[un- easi-]-er ]
(9) Nuclear physicist: [nuclear [physics- -ist]]⇒ [[nuclear physics-]-ist]
(10) Gros fumeur: [gros [fum- -eur]⇒ [[gros fum-]-eur]
146
3 Lexical Ambiguity
In our analysis, the two meanings of the adjective gros are captured by assigning it two
different semantics values. Since the intersective adjective gros is of type he,ti while its nonintersective counterpart is of type hvt,vti, this means that there exists two different lexical
items in the lexicon for gros. This is comparable to the analysis proposed by Despic´ and
Sharvit (2008), in which they claim that some Non-Intersective are lexically ambiguous.
However, for our own analysis we could imagine a different story in which the
semantic value of the adjective is underspecified and thus can take predicates of either
individuals or events as its argument. The semantic type of gros would then be hhσ,ti,hσ,tii,
where type σ denotes the union of the domain of individu- als and the domain of events
(De∪Dv). This would mean that even the intersective adjective is of a complex type, but
would also have the advantage of being a more uniform analysis for the ambiguous adjective
gros.
References
Alexiadou, A. and Schäfer, F. (2010). On the Syntax of Episodical vs. Dispositional -er Nominals. In The Syntax
of Nominalizations across Languages and Frameworks.
Bouchard, D. (2002). Adjectives, Number and Interfaces: Why Languages Vary. Elsevier Science Ltd.
Cinque, G. (2010). The Syntax of Adjectives: A Comparative Study. MIT Press.
Despić, M. and Sharvit, Y. (2008). Some “Non-Intersective” Adjectives are Genuinely Noun-Taking. In NELS
39.
Kamp, H. and Partee, B. H. (1995). Prototype Theory and Compositionality. Cognition.
Larson, R. K. (1998). Events and Modification in Nominals. In Proceedings of Semantics and Linguistic Theory
(SALT) 12.
Larson, R. K. and Marušič, F. (2004). On Indefinite Pronoun Structures with APs: Reply to Kishimoto.
Linguistic Inquiry.
Roy, I. and Soare, E. (2014). On the Internal Eventive Properties of -er Nominals. Lingua.
Shlonsky, U. (2015). Some Notes on ne-Cliticization. In 41st Incontro di Grammatica Generativa.
Spencer, A. (1988). Bracketing Paradoxes and the English Lexicon. Language 64(4): 663–682.
147
Events, individuals, their subparts, and two types of
multiplicative adjectives in Polish
Marcin Wągiel
Palacký University in Olomouc / Masaryk University in Brno
[email protected] / [email protected]
There are two classes of multiplicative adjectives in Polish: i) adjectives with the suffix
-krotn- (e.g., dwukrotny ‘two-time’) and ii) adjectives with the circumfix po⟨ ⟩n- (e.g.,
podwójny ‘double’). Both classes are derivationally complex and consist of a numeral root
introducing a number, e.g., ⟦√dw-⟧ = 2. Nevertheless, multiplicative adjectives differ
semantically from cardinal numerals in a significant way. While (1a) dwaj prezydenci (‘two
presidents’) and (1b) dwie korony (‘two crowns’) denote pluralities of two individuals,
(1c) dwukrotny prezydent (‘two-time president’) and (1d) podwójna korona (‘double crown’)
refer, respectively, to a plurality of two events of acquiring a property and a plurality of two
subparts constituting an entity denoted by the noun. Examples such as (1c)–(1d) show that
although both classes of adjectives in question syntactically modify the NP, they semantically
target different items. Furthermore, both -krotn- and po⟨ ⟩n- adjectives can also modify event
nouns, e.g., (2a) dwukrotne morderstwo (‘(series of) two murders’, lit. ‘two-time murder’) and
(2b) podwójne morderstwo (‘double murder’). The difference between the phrases is that (2a)
refers to two separate events of murdering an undefined number of victims whereas in the
case of (2b) there is a single event of murdering two individuals.
In this paper I propose a semantic analysis of phrases such as (1c)–(1d) and (2a)–(2b).
Let us first consider adjectives with the suffix -krotn-. These adjectives target events, thus
they either combine with nouns denoting properties of events, see (2a), or create events as in
(1c). In general, I assume derivations of (1c) and (2a) in the spirit of Krifka’s theory of
measuring and counting (Krifka 1986, 1990, 1995). The suffix -krotn- is basically a measure
word introducing an operator on events BC (for ‘become/cause’) and an additive measure
function μ(BC). The BC operator functions as a predication trigger, i.e., it represents an
eventive requirement for a property to be assigned (cf. primitive predicates in the theory of
the bipartite structure of verb meaning by Rappaport Hovav & Levin 1998). Hence, I model
the relationship between BC and the property assigned in terms of material implication. On
148
the other hand, μ(BC) yields the number of events of assigning a property to an individual.
Subsequently, the numeral root dw- in the multiplicative adjective introduces the cardinality
of denoted events. The resulting predicate can be represented as a deverbal noun in NeoDavidsonian frameworks (Carlson 1984, Dowty 1989, Parsons 1990), i.e., as a relation
between events (the semantic argument) and entities (the syntactic argument). Thus, we get
(4) as a representation of (3a).
(3)
(4)
a.
Jan to dwukrotny
prezydent.
Jan is two-time-NOM president-NOM
‘Jan is a two-time president.’
b.
Jan dwukrotnie został prezydentem.
Jan twice
became president-INSTR
‘Jan became a president twice.’
⟦Jan to dwukrotny prezydent⟧ =
= ∃e[(BC(e) ˄ Exp(e) = jan) → president(jan) ˄ μ(BC)(e) = 2]
In prose, the formula in (4) states that there were two events which triggered assigning
a property of being a president to the Experiencer of those events, i.e., Jan. Such an
interpretation seems to capture the intuition that (3a) can be felicitously paraphrased as (3b) in
which the BC operator is overtly realized by the copular verb został (‘became’). It follows
that adjectives with the suffix -krotn- can modify only nouns denoting a particular stage-level
property that can be repetitively acquired and lost by an individual. I propose that such a
property is denoted only by nouns that can combine with the VP (5) ponownie został
(‘became again’), see (6a)–(6c).
(6)
a.
Jan ponownie został prezydentem.
Jan again
became president-INSTR
‘Jan became a president again.’
b.
#Jan ponownie został mężczyzną.
Jan again
became man-INSTR
c.
#Jan to dwukrotny
mężczyzna.
Jan is two-time-NOM man-NOM
Since being a man is not a stage-level property which can be repetitively assigned to an
individual as witnessed in (6b), the predicate mężczyzna (‘man’) cannot be modified by an
adjective with the suffix -krotn-. Hence, (6c) is infelicitous.
149
In the case of (2a) the -krotn- adjective behaves similarly as in (1c), i.e., it quantifies
over events. However, since the modified noun denotes a property of events, no Experiencer
is needed and (2a) can be represented by the formula in (7) which states that the murder event
was triggered twice.
(7)
⟦dwukrotne morderstwo⟧ =
= λxλy∃e[BC(e) → murder(e) ˄ Ag(e) = y ˄ Th(e) = x ˄ μ(BC)(e) = 2]
In quantificational terms (2a) is equivalent to (8) dwa morderstwa (‘two murders’)
which refers to a plurality of two murder events.
On the other hand, adjectives with the circumfix po⟨ ⟩n- in examples such as (1d) and
(2b) either target subparts of individuals, or subevents of events denoted by the noun. Let us
first discuss (1d). I propose that again the circumfix po⟨ ⟩n- is a measure word. This time,
however, it introduces the SSS operation (for ‘self-sufficient subpart’) which selects a
property P and yields a measure function which takes an individual and returns a number of
its subparts such that each subpart constituting that individual is considered self-sufficient,
i.e., it has itself a property P. After the integer argument gets saturated by the number denoted
by the numeral root we get (9) as a formal representation of (1d).
(9)
⟦podwójna korona⟧ = λx[crown(x) ˄ SSS(crown)(x) = 2]
In other words, if an individual x is a double crown, then it is constituted by two
subparts y and z each of which is a crown (note that it does not mean that predicates such as
(1d) are divisive). As a result (1d) denotes a set of crowns such that they consist of two subcrowns, e.g., a set of ancient Egyptian pschents.
Similarly, in (2b) the adjective with the circumfix po⟨ ⟩n- introduces the SSS
operation; this time, however, for an event property P it yields a measure function measuring
a number of self-sufficient subevents of an event. Likewise self-sufficient subparts, such
subevents have an event property P themselves. Therefore, as a result we get a semantic
interpretation of (2b) in (10).
(10)
⟦podwójne morderstwo⟧ =
= λxλy∃e[murder(e) ˄ Ag(e) = y ˄ Th(e) = x ˄ SSS(murder)(e) = 2]
The formula in (10) states that there were two subevents of a murder event which were
murders themselves. Since a murder event requires a Theme, i.e., victim, and one individual
cannot be murdered repetitively, a side effect of counting self-sufficient subevents by the SSS
150
operation in (10) is quantification over individuals referring to the Theme of an event, i.e.,
murdered victims. Such an analysis seems to capture the fact that (2b) cannot be paraphrased
by (8), but rather by (11) morderstwo dwóch istot (‘murder of two beings’).
In conclusion, adjectives with the suffix -krotn- semantically target events and in order
to quantify over them they either combine with nouns which include an event variable in their
semantics, see (2a), or create an event as in (1c). On the other hand, adjectives with the
circumfix po⟨ ⟩n- quantify over self-sufficient subparts of individuals, see (1d), or events, see
(2b). Similarly to complex entities or events they constitute, such subparts have themselves
the property denoted by a modified noun.
References
Carlson, G. (1984). Thematic Roles and their Role in Semantic Interpretation. Linguistics 22: 259–279.
Dowty, D. (1979). Word Meaning and Montague Grammar. Dordrecht: Reidel.
Dowty, D. (1989). On the Semantic content of the Notion ‘Thematic Role’. In B. Partee, G. Chierchia & R.
Turner (eds.), Properties, Types, and Meanings Vol II. Kluwer. 69–130.
Krifka, M. (1989). Nominal Reference, Temporal Constitution and Quantification in Event Semantics. In R.
Bartsch, J. van Benthem & P. von Emde Boas (eds.), Semantics and Contextual Expression.
Dordrecht: Foris Publication. 75–115.
Krifka, M. (1990). Four Thousand Ships Passed through the Lock: Object-Induced Measure Functions on
Events. Linguistics and Philosophy 13: 487–520.
Krifka, M. (1995). Common Nouns: A Contrastive Analysis of English and Chinese. In G. N. Carlson & F. J.
Pelletier (eds.), The Generic Book. Chicago: University of Chicago Press. 398–411.
Parsons, T. (1990). Events in the Semantics of English: A Study in Subatomic Semantics. MIT Press.
Rappaport Hovav, M. & Levin, B. (1998) Building Verb Meanings. In M. Butt & W. Geuder (eds.), The
Projection of Arguments. CSLI Publications. 97–134.
151
Lexical strata and vowel (dis)harmony: The Turkish
transformation of a Balkan hypocoristic
Mary Ann Walter
Middle East Technical University, Northern Cyprus Campus
[email protected]
Like other languages of the Balkan sprachbund, Turkish employs a hypocoristic suffix
of the form /-oʃ/ in personal names (probably of Albanian origin (Camaj 1984, Newmark et
al. 1982), but also attested in Romanian, Hungarian, and Slavic). In this study I explore the
phonological behavior of this affix in Turkish and its implications for theories of lexical
strata, in which subclasses of the lexicon operate according to different phonological rules (or,
in a constraint-based system, according to co-phonologies (Ito & Mester 2009) or indexed
constraints (Pater 2009)).
Data
1) forms in reference grammars (Lewis 2000), naturalistic contexts, and from 2 Turkish
informants (n=32)
2) a survey questionnaire (n=50 potentially suffixable forms, n(participants)=9 native Turkish
speakers who report that they sometimes (on a scale below often but above never and a
little bit) hear and use this type of nickname
Results
Productivity:
This suffix is not fully productive, with all respondents providing only null outputs for
at least 10% of the potentially suffixable forms in both datasets 1 and 2. There were no forms
which yielded null outputs for all participants, however.
Variation:
1) Many forms have alternants with different suffix vowels (2/32, Dataset 1, 139/450, Dataset
2). One participant alternated for all but 4 non-null stimuli (excluded from Table 1). (2)
While generally suffixation occurred to the maximal amount of root phonetic material
yielding a bisyllabic suffixed form (a restriction not observed in the other Balkan
languages), 2 participants displayed 2 cases of syllable simplification.
2) Four participants generated trisyllabic forms though only sporadically (max 5/participant).
152
Vowel quality:
Despite lack of complete productivity, Turkish has innovated an alternative form of the
suffix with a high front vowel, in line with expectations for this vowel-harmonic language. In
Dataset 1, the forms produced are relatively evenly split between /o/ and /i/ vowels, while
other (high) vowels are attested only sporadically. In Dataset 2, /u/ is also rare, and other
vowels even more so. Replicating Dataset 1, /i/-/o/ predominate with the latter leading.
Table 1. Vowel attestations in hypocoristic /-Vʃ/
suffix vowel null ü ı ö u I
Dataset # tokens
3
1 1 0 2 12
One
# harmonic -1 1 -- 1 8
Dataset # tokens
139 1 0 2 12 114
Two
# harmonic -1 -- 2 8 53
o
15
2
188
66
However, the existence of a potentially harmonizing variant does not result in
consistent vowel harmony in outputs. On the contrary, well over half the suffixed forms are
disharmonic. This surprising finding is not attributable to any property of the root-final
consonant. Individual item analysis is ongoing. While some forms seem to be disharmonically
lexicalized for all participants, others show between-speaker variation.
Conclusions
In adapting a Balkan hypocoristic suffix into a different phonological system, Turkish
speakers unsurprisingly introduced a new front high vowel variant, presumably to satisfy
native vowel harmony requirements. However, such /i/-suffixed forms are still a minority, and
moreover, forms with both /i/ and original /o/ still harmonize less than half the time. Such
behavior parallels what has previously been observed for loanwords in Turkish. Some such
loanwords, though not all, display anti-harmonic behavior in suffixation. Other loanwords
requiring epenthesis surface with disharmonic high front vowels in large numbers (Walter
2014). These vowel harmony violations provide evidence for stratified theories of the lexicon.
The hypocoristic forms analyzed here may be fully nativized with respect to vowel harmony,
but may also be treated as a separate class to which harmony rules do not and cannot apply.
References
Camaj, Martin. (1984.) Albanian grammar. Otto Harrassowitz Verlag.
Ito, Junko and Armin Mester. (2009). Lexical classes in phonology. In Miyagawa, Shigeru, and Mamoru Saito,
eds, Handbook of Japanese linguistics. Oxford University Press.
Lewis, Geoffrey. (2000). Turkish grammar. 2nd edition. Oxford University Press.
Newmark, Leonard, Philip Hubbard, and Peter Prifti. (1982). Standard Albanian: A reference grammar for
students. Stanford University Press.
Pater (2009). Weighted constraints in generative linguistics. Cognitive Science 33: 999 –1035.
153
Walter, Mary Ann. (2014). The significance of vowel insertion in Ottoman Turkish. Paper presented at the 17th
International Conference on Turkish Linguistics. Rouen, France.
154
Numeral phrases as subjects and agreement with
participles and predicative adjectives
Jacek Witkoś and Dominika Dziubała-Szrejbrowska
Adam Mickiewicz University in Poznań
[email protected], [email protected]
The goal of this presentation is to account for the patterns of agreement found between
the verb and the participle/predicative adjective on the one hand and the subject quantified by
higher cardinal numerals (QHP, 5 and onwards) in Polish (and Russian) on the other, e.g. (1-2)
and (3). Subject QHPs in Polish cause default agreement on the verb and do so only optionally
in Russian:
(1) a. Te
trzy
dziewczyny pracowały/*pracowało tam.
these-NOM.PL three-NOM girls-NOM.PL worked -PL/SG
there
b. Te
pięć
dziewczyn *pracowały/pracowało
these-ACC.PL five-ACC girls-GEN.PL worked-PL/SG
(2) a. Pjat’ devushek rabotali/rabotalo
Five girls-GEN.PL worked-PL/SG
tam.
there
tam.
there
b. Eti
pjat’ devushek rabotali/*rabotalo
these-NOM.PL five girls-GEN.PL worked-PL/SG
(Polish)
(Polish)
(Russian)
tam.
there
(Russian)
(3) Pięć nauczycielek
było
wybranych/wybrane do pomocy w egzaminach końcowych.
in examinations final
five teachers-FEM.GEN.PL was-3SG.NEUT chosen-PL.GEN/-PL.ACCto help
‘Five teachers were chosen to help in the final examinations.’
(Polish)
In the majority of analyses, e.g. in Babby (1987; Franks 1994, 1995, 2002; Bailyn
2004; Bošković 2006; Przepiórkowski 2001; Przepiórkowski & Patejuk 2012; Watanabe
2012; Willim 2014), example (3) has been accounted for by resorting to distinct operations of
Agree, i.e. involving either a relation with the accusative numeral, hence the adjective marked
with accusative, or the genitive head noun giving rise to genitive adjective. Typically,
accounts that could elegantly account for (1-2) (cf. the so-called Accusative Hypothesis), face
problems with (3), where two case forms are possible on the participle/adjective.
In our proposal, the agreement patterns above result from a novel application of the
nano-syntactic theory of case (cf. Starke 2009; Caha 2009, 2010). Following the tenets of
155
nanosyntax, case distribution proceeds via movement of a nominal to a position within an
articulated Kase Phrase (KP, Willim 2000; Franks 2002) to acquire a proper case suffix, i.e. to
obtain a morphological case required by a syntactic probe, e.g. from Caha (2010). Case
suffixation can follow either the analytic or synthetic pattern which is tantamount to
successive cyclic movement with pied-piping or a single step movement with no overt
evidence of pied-piping. Movement of QP/NP to a specifier position of a given case
projection is triggered by the external selector which activates a given case region by handing
down its EPP property, e.g. v activates AccP, while T activates NomP:
(4)
[InsP [Inst [LocP [Loc [DatP [Dat [GenP [Gen [AccP [Acc [NomP [Nom [NP ]]]]]]]]]]]]].
...........
After movement(s) within KP is completed the remaining case projections are neglected.
The analysis
We assume the following structure for QHPs where the numeral acts like a nominal
(bears features number, gender and person) and absorbs the structural case, while the NP
complement receives genitive, e.g.:
(5)
[vP v [VP V [QHP Num FQ [FP F [NP N ]]]]]
We also follow Franks (1994, 1995, 2002) and Przepiórkowski (2004) and assume that
QHPs spell out the structural case as accusative. In terms of a split KP analysis T activates the
head of AccP in (4) and the entire QHP moves to [spec, Acc]; accusative on QH is licensed.
But the NP complement is still caseless and needs to move to [spec, Gen] pied-piping the QHP
level dominating it:
(6) [GenP[QHP pięć FQ [NP. nauczycielek]] Gen[AccP [QHP pięć FQ [NP nauczycielek]] Acc [NomP Nom[QP]]]]
step2
step1
The result is the following case (projection) stack, with AccP placed in the specifier
position of GenP:
(7) X →
[GenP [AccP pięć nauczycielek] Gen]
five teachers
Now, consider the definition of closeness from Pesetsky and Torrego (2001), on the
strength of which the maximal projection and its specifier are equally close to a ccommanding probe:
156
(8) a. Attract Closest: If a head K attracts feature F on X, no constituent that bears F is closer
to K than X.
b. Closeness: Y is closer to K than X if K c-commands Y and Y c-commands K.
The structure of the QHP in (7) has the following property: both the maximal
projection of GenP and the specifier of AccP are close(r) to X. Consequently, X can become
involved in Agree with either GenP (for genitive) or AccP (for accusative). When X is T,
default agreement comes out in Polish, as neither layer of the case (projection) stack on the
QHP subject is NomP, cf. (1b). We assume, following Bošković (2006), that Russian QHPs
come in two options (nominative or accusative) where option (9a) can return nominative
agreement on T, cf. (2), e.g.
(9)
a. T
b. T
[GenP [NomP …] … Gen
[GenP [AccP …] … Gen
(Russian)
(Russian)
When X = Participle/Adjective, these are incomplete φ-probes miss the [_person]
feature and function as passive recipients of the features provided by their nominal goal (and
probe T). A default T makes no claims on the features of Part, whereas both GenP and AccP
are close(r) to Part/A on the strength of (8), providing it with a free option. Therefore Part/A
can become involved in Agree and valuation either with AccP or GenP.
In conclusion
Our account is based on multiple movements within a single Kase Projection, which
must provide both QH/Num and NP of (5) with separate cases. The account based on
articulated KP and case-driven movement does not overgenerate in that it also predicts that
optional agreement does not apply in cases of the concord holding between a verb (selected
by a φ-complete T) and a nominative subject containing a specifier in genitive. In such
complex nominal structures the genitive is a case of an extended nominal projection separate
from the projection of the nominal head książka/book. On the basis of the assumption that KP
is projected on top of each TNP, we predict the following structure:
(10) T [NomP [TNP [GenP jego] [N’ książka…]]] …
his
book
In (10) the possessive GenP is inactive for Agree for φ-features, as its derivational
cycle (phase) came to its end when N accessed the KP projection level of [TNP jego].
157
References
Babby, Leonard. (1987). Case, prequantifiers, and discontinuous agreement in Russian. NLLT 5: 91–138.
Bailyn, John. F. (2004). The case of Q. In O. Arnaudova (ed.), Proceedings of Formal Approaches to Slavic
Linguistics: The Ottawa Meeting 2003. Ann Arbor, MI: Michigan Slavic Publications. 1–36.
Bošković, Ž. (2006). Case and agreement with genitive of quantification in Russian. In C. Boeckx (ed.),
Agreement Systems. Amsterdam: John Benjamins. 99–120.
Caha, P. (2009). Nanosyntax of case. PhD dissertation. University of Tromso.
Caha, P. (2010). The parameters of case marking and spell out driven movement.
http://ling.auf.net/lingbuzz/001026.
Cinque, G. (2005). Deriving Greenberg’s Universal 20 and its exceptions. Linguistic Inquiry 36: 315–332.
Franks, S. (1994). Parametric properties of numeral phrases in Slavic. NLLT 12: 597–674.
Franks, S. (1995). Parameters of Slavic morphosyntax. New York: Oxford University Press.
Franks, S. (2002). A Jakobsonian feature based analysis of the Slavic Numeric Quantifier Genitive. Journal of
Slavic Linguistics 10: 141–181.
Pesetsky, D. and E. Torrego. (2001). T-to-C movement: Causes and consequences. In M. Kenstowicz (ed.), Ken
Hale, a Life in Language. Cambridge, MA: The MIT Press. 355–426.
Przepiórkowski, A. (2001). Predicative case agreement with Quantifier Phrases in Polish. In A. Przepiórkowski
and P. Bański (eds.), Generative Linguistics in Poland: Syntax and Morphosyntax. Warsaw: Polish
Academy of Sciences. 159–169.
Przepiórkowski, A. (2004). “O wartości przypadka podmiotów liczebnikowych”. Bulletin De La Société
Polonaise De Linguistique, fasc. LX: 133–143.
Przepiórkowski, A. and A. Patejuk. (2012). The puzzle of case agreement between numeral phrases and
predicative adjectives in Polish. In M. Butt and T. H. King (eds.), The Proceedings of the LFG'12
Conference. Stanford, CA: CSLI Publications. 490–502.
Watanabe, A. (2012). A morphological solution to agreement puzzles in Slavic. In Proceedings of the 7th
Mediterranean Morphology. 112–122.
http://www3.lingue.unibo.it/mmm2/wp-content/uploads/2013/MMM7_Cyprus_Proceedings.pdf.
Willim, E. (2000). On the grammar of Polish nominals. In R. Martin, D. Michaels and J. Uriagereka (eds.), Step
by Step. Cambridge: The MIT Press. 319–346.
Willim, E. (2014). AGREE drives syntactic movement: Evidence from predicate A(P) movement with subjects
with Genitive of Quantification in Polish. Paper presented at FDSL 10.5, Masaryk University.
Starke, M. (2009). Nanosyntax: A short primer to a new approach to language. Nordlyd 36(1): 1–6.
158
On the discourse properties of dedicated impersonal
pronouns in episodic sentences
Sarah Zobel
University of Tübingen
[email protected]
In this talk, I address the discourse properties of dedicated impersonal pronouns in the
restricted contexts of episodic sentences. These are sentences that describe a specific
situation, and do not make any general claims about the existence of a regularity. This talk
focuses on German man in episodic sentences, “existential man” (EM), which contributes a
similar meaning to that of indefinite pronouns, see (1). In contrast to the existential use,
German man, like all dedicated impersonal pronouns, has a generic use which is only
available in generic sentences (pace Malamud 2013), see (2).
(1)
Gestern hat man die Uni angezündet.
yesterday has MAN the uni set-on-fire
‘Yesterday, someone set the university on fire.’
(2)
Man muss seine Eltern respektieren.
‘One has to respect one's parents.'
(existential man ≈ someone)
(generic man = one)
Note: the existential use is not cross-linguistically universal – most notably English
one can only occur in generic sentences, see translation of (2); e.g. French on (cf. CabredoHofherr 2004), Italian si (cf. Cinque 1988), and other European languages behave like
German man.
The aim of this talk is to address one persisting point of contention for dedicated
impersonal pronouns: their (in)definiteness. They have been argued to be definite expressions
(e.g. Alonso-Ovalle 2002, Kratzer 1997), indefinite(-like) expressions (e.g. Condoravdi 1989,
Malamud 2013, Moltmann 2012), and even suggested to be “a-definite”/implicit argumentlike expressions (Koenig & Mauner 1999). Most of these investigations focus on the
pronouns’ semantic behavior, and either only discuss English one in its generic use, or mix
existential and generic uses. Since in the generic use, any contribution of the pronoun is
masked by a quantificational generic operator Gen, I argue that the existential use provides
better evidence for the (in)definiteness of dedicated impersonal pronouns. Previous results
based on their semantic behavior group dedicated impersonal pronouns with indefinites
159
since they allow for QVE with adverbs of quantification (cf. Malamud 2013 and references
therein). But: EM does not share the same scope-taking behavior with either indefinite
pronouns (Zifonun 2000) or indefinite singular DPs. Central remaining question: Is EM
indefinite? Or does it behave like implicit arguments, as argued by Koenig & Mauner? I give
an in-depth analysis of the discourse anaphoric potential (DAP) of EM which supports Koenig
& Mauner’s suspicion that in their existential use, dedicated impersonal pronouns behave like
implicit argument-like expressions (pace Malamud 2013). The analysis that I propose below
captures this behavior, as well as Zifonun’s (2000) result that EM differs substantially from
indefinite pronouns in its DAP – without running into the formal problems noted by Malamud
(2013) for Koenig & Mauner's account. Informally, EM is best conceptualized as referring to
an indeterminate “group” of individuals (possibly a singleton) for which I write “X”.
1. Preceding sentence/context: EM can occur in out-of-the-blue contexts, see (3), and
does not require its referent to be previously introduced.
(3)
Hast du das gehört? Gestern hat man die Uni angezündet.
‘Did you hear? Yesterday, X set the university on fire.’
Previously introduced (groups of) discourse referents can be understood as referents of
EM, but have to be inferred as such. In general, highly topical/salient discourse referents are
considered as referents of EM, depending on the discourse relation between the utterances
(Asher & Lascarides 2003), see (4).
(4)
Eine Gruppe von Studenten ist für ihren Vandalismus bekannt. Gestern hat man zum
Beispiel die Uni angezündet.
‘A group of students is known for their vandalism. For example, yesterday X set the
university on fire.’
(X = the group of students)
Since zum Beispiel (Engl. ‘for example’) marks an elaboration, the group of students
introduced in the first sentence is easily inferred as the referent of EM in the second sentence.
But note: the speaker in (4) does not explicitly claim that the group of students is indeed
responsible, i.e., man is not anaphoric.
2. Subsequent sentence/context: EM cannot be taken up by 3rd sg. pronouns or
arbitrary definite descriptions (see Cabredo-Hofherr 2008, Zifonun 2000). However, it can be
taken up (i) by 3rd pl. personal and demonstrative pronouns with a corporate reading (pace
Malamud 2013), (ii) bridging definites (pace Malamud 2013), and (iii) EM. Arguably, these
expressions all involve inferred referents, i.e., they are not strictly anaphoric to EM; this is
160
also supported by the different number specifications. Example (4) can be continued by (5)
(illustrates i & ii) or (6) (illustrates iii).
(5)
Sie /die
/die Brandstifter haben ein Streichholz in einen Mülleimer
geworfen.
they /d-pron /the arsonists
have a match
in a
garbage-can thrown
‘They / the arsonists threw a match in a garbage can.’
(6)
Man hat ein Streichholz in einen Mülleimer geworfen.
MAN has a match
in a
garbage-can thrown
‘X' threw a match in a garbage can.’
(X' = X of (4))
EM in a subsequent sentence does not have to co-refer with occurrences in preceding
utterances: (7) can also naturally continue (4).
(7)
Man sucht noch nach den Brandstiftern.
‘X'' is still looking for the arsonists.’
(X'' = the police)
Note: Malamud (2013) claims that EM cannot be taken up by either personal or
demonstrative pronouns or definite DPs, and concludes from this that Koenig & Mauner's
parallel between implicit arguments and dedicated impersonal pronouns cannot be confirmed.
As example (5) shows this conclusion is premature.
3. Sentence internally, EM can only co-refer with a select few elements: (i) the 3rd
sg. reflexive pronouns, (ii) PRO, and (iii) EM (Cabredo-Hofherr 2008, Zifonun 2000). It
cannot co-refer with any suppletive forms in the accusative and dative or with possessives (cf.
Kratzer 1997).
(7)
a. Man hat sich entschlossen PRO zu bleiben.
MAN has self decided
PRO to stay
‘X decided to stay.’
(only exhaustive control, cf. Landau 2010)
b. Man hat beschlossen, dass man bleibt.
MAN has decided
that MAN stays
‘X decided that X' stays.’
(X = X and X = X' possible)
As Koenig & Mauner (1999) and Landau (2010) show, the same discourse behavior, as
sketched for EM above, is also exhibited by implicit arguments, e.g. the demoted agents of
passives. This is challenged by Malamud (2013) who argues that implicit arguments do not
show QVE unlike EM. However, it is unclear at this point whether silent elements can enter
into QVE configurations at all. Further support that EM parallels implicit arguments comes
from a corpus study in which uses of man in German literary texts are compared to their
English translations. Since English one lacks an existential use, translators need to use
161
different but similar strategies: EM is translated either (i) as an implicit argument or (ii) as a
personal pronoun corresponding to the inferred set of referents.
(8)
a. Man hörte Regentropfen auf das Fensterblech aufschlagen.
‘X heard rain drops hitting the window pane.’
T : ‘Drops of rain could be heard hitting the pane.’ (passive → implicit argument)
b. Man sah auf den Vorplatz der Wohnung hinaus.
‘X saw out onto the landing of the flat.’
T : ‘He could see onto the landing.’ (‘he’ is the currently most salient male individual)
(Kafka, 1917, “Die Verwandlung” (‘The Metamorphosis’); Translation (T: David Wyllie, 2002)
Formal proposal: The central idea of my analysis is that EM is a weak nominal. It
contributes a (contextually constrained) set of animate individuals, but no distinguished
referent that can be taken up by anaphoric pronouns. This idea is similar in spirit to
Chierchia's (1995) proposal for Italian si and Koenig & Mauner's (1999) proposal for implicit
arguments. As Malamud (2013) correctly argues, however, a DRT account in the spirit of
Koenig & Mauner cannot be maintained for EM since the introduction of a discourse referent
is needed in DRT to derive QVE for dedicated impersonal pronouns. Nevertheless, a dynamic
treatment is needed to capture EM’s DAP. I adapt McNally & Van Geenhoven's (1998)
proposal for the semantics of weak nominals:
(9)
[[man]]g = {gxg' | x ∈ {y : y is an animate individual} & g = g'}
According to (9), EM contributes a variable x of type e that ranges over the set of
animate individuals, but which does not change the variable assignment to add a new referent
which could be picked up by anaphoric pronouns. The existential flavor that motivates EM’s
translation as ‘someone’ is contributed by existential closure of the property denoted by the
sentence containing EM, which is a straightforward result of EM combining with the verb.
Furthermore, I assume that reflexivization and control are not mediated by the variable
assignment (see Chierchia 1995, Landau 2010), and can hence occur with EM.
Generic man: The context-change potential proposed for EM in (9) already provides
everything needed to analyze the use of man in generic sentences. Following previous
proposals, I assume that in the generic use, the generic operator Gen binds the variable
contributed by man (cf. Zobel 2014). Similarly, the QVE cases are derived by assuming that
the adverb of quantification unselectively binds the variable (cf. Malamud 2013 and
references therein). For any potential cases of donkey-pronoun uses of man, I follow
162
Chierchia (1998), who proposes that Gen or adverbs of quantification bind the variables
contributed by both occurrences simultaneously.
References
Alonso-Ovalle, Luis. (2002). Arbitrary Pronouns Are Not That Indefinite. In C. Beyssade, R. Bok-Bennema, F.
Drijkoningen & P. Monachesi (eds.) Romance Languages and Linguistic Theory 2000. John
Benjamins.
Asher, Nicholas & Alex Lascarides. (2003). Logics of Conversation. Cambridge University Press.
Cabredo-Hofherr, Patricia. (2004). Impersonal pronouns in Somali, German and French. Handout, Syntax of the
World’s Languages 1, Leipzig.
Cabredo-Hofherr, Patricia. (2008). Les pronoms impersonnels humains - syntaxe et interprétation. Modèles
linguistiques tome XXIX-1 57: 35–56.
Chierchia, Gennaro. (1995). The Variability of Impersonal Subjects. In E. Bach, E. Jelinek, A. Kratzer &
B.H.Partee (eds.) Quantification in Natural Languages. Kluwer. 107–143.
Chierchia, Gennaro. (1998). Reference to kinds across languages. Natural Language Semantics 6: 433–466.
Cinque, Guglielmo. (1988). On si constructions an the theory of arb. Linguistic Inquiry 19: 521-582.
Condoravdi, Cleo. (1989). Indefinite and generic pronouns. In E. Jane Fee and K. Hunt (eds.) Proceedings of the
Eigth West Coast Conference on Formal Linguistics.
Koenig, Jean-Pierre & Gail Mauner. (1999). A-definites and the Discourse Status of Implicit Arguments.
Journal of Semantics 16: 207–236.
Kratzer, Angelika. (1997). German impersonal pronouns and logophoricity. Presentation at Sinn und Bedeutung
II. Berlin, Germany.
Krifka, Manfred et al. (1995). Genericity: an introduction. In G.Carlson & F.J.Pelletier (eds.) The Generic Book.
University of Chicago Press.
Landau, Idan. (2010). The explicit syntax of implicit arguments. Linguistic Inquiry 41: 357–388.
Malamud, Sophia A. (2013). (In)definiteness driven typology of arbitrary items. Lingua 126: 1–31.
McNally, Louise & Van Geenhoven. (1998). Redefining the weak/strong distinction. CSSP 1997.
Moltmann, Friederike. (2012). Two kinds of first-person-oriented content. Synthese 184: 157–177.
Zifonun, Gisela. (2000). Morphosyntax und Semantik des Pronomens "man". Deutsche Sprache 28: 232–253.
Zobel, Sarah. 2014. Impersonally Interpreted Personal Pronouns. University of Göttingen.
163