1
Indeterminate Pronouns:
The View from Japanese
*
Paper presented at the 3rd Tokyo Conference on Psycholinguistics
March 2002
Angelika Kratzer Junko Shimoyama
UNIVERSITY OF MASSACHUSETTS KYUSHU INSTITUTE OF
AMHERST TECHNOLOGY
1. Cross-linguistic Variation
The traditional view on natural language quantification is that languages
have determiner quantifiers projecting DPs that scope overtly or covertly, subject
to the usual constraints on movement. The quantification system in Japanese does
not seem to conform to this picture. In Japanese, quantifier phrases are built using
what scholars of Japanese commonly refer to as ‘indeterminate pronouns’ (Kuroda
1965):
(1) dare ‘who’ doko ‘where’
nani ‘what’ itu ‘when’
dore ‘which (one)’ naze ‘why’
dono ‘which’ (Det) doo ‘how’
*
A slightly shorter version of this paper appears in Yukio Otsu (ed.): The Proceedings of the Third Tokyo
Conference on Psycholinguistics. Tokyo (Hituzi Syobo), 2002, 1-25. We would like to thank the audiences at the
13th Amsterdam Colloquium, the Third Tokyo Conference on Psycholinguistics, UCLA, UMass Amherst, and in
particular Bernhard Schwarz for helpful comments.
2
Depending on the operator they ‘associate with’, Japanese indeterminate phrases
can take on existential, universal, interrogative, negative polarity, or free choice
interpretations. 2(a) is an example of the universal construction, and 2(b) is an
example of a wh-question.
(2) a. [[Dono hon-o yonda] kodomo] -mo yoku nemutta.
which book-ACC read child -MO well slept
‘For every book x, the child who read x slept well.’
b. Taro-wa [[dare-ga katta] mochi]-o tabemasita ka?
Taro-TOP who-NOM bought rice cake-ACC ate Q
‘Who is the x such that Taro ate rice cakes that x bought?’
A first connection between quantification in Japanese and in English was
established by Nishigauchi 1986, 1990. Nishigauchi argued that some Japanese
quantifiers are adverbial quantifiers, and that they can unselectively bind variables
made available by indeterminate phrases and bare NPs, as proposed in Heim 1982.
The semantic part of Nishigauchi’s analysis was criticized by Ohno 1989 and von
Stechow 1996. Von Stechow 1996 also attempted to provide an explicit analysis
of the syntax-semantics mapping. However, his analysis relies on assumptions for
Japanese that are idiosyncratic and ad hoc (see Shimoyama 2001), hence moves us
away from a possible explanation for how apparently different types of quantifiers
can be acquired by children.
In his typological survey of indefinite pronouns Haspelmath 1997 shows that
indeterminate pronouns in the Japanese sense constitute a unified class cross-
linguistically. An example is Latvian (Haspelmath 1997, p. 277, diacritics
omitted):
3
Interrogative
kaut-series
ne-series
jeb-series
person
kas
kaut kas, kads
ne-viens
jeb-kads
thing
kas
kaut kas
ne-kas
jeb-kas
place
kur
kaut kur
ne-kur
jeb-kur
time
kad
kaut kad
ne-kad
jeb-kad
manner
ka
kaut ka
ne-ka
determiner
kads, kurs
kaut kads
ne-kads
jeb-kads,
jeb-kurs
The Latvian ‘bare’ series has interrogatives. The kaut- series has existentials. The
ne-series appears under the direct scope of negation, and the jeb-series is found in
indirect negation contexts, in comparatives, and also with a free choice
interpretation. If indeterminate phrases form a natural class cross-linguistically, the
question arises as to what it is that makes Japanese quantifier and interrogative
constructions look so different from their counterparts in Indo-European
languages. In this paper, we will take a first step towards answering this question
by presenting an analysis of the German indeterminate pronoun/determiner
irgendein from a Japanese point of view. We will show that the Japanese
perspective allows us to explain some puzzling properties of irgendein and free
choice indefinites more generally. We will also address the differences between
Indo-European and Japanese indeterminate pronouns and speculate that they might
ultimately reduce to the presence or absence of uninterpretable features that give
Indo-European indeterminate pronouns their selective look.
2. A Hamblin Semantics for Indeterminate Pronouns in Japanese
Hamblin originally designed his alternative semantics for run-of-the mill
questions in English. As far as we are aware, the first extension to quantification
in languages other than English is Ramchand 1997. Hagstrom 1998 applied
Hamblin’s semantics to existentially quantified sentences in Japanese.
4
Shimoyama (1999, 2001) established the connection with universally quantified
sentences, and argued moreover for an in situ interpretation of indeterminate
phrases. The guiding idea behind a Hamblin semantics for indeterminate phrases
is that they introduce sets of alternatives that keep ‘expanding’ until they meet an
operator that selects them. The alternatives can be of different semantic types,
such as individuals, properties and propositions, and consequently, we expect the
existence of quantifiers that can operate over alternatives of different semantic
types. Determiner quantification falls out as a special case, the case where the
alternatives are individuals.
1
On such an analysis, the Japanese universal
quantifier mo can be analyzed as a regular generalized quantifier, and the
semantics of sentence 2(a), for example, can be sketched as in (3).
(3) All members of A slept well: A = {the child who read book a, the child who
read book b, the child who read book c, ….}
A second important consequence of a Hamblin analysis for indeterminate phrases
in Japanese is that while allowing for long-distance association between
indeterminate phrases and particles as in (2), it automatically derives the locality
conditions for this association without any stipulations. Indeterminate phrases in
Japanese must associate with the closest available operator:
1
Thanks to Akira Watanabe for bringing to our attention the fact that Japanese lacks a quantificational
particle meaning most that takes indeterminate phrases. Here is how ‘most’ is expressed in Japanese (the ‘floated’
versions are also possible). NO = pre-nominal modification marker.
(i) Hotondo-no gakusei-ga utatta.
most-NO student-NOM sang
‘Most students sang.’
Watanabe pointed out that the question of whether the above fact is an accident in Japanese, or it holds across
languages that has Japanese-type systematic indeterminate phrase quantification, should have consequences in the
validity of the claim that determiner quantification is reduced to a special case.
5
(4) * [….[…. ind …. ka/mo]……]-ka/mo
On a Hamblin analysis, the alternatives created by an indeterminate phrase can
expand across relative clause boundaries as illustrated in (2). They are, however,
caught by the first relevant operator in their way. In (4), which illustrates a typical
intervention effect, the alternatives created by the indeterminate pronoun must
associate with the lower ka/mo. It could not be otherwise. The intervention
effects follow from the very architecture of the interpretation system, in interaction
with structural configurations. No locality principles have to be stated in the
grammar.
Here is a brief illustration of how the interpretation of the simple sentence Dare(-
ga) nemutta is computed in a Hamblin semantics. We give an overview of the
essential definitions of the interpretation system in the Appendix of section 3,
which can be skipped by readers who are not interested in the technical parts of our
proposal. In a Hamblin semantics, all expressions denote sets of ‘traditional’
denotations. These are the alternatives. Most lexical items denote singleton sets.
The main innovation comes with indeterminate pronouns and phrases. Those
denote sets of individuals. We have to think of those sets as individual
alternatives, rather than as properties. This is the major conceptual shift in
Hamblin’s system. Via pointwise functional application, the alternatives created
by indeterminate phrases can ‘expand’. More formally, we have (for all possible
worlds w and variable assignments g):