The semantics of Japanese verbs by MikeJenny

VIEWS: 163 PAGES: 92

									                              The semantics of Japanese verbs.
                      Cases, semantic roles and graphic representations

                                   Dr. Simone Dalla Chiesa
                                The University of Milano, Italy
                     Department of Contemporary Languages and Cultures

General introduction
        In this series of four lectures I am going to discuss the technique I use to teach
Japanese verbs and their constructions to my Italian students. During the years I
developed a method of teaching Japanese verbs that is founded on two kinds of graphic
representations: a funny looking "Balloon Graph" and a grid, which I call "Table Graph". I
developed both graphs by myself, but the Balloon Graph comes from some intuition my
own teachers had in Japan, many years ago, and the Table Graph is actually very similar
to the theta-role grid used in Generative Grammar. I am going to explain how these graphs
are composed later. For the time being, let me just show one example of each.

        Sample of Balloon Graph


                                                                                で de            に ni
         が ga                                                                  Dynamic         Static
       Nominative                                                              locative       locative
                                                                                 PLACE         TIME
          AGENT           Accusative                 に ni
         SOURCE             PATIENT                  Dative
                           LOCATUM                   GOAL

Sample of Table Graph

Predicate pairs of       教える                    教える                 教わる・習う                教わる・習う
teaching/learning        oshieru                oshieru             osowaru, narau        osowaru, narau

    AGENT            nominative が ga       nominative が ga           agentive に ni
    SOURCE                                                                                 ablative から
    P.LOCATUM         accusative を o                                accusative を o        accusative を o
    GOAL               dative に ni
    AGENT                                                           nominative が ga       nominative が ga
    PATIENT                                 accusative を o

                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
        Both kinds of graph make use of certain terms, like AGENT, PATIENT or GOAL. They
are the labels of three of the so called semantic roles. I believe that one could grasp the
meaning of these terms with little effort, even if she has never heard of them before. But
since this is a notion that I use extensively when I teach Japanese verbs, a good
understanding of it is crucial. So, during this first introductory lecture I will discuss the
notion of semantic role in some detail. My graphs also contain words such as nominative,
accusative or dative. These are case labels, and I am absolutely positive that you, as
native speakers of the Finnish language, are very familiar with them. What you may not be
accustomed to is the use of the notion of case in teaching Japanese grammar. This is why
I believe it will be necessary to explain what cases are in my particular approach to
Japanese verbs. Therefore, during this lengthy introductory lecture I will spend some time
dealing with this notion as well.
        After this introduction, my lectures will consist in an analysis of a number of
Japanese simple clauses, grouped according to the type of predicates 1 that head them.
For each of the most important predicates or classes of predicates I am going to project a
Balloon and a Table Graphs that display together that predicate, its several phrases and
their particles. The graphs are meant to show in an easy way the relation between
semantic roles, cases and case markers. Of course I will also discuss each graph in detail.
        What is important to understand is that what I am going to do during these lectures
is not formulating an explanation of why certain verbs are accompanied by certain specific
sets of cases and particles. Rather, it is a discussion of how certain typical, recurrent
combinations do happen. The "why" is one of the most debated problem of semantics and
syntax, and even if I obviously had to deal with several linguistic theories in order to build
my teaching technique and give it consistence, I am not discussing linguistic theory here. I
will briefly deal with the most relevant theoretical issues in the Conclusion chapter, for the
few survivors that will follow me till the very end. For the time being, it should be
remembered that all the reasoning I will be doing here has a didactic purpose, and is the
very same reasoning I do in the classroom in front of my students when I teach Japanese
grammar. I will be also trying to use the same wording. You will see that this is a sort of
new, and surely peculiar way of teaching Japanese verbs – but it's a way of teaching them,
not of explaining the cognitive basis of their constructions.
        I believe these lectures will be interesting and useful for students of all levels of
Japanese language, because they will learn something totally new or re-learn something
old in a new way. But they might be useful for advanced students as well, and even for
some fellow Japanese language instructor, who will be introduced to some unusual,
alternative way of teaching Japanese verbs.

Personal history
      I am a graduate from the University of Venice, and have been teaching Japanese
language since 1987. In the past as a Contract Professor I held a number of posts at the
Universities of Pavia and Venice. Perhaps the most relevant for these lecture series are
the courses in Japanese Linguistics and in Japanese Language Teaching I taught at the
University of Venice as recently as in 2007-08. Since 2002 I have been Research
Professor at the Department of Contemporary Languages and Cultures of the University of
Milano. There I coordinate all Japanese language courses and teach five Japanese

  Throughout my lectures I will use the term verb to refer to the lexical and morphologic category of verbs,
and to verbs as the instrument used to describe events. I will also maintain the usual denomination of
transitive and intransitive verbs. I will use the term predicate to refer to the heads of clauses, verbs and
adjective alike, which assign case, have a meaning and are characterized by aspectual properties. With the
term clause I refer to a sequence of (mostly) nominal phrases headed by a verb, while with sentence I refer
to a structured set of clauses.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
grammar classes, in the undergraduate Program in Cultural and Linguistic Mediation and
in the graduate program in Languages, Cultures and International Communication.
Teaching grammar means that I do have to follow the basic topic flow of our textbook
(ICU's Modern Japanese for College Students), but that I am also free to build my own
graphs and tables, formulate general and detailed rules using the wording of my choice,
rushing ahead to meet advanced grammar topics whenever I want. I am free to do this
because the real, hard job – like drills, kaiwa, sakubun, kanji tests and so on – is done by
my wonderful Japanese colleagues.
       My interest for the topic of this lecture series started very early in my life of study. In
the middle of the '80s, just about twenty-five years ago, I was a Monbushō research
student at the University of Tsukuba. At the time the head of the Institute for Japanese
Language Teaching was professor Teramura Hideo. Teramura was a linguist very well
known and appreciated in Japan. His main research interests were, as it happens, the
analysis and classification of the case structure of predicates. He was clearly under the
influence of Fillmore's case grammar,2 but he build his own system, called Teramura
bunpō, or "Teramura grammar". I never had the chance of being taught by Teramura
himself – he moved elsewhere and then died shortly after my arrival in Tsukuba – but I
absorbed his interests, methods and general approach through my own teachers, who had
been pupils of Teramura themselves.

Teramura Grammar
       What Teramura was saying about the relation between a predicate and its case
structure was something like this.
          Painkiller: the Monster

    Filmore 1968.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
       This is a screenshot taken from a game named Painkiller. Basically, what you have
to do in this game is getting around killing monsters. This is one of them. It's a swamp
monster made of muck. The monster is walking around you in the slush, during very little
to harm you.

              Painkiller: the Claw

       Occasionally, however, the monster would stop and stick its arms in the mud. Its
claws would just disappear, then they would surface again, beyond your back, and try to
snatch you.
       What this monster is doing is exactly what predicates are doing according to
Teramura grammar. A predicate would bury part of itself beneath the visible surface of a
clause, and have these parts re-appear somewhere else in the form of case particles.
Thus, case markers are actually the "fingers" of a predicate that re-surface in the clause.
But there is no visible connection between them and the predicate, and their shapes are
completely different from that of the predicate, so that they are not very easy to relate to
the body of their far-away parent entity.
       Such relation may be represented by this graph.

      Painkiller: the Idea

           が                    を                     に                   おしえる

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
        This simple graph displays predicate oshieru "teach" and the three case particles it
assigns. The predicate is shown in a box at the end of the clause, but its fingers, or tendrils,
or they could be called threads of mycelium, extend underneath the ground till, like
mushrooms, they pop up again in the visible surface structure of the clause as
postpositional particles. When they surface in this way they are represented by those
round balloons. Therefore, postpositions seem to be independent units of the clause, but
actually they are not. Rather, they're part of the predicate's body, they are its long distance
manifestation within the landscape of the clause. Let me use a last metaphor: they are the
epiphany of the verb. And this is true both when case is expressed by a particle attached
to an unchanging noun (as it is in Japanese) and when is expressed my a morpheme that
is part of the noun itself (as it happens in Finnish or Latin).

        This strongly predicate-centered approach of Teramura was ultimately intended to
serve as a way of cataloguing predicates according to their characteristic sets of
accompanying particles. In this it was not very different from Fillmore's case grammar.
Teramura wrote some excellent book using this approach, including a couple of textbooks
for intermediate students which are impressive in their consistency and crisp logic.3

The semantics of particles
        Teramura was saying that predicate and particles are one and the same thing.
However, he was saying very little about why a given predicate always occurs in
conjunction with certain particular set or sets of case particles. What makes a predicate
class to take certain particles instead of others? If we want to discuss this question we
should look for help somewhere else. For instance, one can motivate the use of certain
particles instead than others by means of their meaning. It is the so-called semantic or
cognitive approach to Japanese particles. This approach says that the use of all Japanese
particles can be analysed in terms of meaning. I am sure you have already had some
encounter with an analysis of this kind, because it looks like we teachers of Japanese love
it very much. Let me formulate just one example, that of the particle ni. It is time-
consuming but interesting.4 The basic idea is that ni functions as the marker of a place in
which no movement, and consequently no action, happens.

          Niwa-ni ki-ga aru.
          Garden-LOC tree-NOM be
          There are trees in the garden.

          Niwa-ni kodomo-ga iru.
          Garden-LOC children-NOM be
          There are children in the garden.

          Nanoka-ni kaeru.
          7th-LOC return
          I am going back on the7th.

          (1)     NI locative                                  │

    Teramura 1972, 1982, 1988.
    See another way of discussing the semantics and cognitive content of ni in Sugai 2007.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
         These are examples of place and time ni locatives. No action happens in the garden
– if it were, the particle would be de. Some action does take place on the 7th, however.
But the event kaeru is instantaneous, which means it happens in a point of time placed
within the day 7th. This point has no dimension and cannot contain action, because
actions last in time and are supposed to spread, or move, along the timeline, taking
segments of it. Using an expression taken from physics, you can tell that ni marks a point
in spacetime. In the simple graph at the bottom, I represented such a static point as a
"wall" instead of a dot.

      Haha-ni tegami-o okuru.
      mother-TRA letter-ACC send
      I will send a letter to my mother.

      Kodomo-ni omocha-o ageru.
      children-DAT toy-ACC give
      I give a toy to a child.

      Itaria-ni kaeru.
      Italy-TRA return
      I am going back to Italy.

      Tomodachi-ni au.
      friend-TRA meet
      I am meeting a friend.

      Haru-ni nat-ta.
      Spring-TRA become-PAST
      Spring has come.

      (2)    NI dative/translative                     →│

       Again, ni marks a static point, that where some movement ends. Some movement
or action did happen before the event, and there might be some afterward, but the end of a
movement is the absence of movement. Hence the use of ni to mark the endpoint. This
analysis applies to both transitive and intransitive verbs, and works when the endpoint is
an animate being, a place, or an instant along the timeline reached by the flow of "reality".
This is why I represent this point as a "wall" reached by an arrow.
       However, there are instances in which the particle ni seems to have exactly the
opposite meaning. This happens, for example, when it marks the source of a favor with the
predicates of receiving.

      Jon-ga tomodachi-ni nihongo-o narat-ta.
      John-NOM friend-AGE Japanese-ACC learn-PAST

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
      John was taught Japanese by a friend.

             (3)    NI agentive                                 ←│

        All considered, this use of ni can be said to follow the same logic observed in the
other examples: ni marks the unmoving starting point of a unidirectional movement, the
source of both a favor and the transfer of an object.
        This requires to reformulate the semantics of ni as follows: Ni marks a spot in which
no movement takes place, but that may serve as the reference point (the starting or ending
point) for a single arrow (better: vector) of movement. This reformulation is necessary also
in consideration of the contrasting use of particle to, as in the instances that follow.

      Keiko-ga Masao-to kekkonsuru.
      Keiko-NOM Masao-REC marry
      Keiko will marry Masao.

      Keiko-ga Masao-to rikonsuru.
      Keiko-NOM Masao-REC divorce
      Keiko will divorce from Masao.

      (4)    TO reciprocal           →│← or ←│→

       Just like ni, to can mark either the starting or the ending point of a movement. This
movement can be either converging (as in a wedding, which entails a union) or diverging
(as in a divorce, which entails a separation). Hence, the semantics of to may be
conversely stated as follows: To marks a spot that serves as the reference point for two
opposing vectors of movement (converging or diverging as they may be).

        By the way, this is the very explanation we were given as students of Tsukuba
University by instructors under the influence of the Teramura model, even if Teramura
grammar is an approach focussed on the predicate rather than on the semantics of
particles. But this approach is surely consistent with that of Teramura's. And indeed this
way of explaining particles is at the same time fascinating and quite effective. However, it
means teaching the function and semantics of Japanese particles, instead than those of
verbs. Would it be possible teaching the case structure of Japanese using only such
approach? I believe this approach should at least be supported by a more detailed form of
classification of particles.

Case markers and postpositions
        In fact, those accustomed to some Japanese linguistics – for example because they
have read the textbook An Introduction to Japanese Linguistics by Tsujimura Natsuko
(1996) – may have already noticed that looking for a meaning of all particles is not
generally accepted by present linguistic theory, at least not for Japanese. Japanese
linguistics (that branch of general linguistics that deals with Japanese language, not the
traditional study of Japanese language undertaken by Japanese scholars) distinguishes
between case particles and postpositions.

Case particles                                       Cases
が ga                                                 nominative, accusative
                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
を (w)o                                               accusative
に ni                                                 dative
の no                                                 genitive
は wa                                                 (topic)

に ni                                                 locative (static), translative, agentive
で de                                                 locative(dynamic), instrumental
へ (h)e                                               translative
から kara, より yori                                     ablative (elative)
まで made                                              allative
をo                                                   accusative (of path)
と to                                                 reciprocal

       Case particles are few, just nominative ga, accusative o , dative ni and genitive no.
Postpositions are many more, and include locative, agentive and translative ni, locative
and instrumental de, other path-marking particles such as e, kara, made, o, and so on.
According to this distinction, case markers are supposed to be assigned by the predicate –
they are, exactly as I told you, a kind of long distance morphological extension of the
verb's body. They have no meaning by themselves; the meaning taken by their whole
phrases (called noun phrases, or NPs) comes directly from the verb, it is determined by it.
On the contrary, postposition do have a meaning and can of course be analysed in
semantic terms just as I have done above. As such, they are the elements that provide
with meaning and control the phrases where they occur (and which they "head"). These
phrases are indeed called "postpositional phrases" (or PP).

      Tree of Japanese NP and PP

                NP                                                  PP

                 N                                          NP                P

            kuruma-ga                                    kuruma               de

      Diagrams of this kind are now generally replaced by the more powerful X' trees. I
added this old fashioned tree diagram just to show the difference between PP and NP. I
am not commenting on it.
      How the distinction between case particles and postpositions is made? Mostly in
terms of the syntactic distribution of quantifiers (QUANT). With case-marked phrases, the
QUANT can follow the case particle, as in

      Kuruma-ga nidai aru.
      car-NOM two(QUANT) be

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
       There are two cars.

        In other words, the NP is unbreakable, and QUANTs must be placed before or after
it. With postpositional phrases, however, the QUANT just cannot be placed after the
postposition. We cannot say, for example

       *Kuruma-de nidai ik-ō.
       car-INS two(QUANT) go-VOLITIONAL
       Let's go in two cars.

        but we have to place the QUANT before the PP as an genitivized adjective, saying
nidai-no kuruma-de, or in-between noun and postposition, saying kuruma-nidai-de.
        Useless to say, there are several inconsistencies with this syntactic model. For one
thing, markers no and wa syntactically behave as case particles, but they are definitely not
assigned by the predicate. Wa is assigned by the speaker for discoursive or pragmatic
reasons, whereas no has the very superficial function of signalling a relationship between
two nouns, not between a noun phrase and the predicate. Moreover, there is a number of
situations in which any basic, intuitive knowledge of the language goes against the
prescriptions of this model. To say Kodomo-ni omocha-o ageru or Kodomo-ni omocha-o
okuru is saying very much the same thing, at least for as much as the ni phrase is
concerned. But in the former clause the ni phrase is supposed to be a NP, in the latter a
PP. Also: Why should not be possible to discuss the meaning of case particles as well?
Does not the ni used with ageru have a meaning just as that used with okuru? And why
not discussing the meaning of the accusative particle o as well? We could say, for
example, that "o indicates that the noun to which is attached refers to a thing changed or
moved (in a word, affected) by some external agent". Finally, certain distributional
properties of postposition can be observed with case markers as well. For instance, one
can surely say

       Tefuda-ichimai-o suteru.
       Hand-one(QUANT)-ACC throw away
       Discard one card from your hand.

      (this example is taken from the card game Yu-gi-oh), but the placing of the QUANT
ichimai in that position is not supposed to be allowed if o is a case marker.

        So we are left with the doubt: teaching the grammar of predicates using Teramura's
method or the approach of linguistics, more refined and consolidated, but complex and
somehow unsuitable for the classroom? It should be understood that these are not the
only two possible approaches. There are at least three other approaches, which actually
overlap to a certain extent. I want to put it clearly now that none of them is completely
satisfying either – that's why in the end I had to discard all of them and resort to the notion
of semantic roles. But let me quickly deal with each of them.

       The first approach is based on the notion of verb valence or "subcategorization".
The valence of a predicate is the minimum number of phrases that must occur in its clause
in order to thoroughly describe the event. These necessary phrases are called arguments
or terms. There are predicates that require only one argument, and are called monovalent;
other predicates that requires two arguments, and are bivalent, and finally predicates that
require three arguments, and are therefore trivalent. While there are certain predicates that

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
may require no argument at all, no predicate exists that require four arguments, in any
language of the word. This is surely because for the human mind the maximum number of
entities that may be significantly involved in an event is three. Of course many more
entities can be mentioned in a clause, but these entities are not crucially involved in the
event as "true participants". Rather, they're mere "background". These entities are
represented in a clause as non-argument phrases, or Adjuncts.

0      fubuku                                no argument
1      asobu, aku:                           NP (subject)
2      taberu, tsukuru, miru, aisuru         NP (subject)           NP (object)
3      ageru                                 NP (subject)           NP (object)    NP (indir. object)
3      oku, okuru                            NP (subject)           NP (object)    PP (indir. object)

       For example, let me first mention one of the very rare Japanese predicates that
have a valence of zero. I say "very rare", but it looks like they are only two, or possibly just
this one.5 It is verb fubuku 吹雪く, which appears to be a mountaineering term derived
from the noun fubuki, "snowstorm". Other languages are richer than Japanese in zero
valence predicates. Indeed, they usually refer to weather phenomena. For instance, the
English predicates rain, snow, et cetera, have a valence of zero, as do their Italian
equivalent piovere, nevicare, and so on. You can tell me for Finnish.

       A monovalent or "one-place" Japanese predicate is asobu, in that requires only one
argument to be expressed. This is the argument referring to the entity performing the
action. Also a verb of spontaneous happening such as aku is monovalent. Its single
argument refers to the entity that spontaneously undergoes a change. In one word, we are
dealing where with intransitive verbs. When only one argument is present, as in such
instances, it is the subject.

        Two argument predicates are the basic transitive verbs. Their two arguments refer
respectively to an entity performing an action and to a second entity that is affected or
controlled by the first one. There is plenty of examples in Japanese. Above I quoted taberu
(eat), tsukuru (build, make), miru (watch) and aisuru (love) just to mention predicates that
exhibit different degrees of transitivity. Actually, the transitive is the most diffused type of
verb in any language, because the transitive event it describes is considered to be the
most basic type of event the human mind can represent. With these "two-place" predicates,
as bivalent predicates are also called, the argument referring to the initiator or agent is the
subject; the second argument is the direct object.

        Three argument predicates describe events in which one entity initiates an action
that affects a second entity by displacing it in space. A third entity is also involved, that
receiving the displaced thing. The large class of trivalent predicates includes the so-called
"ditransitive" verbs of giving (like ageru) and the verbs of putting (like oku), but also those
sister predicates that describe the same event "from opposite points of view". When a
couple of verbs does that, they are called conversive verbs. In this instance, the verbs of
giving are conversive with respect to the verbs of or receiving (like morau). With three-
place transitive predicates of this kind, the argument denoting the agent initiating the event

  The second one might be shigureru 時雨れる "rainstorm", but one can say tenki-ga shigureru, realizing as
the subject the one argument "weather", so it is probably better to consider shigureru to be an ordinary
monovalent predicate.
                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 10 -
is the subject; the second most relevant argument – the one denoting the affected entity –
is the direct object, and the third one is the indirect object.

        You may have noticed that I never mentioned anything resembling cases:
nominative, accusative, dative or whatever. Rather, the subcategorization scheme refer to
the category of Subject, Direct Object and Indirect Object. These are the so called
grammatical relations, which represents notions more basic and universal than cases.
While grammatical relations are present in all languages with any type of predicate, cases
widely changes across languages. For example, even if the basic idea is the same, the
case system of Finnish is quite different from that, much simpler, of Ancient Greek; which
means that the number, label and uses of morphological cases match only slightly in these
two languages. And there are languages that use completely different case patterns, like
the ergative languages. In an ergative language, the ergative and absolutive cases are
used in the place of nominative and accusative, and with a different distributional pattern.
Indeed, how morphosyntactic cases are used to mark, or "realize", the grammatical
relations of subject and object is one of the most studied topic of contemporary syntax and
semantics. There is a large number of approaches, each proposing different mapping
principles and marking rules. I will not deal with any of them here, because this topic is too
vast and is of little value when teaching grammar.
        What I want to do, instead, is briefly discussing why I believe this subcategorization
approach is not suitable as it is to explain Japanese verbs to students. First of all there is
the problem of how to classify the verb au "meet". It is a two-place, bivalent predicate, and
therefore (according to a "geometrical" notion of transitivity), should be placed in the
transitive verbs class, but its ni phrase cannot be considered to be a direct object.
Secondly, how to classify motion verbs? With motion verbs, such as iku, one entity moves,
and that's for sure the subject, because without it, no motion event would ever take place.
But are therefore motion verbs as intransitive as, say, asobu, shinu or aku? And also: The
clause of iku usually contains a reference to the path of movement, which can be realized
by a set of three phrases denoting the point of departure (kara, yori), the destination (ni, e,
made) and the path proper (by the way, marked by an accusative o). Now, is expressing a
path necessary? If so, the path would count as one argument, and iku would be bivalent,
just like the transitives. But does this path really count as just one argument? Should not it
rather count as two arguments at least, if one considers the ablative and translative
phrases to be two separate arguments? And how to consider the kara-made phrases?
When the are coupled, are they to be counted as one or two phrases?
        Another weak point of the valence approach is the inconsistence between the
classification of phrases in NP and PP and the type of phrases to be counted for valence.
One would expect that the arguments would all be realized as NPs, since NPs receive
their meaning and function directly from the verb and can therefore be used to define
semantic verb classes. The meaning of a PP, on the other hand, is independent from that
of the verb, because it comes from the particle, or from the event itself. Accordingly, all
PPs should be non-argument adjuncts. This is consistent with the "background" status of
adjuncts. Adjuncts specify place, time, instrument, and all the other conditions that set the
stage for the action, but do not specify the action in itself. Indeed, their type and marking is
the same across most kind of clauses. For instance, in Japanese the place of action is
realized by a de locative with most types of predicates. But certain classes of predicates
have their valence defined by a PP, not only by NPs. Such is the case of the verbs of
putting of the oku-type, which subcategorize as trivalent predicates but whose indirect
object, expressing the endpoint, is an adjunct PP. And how to consider the agentive ni
phrase of morau (receive)-type verbs?

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
                                                     - 11 -
The Japanese classification of particles
       Perhaps some help in defining the boundaries between classes of phrases on the
basis of their markers might come from the traditional classification of particles build by the
Japanese grammatists themselves. "Particles" (or joshi) form a very broad category in
Japanese grammar. Those joshi that are relevant to us are those assigned to nouns,
belonging to the two classes of kakujoshi and fukujoshi or kakarijoshi.

Kakujoshi (case particles) Syntagmatic                  Cases
が ga                                                    nominative
を (w)o                                                  accusative
の no                                                    genitive
に ni                                                    locative, dative, translative
へ (h)e                                                  translative
と to                                                    reciprocal, comitative
から kara                                                 ablative
より yori                                                 ablative
で de                                                    locative, instrumental, essive

        The group of kakujoshi deals with cases, since kaku means exactly that. A list of
kakujoshi would include ga, o, no, ni, e, to, kara, yori and de.
        All the particles of this list mark cases (as they are indeed supposed to do…). The
list therefore entails no distributional or semantic distinction between "case markers" and
"postpositions", and has no bearing on the criteria of subcategorization. Moreover, cases
themselves are not labelled by any other mean than their particle. Here I do label them,
but just to show that several particles can mark more than one case, as it is with ni
(marking dative, translative, locative and agentive) or de (instrumental, locative, essive). (I
am going to explain later the use I do of the term translative). In short, the kakujoshi list is
not a list of cases but of markers. There is nothing to prevent analyzing kakujoshi on the
basis of their meaning.

                                                            syntagmatic axis

                                              kakujoshi                      x

                 fukujoshi, kakarijoshi

         y         paradigmatic axis

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 12 -
       It should also be noticed that all the particles of this list but no relate a predicate
with the nouns occurring in its phrase. Indeed kakujoshi are defined as being
"syntagmatic". This definition has two meanings. First, it says that kakujoshi have the
function of connecting a noun to the predicate or to another noun to form a string. Second,
it says that kakujoshi are laid on the horizontal axis of a clause without disturbing each
other or competing for the same slot. Therefore, they can theoretically all appear together
(cooccur). This means that, with some effort, one can build a clause with one occurrence
of each single particle, such as:

       Ani-ga haha-ni shorui-o shashin-to(isshoni) takkyūbin-de Amerika-kara Aomori-no
       ie-e okut-ta.
       my elder brother-NOM mother-DAT documents-ACC photograph-COM courier-INS
       America-ABL Aomori-GEN house-TRA send-PAST
       My elder brother sent the documents from the States by courier to my mother at the
       house in Aomori along with some pictures.

        The clause above does indeed contain all the case particles but one, the to
reciprocal case marker. This lack is due to the fact that the reciprocal case marker has a
restricted function, which I am going to discuss later, and is assigned only by the particular
class of symmetric predicates. In the event of a non-symmetric predicate, a similar function
is taken by the comitative case marker to or toisshoni.

Fukujoshi and kakarijoshi
        The term fuku comes from fukushi, which means "adverb", so that the particles of
this class are also referred to as "adverbial particles". This is consistent with the fact that
quite often a group noun+fukujoshi represents an adverbial phrase, which can be omitted
without disrupting the grammaticality of a clause. Kakarijoshi means something like
"connecting particle", in the sense that these particles put their phrases in relation with
distant phrases in other clauses. The boundary and functions of these two particle classes
are somehow blurred, and Japanese grammatists do not agree on how to group them.
Therefore I will treat them as one single class and call them collectively fukujoshi.
Fukujoshi include wa, mo, koso, hodo, nado, bakari, dake, nomi, kiri, made, shika, sae,
and demo.

  Fukujoshi (kakarijoshi) Paradigmatic
   Adverbial or connecting particles
                 は wa
                 も mo
               こそ koso
               ほど hodo
              ばかり bakari
               など nado
               だけ dake
               のみ nomi
                きり kiri
               しか shika
                さえ sae
               でも demo

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
                                                    - 13 -
       The basic characteristic exhibited by the particles of this class is that they are
"paradigmatic" in nature. In the graph above, they would be stacked along the vertical y
axis. This means that fukujoshi can be added into the same slot already occupied by a
kakujoshi, enriching the meaning of that phrase and of the clause as a whole without
changing the basic skeleton of its structure. In this operation, fukujoshi are either simply
added to the existing kakujoshi, or they substitute and "obscure" them. This happens in the
spoken language and only with ga and o. (Also no behaves in a peculiar way in
combination with a fukujoshi, but it is a peculiar kakujoshi anyway).6
       Fukujoshi have a strong meaning, much more definite than that of kakujoshi. Just
one example. Particle nomi "excludes other possibilities and establishes a limit".7 By
attaching it to a noun the speaker implies that only the specific entity marked by nomi can
take part in the event token she is relating. In short, nomi means "only". It should be
noticed that the new meaning brought about by a fukujoshi is not determined by the verb
and has nothing to do with an objective description of an event such as that provided by
the verb. Rather, this meaning comes from the speaker, who is the only one to decide
weather to use a fukujoshi or not according to her own pragmatic or discoursive
considerations. In other words, fukujoshi are used by the speaker for emphasis, when she
wants to underscore certain subjective facets of the event she is relating.

       This classification of particles is very interesting, but still does not say very much
about the special relation of a predicate with its particles, and why only certain particles
are selected by a given predicate. It should be clear at this point that we should look in a
completely different direction.

Abstract, grammatical cases and concrete, local cases.
       The direction I find more promising is that of substituting a particle-based approach
with an approach anchored onto some deeper (I use a dangerous word here) relation
between a predicate and the nouns that compose its clause. This relation is expressed by
the notion of case. There are several theories of case and entire books devoted to them,
but this is not the place to give a short outline of even the most important of them. Let me
only put it clear that I use here a definition of case quite different from that of deep case of
Fillmore and from the case of Generative Grammar. For Fillmore, case is actually closer to
the semantic role that I am going to deal with in a short time, whereas Generative
Grammar says that only the valence arguments have "case". Rather, what I do here refer
to as "case" is surface case, which I define as

      Case – the grammatical encoding of the relation of meaning between a verb and the
nouns in its clause.

      I use this rather "traditional" definition of case because I need an approach that
sees cases as "handy tool[s] marking semantic relationships between nouns and verbs",8
no matter the degree of necessity that binds a given noun to the verb and the event the
verb describes. This is to say that I want all phrases to bear a logical relation with the

  The kakujoshi obscured by fukujoshi are the ones marking the two "direct" cases of nominative and
accusative. Taking into consideration the syntactic behavior of fukujoshi would therefore allow to identify an
important distributional difference among kakujoshi themselves. Alas, the distinction stops here, does not
extend to ni, and therefore does not permit to isolate a class of kakujoshi corresponding to the markers of the
three arguments (grammatical relations) subject, direct object and indirect object.
  Kawashima 1996: 166.
  Butt 2006: 4; see also Blake 2001: 1.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 14 -
predicate heading a clause, without drawing any distinction between argument and non-
argument phrases, or between "noun phrases" and "postpositional phrases". In other
words, since all the groups in a clause are bound to the predicate and express an
extension of its meaning, all have case – arguments and adjuncts alike.
        That all phrases have case is a notion that is consistent with a very basic linguistic
distinction drawn in traditional grammar between two classes of cases: grammatical or
abstract cases and local or concrete cases. 9 This distinction has no direct bearing on my
treatment of Japanese grammar, but I want to mention it here because in the Finnish
grammar a clear distinction is drawn between these two classes of cases, and in several
instances I did borrow from the terminology of Finnish.
        According to this classification, cases themselves or their use (or "function") can be
divided in two groups. The first group includes those cases that express a relation between
phrase and predicate that is necessary for the verb to thoroughly express its meaning. The
action described by oshieru, for example, entails the involvement of three entities to be
completed: an animate giver, an animate receiver or "givee", and an inanimate transferred
thing, the "given". Predicate oshieru expresses itself by means of three distinct particles ga,
ni and o, selected because their meanings are consistent with that of oshieru itself. We
indicate the mode in which predicate oshieru expresses its meaning through these
particles by labelling their instances with the case names of nominative, dative and
accusative. The important function these three cases have for a complete expression of
oshieru is the raeson for calling them grammatical. It should be noticed also that
grammatical cases express a kind of abstract participation to the event, quite different from
concrete, contingent, "background" specifications such as those of "when", "where" or
"how". This is the reason why grammatical cases are also labelled abstract.
        On the other hand, to the group of "local cases" belong those cases whose function
is not bound to the expression of certain particular predicate classes, but can occur in the
description of nearly any type of event. These are the "background", concrete cases that
indicate the placement of events in time and space, the direction of movement, the
instruments used and so on – all specific, limited functions quite disconnected from the
abstract notions of movement and action in general. Beside its grammatical cases, oshieru
could, for example, use locative phrases such as kyōshitsu-de ("in the classroom") or
kayōbi-ni ("on Tuesday"), or instrumental phrases such as tēpurekōdā-de ("by means of a
tape-recorder"). These are indeed the local or concrete functions of cases. The Finnish
language split these functions into the two more detailed classes of Locative and Marginal

      To end up this section, let me just stress again an important point: the separation
between grammatical and local cases does indeed reflect that between arguments and
adjuncts. But it also entails that both kinds of phrases – argument and adjuncts alike –
have cases, for all the phrases in a clause are tools of the predicate.

Semantic roles
       At this point, however, we are left with one last, troublesome question: where does
the meaning of the verb comes from? From what source the verb draws the strength to
incarnate itself within the surface structure of a clause? In order to answer these questions
and give my students a consistent and complete heuristic frame, I decided to use a well
known notion of linguistics, that of semantic role. Semantic roles, or thematic relations (not
to be confounded with the theta roles of Generative Grammar) are originally defined as:

    Lyons 1968: 294-302.
                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
                                                    - 15 -
       "[a] predetermined set of labels that identify arguments according to the semantic
relation they bear to their verb; each verb is associated with a relevant list of semantic

       This definition is still very much predicate-centered, and I cannot see a meaningful
difference between such roles and cases as I defined them. However, in the vast
landscape of linguistic notions concerning the relation of meaning between phrases and
verbs, there is another definition of semantic roles that better suits my needs, and which
opens a door to a wholly new field of reasoning.

        "[A] semantic role is the actual role a participant plays in some real or imagined
situation, apart from the linguistic encoding of those situations."11 (my Italics)

       Now this definition is very peculiar, because when it says apart from the linguistic
encoding of those situations it brakes the connection between roles and predicates, and
indirectly introduces the important concept of event. Drawing from this notion I build my
own working definition of semantic roles.

       "Semantic roles are descriptions of the distinct modes of participation in an event
exhibited by the involved entities."

      In order to make clear what roles are, here I go, at long last, with my own inventory
of semantic roles.
        Inventory of Semantic Roles

AGENT             An animate entity voltionally initiating an event. I distinguish between
                  animate and inanimate, non-volitional initiators. To the latter I assign the
                  label FORCE.
BENEFICIARY       An entity advantaged or disadvantaged by an event. So labelled to
                  distinguish it from the benefactive case. When the BENEFICIARY and the GOAL
                  roles are borne by a same entity, its NP is marked by a dative and the two
                  roles cannot be syntactically or semantically distinguished. A distinct
                  benefactive marking appears only when the BENEFACTIVE and GOAL roles are
                  borne by different entities.
EXPERIENCER       An animate entity represented as undergoing an experience of perception or
FORCE             An inanimate entity initiating an event. An entity is to be considered to bear
                  the role of FORCE when initiates an event as a mechanical or natural cause.
                  If causing an event of perception can be distinguished as STIMULUS.
GOAL              The physical or abstract location where a PATIENT LOCATUM ends its
                  movement or where is directed. Following a weak form of Localist Approach,
                  I usually consider the role of GOAL to also encompass the final state of a
                  change-of-state event (FACTITIVE role) and the "animate goal in an event of
                  physical or abstract transfer of possession" (RECIPIENT), also defined as a
                  consciously affected goal (DATIVE).
INSTRUMENT        An entity used by an AGENT to carry out an action. An effector or immediate
MANNER            The condition under which the process of a dynamic event is carried out or a

   Levin and Rappaport Hovav 2005: 35.
   Payne 1997: 47.
   Strictly speaking, animate entities always bear an EXPERIENCER role, no matter the event type they are
involved in. But this role is relevant only when their perception or feeling is described. This does not mean
that the experiencer role is not event dependent (see the Conclusion), though. The role is always "there", but
has syntactic expression only with certain classes of predicates.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 16 -
                stative event takes place.
PATH            The physical or abstract space through which a PATIENT LOCATUM moves.
                This is the standard definition; sometimes all spatial dynamic roles such as
                SOURCE, GOAL and path proper could be usefully conflated into a generalized
                PATH role.
PATIENT         An entity undergoing a change of physical or abstract state. I consider to be
                a PATIENT the entity representing the resultative object of accomplishment
                verbs of creation, sometimes associated with the role of RESULT.
PATIENT         An entity undergoing a change of place. I keep this role distinct from the
LOCATUM         more general PATIENT role in order to distinguish between the two different
                kinds of affectedness of PATIENTS in a two-patient event, in which either
                PATIENT is a P.LOCATUM.
PLACE           The physical location where an event takes place. Often LOCATION; here
                PLACE to distinguish it from the grammatical case.
SOURCE          The physical or abstract location where a P.LOCATUM initiates its movement
                from. It includes the giver of dative verbs.
THEME           An entity located in a place or condition but not acting or undergoing a
                change of state. This narrow definition is only possible because I gave a
                detailed definition of AGENT, EXPERIENCER and PATIENT.
TIME            The time location at which an event takes place.

       I composed this list by eliminating many of those too fine-grained or unnecessary
roles that usually find a place into more detailed treatments of this topic. (I am still
mentioning some of them within the table). All the roles I will use in my graphs and
throughout these lectures are listed here.
       In order to start discussing these definition in some detail I want to provide a
preliminary insight of how my graphs look like and of how they make use of semantic roles
labels. I am discussing the "Balloon Graph" of oshieru, a verb I have already mentioned
several times and which I am going to use again and again.

        Maishū kayōbi-ni Keiko-ga machi-no bunkakaikan-de kodomo-ni eigo-o oshie-teiru.
        every week(ADV) Tuesday-LOC Keiko-NOM town-GEN culture centre -LOC child-
        DAT English-ACC teach-EXTENSIVE
        Every Tuesday Keiko teaches English to some children at the town's Culture Center.

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
                                                      - 17 -
       Complete Balloon Graph of dative predicate oshieru, with case, role names and adjuncts


                                                                            で de       に ni
        が ga                                                               Dynamic     Static
      Nominative                                                           Locative   Locative
                                                                             PLACE     TIME
         AGENT          Accusative               に ni
        SOURCE          P.LOCATUM                Dative

         The following is a complete discussion of the semantics of oshieru and of how its
meaning conditions the case structure of its clause. It is a far more detailed and time
consuming treatment than that I usually give during classes, and than that I will be doing
here during the next lectures. I am providing now such a full treatment of oshieru because
it is the first time I deal with a predicate in some detail and I want to specify all the different
rings that compose my chain of reasoning. I will also repeat some of the notions I have
already briefly discussed.

        Verb oshieru describes an event in which three entities are involved. As I have
already told, these entities are (in simple terms) an acting giver, a displaced or "given"
entity, and a third, receiving entity. In order to be more specific, one can describe the roles
borne by these entity as follows, using the characters mentioned in the example. The first
entity, Keiko, is the initiator of the event, and without her no such an event would happen.
Saying that this entity acts means, first of all, that she does what it does volitionally and in
control. In other words, this entity is animate and gifted with consciousness. What Keiko
does is taking one thing that she possesses and transferring it to some other entity. In the
event described by oshieru, the transferred entity is not a physical thing, but a complex
and organized set of notions – like the rules, lexicon and so on that form the English
language. But the nature of the transferred thing does not really matter. The fact that Keiko
does not deprive herself of this transferred thing is also interesting but not relevant here.
Important but, again, irrelevant, is the aspectual question whether the transferred
information is a close package (and can be therefore received entirely and learned in full)
or rather an open quantity. What matters us is that, since this flux of information proceeds
from Keiko, one can say that Keiko is also the point of origin of the transferred "thing".
Again, another fascinating but, alas, irrelevant point is whether this dealer of information
was also the original creator of the information itself. It should be pointed out that in order
to be successfully transmitted and received by some other entity, the transferred
information must be efficiently ordered and packaged. Doing so is surely an important part
of the action undertaken by Keiko, and entails a change in the shape and inner

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 18 -
organization of the information. But the most important change affecting the object of
transfer is of course the fact of being displaced in itself. Along with effecting things (which
is "putting them into being") and affecting things by means of destroying them or changing
their shape (which one can see as a way of abstractly displacing them in time), displacing
things in space is what characterizes the most basic type of event, the prototypical
transitive event, in which a force is applied by an acting, primary entity to a secondary
entity. Whenever the entity acted upon by the initiator of the event is displaced in space,
there must a third entity or place (I treat places as entities, in my approach) where it ends
its movement. In the example above, this entity is represented by the kids kodomo. Verb
oshieru implies several things about this third, receiving entity. The first obvious thing is
that this third entity is the final destination of the transfer, the location where the
transferred thing ends its run. Again whether the movement will start again in the future,
and the transferred entity will move further on, is not relevant here. What might be
interesting to notice is only that if it will, it will do it by the force imparted by its new owner,
the entity that is receiving it in the event presently described by oshieru. But that will be a
new event, described anew by another (or possibly the same) verb. There is no language
in the world that cares about ways of describing four-entities events in which the force
imparted by an initiator moves a displaced thing to a receiver via a fourth, intermediate
entity. Simply enough, the human mind gives no cognitive salience to this complex type of
event. The second important thing concerning the receiving entity is that it must be
animate and conscious. The verb oshieru strongly requires that. Indeed, oshieru does not
describe a transfer of a physical stuff and neither of electric impulses or waves that can be
encoded as bits and received mechanically or automatically by an electronic machine.
Such a kind of transfer would be described by a mere verb of sending or of putting. Rather,
oshieru requires at the endpoint another animate entity that would recognize and treat this
abstract knowledge as such. This fact connects in turn with two, very important facts.
        The first fact is that the receiving entity must receive the information volitionally, and
must absorb, re-organize and store it by its own accord. In other words, even if it is not the
entity that initiates the event, the kodomo entity must show an active participation in the
event "oshieru" in order for it to reach a successful conclusion. This point is crucial for my
reasoning. This is exactly the scenario we have here in this classroom right now. I am
sending you a lot of information, some new for you, some old and only packaged in a new
way, but we cannot say that I have taught you anything if you do not act willingly and learn.
From your side, learning requires several acts of will: coming here in the first place, paying
attention, reasoning about what I am saying, judging my models and possibly applying
them yourselves, and who knows, maybe even studying your notes afterwards. In short,
then, the event oshieru requires two agents: a primary agent that starts the event of
teaching, and a secondary agent that allows its accomplishment by initiating at the same
time a sister event of learning. It cannot do without either of them.
        The second fact is that the receiving entity is also affected by the event. Any event
of learning causes a change in the mind of the learner, from a change in the overall
quantity and quality of the information stored in her memory banks to the derived change
in thought and believes that the cognitive processing of the newly acquired knowledge will
originate. (Such an affectedness of the learner will be relevant to the alternative
construction of oshieru that I will discuss later).

       Having so analysed the event related by oshieru, let me show now how semantic
role labels would apply to its participants, and what cases they take.

       Role Structure of dative events

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 19 -
          Maishū kayōbi-ni Keiko-ga machi-no bunkakaikan-de kodomo-ni eigo-o oshie-teiru.
          every week(ADV) Tuesday-LOC Keiko-NOM town-GEN community hall-LOC child-
          DAT English-ACC teach-EXTENSIVE
          Every Tuesday Keiko teaches English to some children at the town's Culture Center.

           AGENT         An animate entity voltionally initiating an event. I distinguish between
                         animate and inanimate, non-volitional initiators. To the latter I assign the
                         label FORCE.
           PATIENT       An entity undergoing a change of physical or abstract state. I consider to be
                         a PATIENT the entity representing the resultative object of accomplishment
                         verbs of creation, sometimes associated with the role of RESULT.
           PATIENT       An entity undergoing a change of place. I keep this role distinct from the
           LOCATUM       more general PATIENT role in order to distinguish between the two different
                         kinds of affectedness of patients in a two-PATIENT event, in which either
                         patient is a P.LOCATUM.
           GOAL          The physical or abstract location where a PATIENT LOCATUM ends its
                         movement or where is directed. Following a weak form of Localist Approach,
                         I usually consider the role of GOAL to also encompass the final state of a
                         change-of-state event (FACTITIVE role) and the "animate goal in an event of
                         physical or abstract transfer of possession" (RECIPIENT), also defined as a
                         consciously affected goal (DATIVE).
           SOURCE        The physical or abstract location where a P.LOCATUM initiates its movement
                         from. It includes the giver of dative verbs.

        The first entity (Keiko) is obviously to be labelled AGENT. Indeed you see that I
define AGENTS as those animate entities that start events. Being the initiator of the event
without whom the chain of transmission of force would not begin, this entity represents the
main participant in the event oshieru. As such, it is realized as the subject, and is assigned
nominative case. But since this entity is also the starting point of a transfer of information,
it bears a second role of SOURCE. Predicate oshieru does not realize this second role
syntactically, in the sense that does not marks this argument in any way that would
explicitly express the role of SOURCE, but this is what its twin predicate narau "learn" does,
as I will explain in due time.
        The second entity (eigo), the transferred "thing", is affected by the act, and must be
labelled PATIENT. There are, however, two kinds of PATIENTS. Entities that change state or
are put into being during the event are labelled just that, PATIENTS. Instead, an entity that
changes place receives a different label, and is named PATIENT LOCATUM. The knowledge
transferred by oshieru belongs to this kind. The need of formally distinguishing PATIENTS
according to their types of affectedness comes from the fact that there are verbs, like
those of "painting", in Japanese nuru, that describe events involving both kinds of PATIENT,
and treat them quite differently. The alternation exhibited by nuru will be the topic of my
last lecture. The entity bearing the role of PATIENT LOCATUM (which I shall shorten in just
LOCATUM or P.LOCATUM) is the second most important participant in the event, and is the
direct object. It is assigned accusative case, by means of a particle (o) that, across the
case structures of several predicates, consistently suggests the condition of being under
the control of an AGENT.13 (Or being the focus of an the attention or conscious commitment
of some entity, which probably gives such experiencer some degree of control over the
        Finally, the third entity (kodomo), where the displaced information ends its run,
obviously bears the role of GOAL. The Localist Approach mentioned in the Role list is an
analytical tool originally developed by Ray Jackendoff, in which all events are treated as

     Aoki 2008.
                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
                                                      - 20 -
involving one entity and one place.14 I will briefly discuss the Localist Approach in the
relevant sections of my lectures; for the time being it will suffice saying that here I do use it,
even if in a weak form, and consequently I label GOALS all kinds of endpoints of a
movement. When no Localist Approach is used, an endpoint could be classified as GOAL
proper, when it's a physical place; as FACTITIVE, when is an abstract place, which means a
condition, and the event is one of change of state; as RECIPENT, when is an animate entity
and the event is a transfer of possession; or even as DATIVE, when the animate recipient is
consciously affected – this probably means that the animate recipient is also a conscious
experiencer. The Localist Approach allows me to unify these too many roles in just the one
role of GOAL, and so show how several similar classes of predicates syntactically treat the
GOAL role in the same way. It should also be clear why this third entity bear a secondary
role of agent. This entity – kodomo in the example – does not initiates the act oshieru, but
since an event of teaching entails a corresponding event of learning, this third entity is the
initiator of the twin act narau. As such, the volitional participation of this entity is also
necessary for oshieru to be accomplished. The two conversive predicates oshieru and
narau describe the same event, and therefore have the same truth-value. As I said above,
they describe the event from two opposite point of view, realizing their subject differently:
oshieru realizes as the subject the main agent, narau the secondary agent. Eventually, the
reason why this third entity is to be considered a PATIENT should also be clear. Being
taught and learning is a complex event that effects a change in the learner.
        The GOAL phrase receives the third grammatical relation, that of indirect object, and
is assigned an oblique case. This phrase is marked by ni, which is the default oblique case
marker in Japanese and, as such, probably the most "abused" marker of this language. I
have already discussed the semantics of ni, and it should be clear by now that ni is indeed
the best possible particle to mark an endpoint. (By the way, ni is also used as the marker
of the agentive case, used to specify the secondary agent with verbs describing two-agent
events. This is a puzzling coincidence, but I am not going to explore the matter). As a label
for this case marker, I believe the term dative is just right, since it is self-explanatory, and
allows to distinguish an animate goal from physical or abstract endpoints.

        Until now I have discussed only the three main phrases of oshieru, the so called
terms or arguments. On the extreme right of the Balloon Graph two other balloons are
sitting, slightly overlapping. They contain the two other phrases of the example above.
These phrases are not necessary for oshieru to completely describe an event type of
"giving". Their function, and the role of the "entities" they represent, is of placing the event
in time and space, as indicated by the role labels TIME and SPACE written inside the
balloons. But specifying time and space is something one can do with any type of event –
it is an instance of those "background" specifications I mentioned earlier on.
Correspondingly, the cases used to mark these roles are two of the so called local or
concrete cases. The role of SPACE is marked by a dynamic locative. Now this definition
may sound a kind of nonsense, since "locative" cases are properly placed in the domain of
static cases, while in fact cases like the dative and translative belong to the domain of
dynamic cases.15 But such a definition is necessary in Japanese for this language uses
two locative cases to specify, respectively, the place of a dynamic event, like an action or a
change or state, and the static location of objects. This is an instance of the former locative,

   Jackendoff 1983: 188-195. A discussion of the localist approach is in Levin and Rappaport Hovav 2005:
   See Lyons 1968: 300-301 on the opposition between "static" and "dynamic".
                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 21 -
whose marker is the particle de.16 The time adjunct, on the other hand, is marked by the
static locative marker ni, which is used to specify unmoving points in both space and time.

       To end up this section, let me draw some partial conclusion. I have defined case as
the expression of the logical and semantic relation between nouns and verbs, and
semantic roles as a description of how entities participate in events. In short, cases are
assigned by the predicate, roles by the event itself. Cases an roles can be used to dissect
a clause, or the marking pattern of an entire verb class, and to label all its parts. I have
used oshieru as an example of this analysis. By assigning case and role labels to each
phrase of a clause, I tried to show that a verb and its case structure are actually related to
the event and its role structure, and must be consistent with them. This relation between
semantics on the one hand, and lexicon, morphology and syntax on the other, is of great
help in the teaching of Japanese verbs, as it makes verbs and their case patterns easier to
remember, allows to draw a relation between predicates with similar case structures, and
eventually serves as a basis on which to build large classes of predicates. The graphic
implements that I use in my lectures are an important aid for doing that. Now it is time to
see them in detail.

The "Balloon Graph"
        In this section I am going to explain where Balloon Graphs come from, how they are
composed, what kind of information can display, and their overall merits.
        As I said earlier on, I am the developer of the Balloon Graph, but not its original
creator. Teramura books did not use any graphic aid, but sometimes my teachers of
Tsukuba University did, and traced on the blackboard prototypical Balloon Graphs such as
the following one.

        Crude balloon graph of oshieru

          だれが                                                だれに
          who?                                             to whom?


       (Here I added the English translation)
       Despite its crude form, this primitive graph has several merits. For instance, by
observing the number and marking of the phrases displayed in the upper ovals, a student
can easily deduce the existence of a "three-place" class of predicates. The question words
in the upper balloons also help identify the kind of entities participating in the event, so that
this graph might be said to link, somehow, semantics to syntax, and to represent argument
structure in a simple fashion. Moreover, if it is read as English text, left-to-right and up-to-
down, the whole balloon sequence captures the "ordinary" word order of a Japanese
   It may be interesting to notice that de is formed by the static locative particle ni (originally the infinitive for
of the copula "be") plus the verbal extensional affix te, to specify the fact that the place is not a single,
unmoving point, but rather a two-dimensional plane used extensively by entities that move and occupy in
succession several different points of it.
                                      Dalla Chiesa -The Semantics of Japanese verbs
                                                 Helsinki 4th-7th May 2009
                                                            - 22 -
        The use of graphs of this kind is quite diffused in contemporary language teaching
at the level of Primary or Junior High School. The following is a similar but more
sophisticated graph taken from an Italian language teaching manual. 17 The verb is the
same, but I changed the language to English.

          Simple Balloon Graph of the English predicate teach



        Keiko                                              children

        Drawing the arguments under the verb effectively hints at the fact that the predicate
can be seen as the dominating feature of a clause, in that it controls the number and
marking of its phrases. The predicate and its particles are boxed in order to distinguish
their shared "harder" nature from that of the nouns, which are just the fillers of the three
slots required by the predicate and are therefore displayed in "softer" oval balloons. Also,
the phrases marked by a direct case are directly linked to the predicate box, but the
indirect object "children" connects to the predicate via an intermediate, smaller box. This
box is attached under the belly of the verb and contains the particle to. I think the logic of
this is quite obvious. Finally, the double line linking the leftmost balloon to the predicate
has the aim to single out that phrase as the subject. A configurational approach of this kind
is the norm in Italian schools, and it adds a nice touch. So I decided to implement the
double line in my Balloon Graphs as well. In its complete form, this Graph would appear as

     D'Alfonso 2001: 13-16.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
                                                          - 23 -
      Starfish-like Balloon Graph of the English dative predicate teach, complete representation

        Instrumental adjunct                                                     Other adjuncts

                                              by                    ...
       Temporal oblique                                                           Locative oblique
         Time locative                               "teach"                       Place locative


             Subject                             Direct object                       Indirect object
            Nominative                           Accusative                               Dative

       This gorgeous starfish-like graph shows the complete set of phrases that compose
the clause of predicate teach, but could be used to represent the construction of any class
of predicates. Its main merit is that it does capture a hierarchy of constituents, with the
most important ones for the predicate prototype displayed at the bottom. These are the
three mandatory arguments of any predicate of giving. The dashed balloons sitting on the
two sides of the predicate represent a second "tier" of phrases, bound to local functions.
They are a group of "background" specifications, but are important because certain
classes of predicates require their presence. The upper ovals contain the other, least
important adjuncts. The actual construction of any given predicate can be obtained from a
matrix of this kind by pruning off the unneeded arms. But there is simply no room on the
page or on the blackboard to implement such a process, the graph is too large and
complicated, and, at least for English, flawed by its impossibility of representing the double
object alternation.
       In a simpler, slightly altered form, however, a graph of the same kind could display
the construction of oshieru.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 24 -
       Balloon Graph of dative predicate oshieru, with boxed particles


             が           を            に

       The merit of this graph is that the particles are represented as "appendices" of the
verb itself. However, with the exception of certain adverbs, in Japanese all the phrases in
a clause are realized with the "mediation" of a particle. Therefore, it is totally superfluous to
remark this by connecting their balloons to the verb via a "particle box". I prefer a "full
balloon approach". I repeat here the Balloon Graph of oshieru.

       Complete Balloon Graph of dative predicate oshieru


                                                                             で de       に ni
        が ga                                                                Dynamic     Static
      Nominative                                                            Locative   Locative
                                                                              PLACE     TIME
         AGENT           Accusative               に ni
        SOURCE             PATIENT                Dative
                          LOCATUM                 GOAL

        Displaying both predicate and phrases in ovals rather than in boxes makes it easier
to understand that particles and verbs are the main conveyors of meaning in a clause.
Moreover, circles are easier to trace on the blackboard and offer more line-connecting
points when drawn by Microsoft Word. Finally, writing case particles within each balloon is
more elegant and simple than boxing them, and better represents the logical relation that
links them to the predicate.
        In my Balloon Graphs, each balloon represents a single phrase, and displays a
case name, its marker, and the label of the semantic role thus realized. The notion of

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 25 -
grammatical relation is not really important in my approach, and I seldom use it when I
teach Japanese verbs to my students. I could find some room for it inside the balloons, or
just outside them, but I decided to do without it altogether.
       The subject is displayed on the extreme left, and its balloon is connected to the verb
balloon with a double line. The other arguments are grouped under the verb. Adjuncts are
drawn on the extreme right, away from the balloons of the argument phrases. All possible
adjuncts could be included in the graph, if space and time allow it, but usually this is not
necessary, as students soon learn that most common adjuncts can be added to any kind
of clause.

       Before proceeding further, I need to remark the fact that my balloon graphs are not
well suited for representing the word order of a Japanese clause. This is probably their
only weak point. If faithful to the syntax, a balloon representation of a clause would appear
as nothing more than a mere circling or boxing of the constituents.

      Japanese syntax, graphic representation of a three-argument clause

           が                   に                      を                      教える

       Balloon Graphs have the purpose of representing in two dimensions a melody of
phrases, in other words, phrases that are originally sequenced in one dimension only. The
second, added dimension helps to graphically visualize those logical and semantic
relations between the constituents of the clause that do not easily show up in one
dimension. I found out that trying to also maintain the linear order of the phrases would be
too complicated and confusing, and eventually would decrease the heuristic power of the
Balloon Graphs. This is why I decided of not concerning myself with word order.
       But I want to add that Chinese clauses can be represented by a Balloon Graph so
that the natural order of the constituents be maintained. The following are the two "insect-
like" Balloon Graphs of the alternate constructions of three-place predicate 教 jiāo.

      Balloon Graph of the double-object construction of Chinese dative predicate jiāo

              Tā jiāo xiǎoháizi yīngwén.
              (NOM)-he teach (DAT)-child (ACC)-English
              He teaches English to children.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 26 -

          他                                          小孩子                              英文
        Subject                                 Indirect Object                   Direct Object
       Nominative                                    Dative                        Accusative

        This Graph represents the two-object construction of jiāo. Constituents are read left
to right, following the natural word order of the clause. I omitted roles labels. I do not want
to spend too much time on these Chinese examples It should only be noticed that
according to my approach Chinese has also case; only, Chinese case is mostly marked
positionally rather than by specific phonological markers.

       Balloon Graph of the alternate construction of Chinese dative predicate jiāo

               Tā gěi xiǎoháizi jiāo yīngwén.
               (NOM)-he DAT-child teach (ACC)-English
               He teaches English to children.



          他                                小孩子                                         英文
        Subject                       Indirect Object                              Direct Object
       Nominative                          Dative                                   Accusative

        This Balloon Graph shows the so called dative alternation of jiāo, characterized by
the presence of an indirect dative phrase between subject and predicate, in the required
position of concrete, local cases. Beside having a fixed position, Chinese local case are
also marked by prepositions. The result is a three-leg "insect" with a anatomical plan
completely different from that of the former Graph, but still displaying a full class dative
predicate. I make no further comment on this chart, but just notice two things: the boxing of
the preposition 给 gěi and the fact that this dative phrase cannot be omitted, and therefore
is not an adjunct.

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 27 -
      Having spent so much time on oshieru I think the right thing to do now is showing
the Balloon Graph of its conversive predicates narau and osowaru.

       Balloon Graph of the predicates of "receiving" narau/osowaru

       Keiko-ga Jon-ni eigo-o narat-ta.
       Keiko-NOM John-AGE English-ACC learn-PAST
       Keiko was taught English by John.

       Keiko-ga Jon-kara eigo-o narat-ta.
       Keiko-NOM John-ABL English-ACC learn-PAST
       Keiko learned English from John.

                                   習う narau
                                  教わる osowaru

        が ga                           から kara                                       をo
      Nominative                       Ablative                                  Accusative
         AGENT                           SOURCE                                  P. LOCATUM

                                         に ni

       Here I omit all possible adjuncts, to better focus on role and case structure. There is
no difference in meaning between narau and osowaru; both mean "to learn" and are
conversive with respect to oshieru. Being conversive means that oshieru and narau
describe not only the same type of event, but also the same individual, specific event, or
event token. Since the event is the same, the involved entities are also identical. There is
an agentive entity that transmits some knowledge as both AGENT and SOURCE, the
transmitted object as LOCATUM, and a receiving entity as GOAL, AGENT and also PATIENT.
Verb narau, however, allows the speaker to describe the event by means of a different
case structure, through realizing as the subject not the AGENT initiator of the act of teaching,
but rather the entity bearing the role of the affected GOAL. There are basically two reasons
for describing the event using narau rather that oshieru. One is to give the GOAL entity the
function of a syntactic pivot, so that it can be the subject of several sentences. The other is

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 28 -
to simply underscore its agency and commitment in the act of learning. Once using
predicate narau, however, the speaker is offered one more choice. She can assign to the
SOURCE entity, which is the agentive subject of oshieru, either the ablative or the agentive
case. The speaker can operate this choice following her own taste: by choosing the ni
agentive marking she will underscore the active, stimulating role played by the teacher in
the learning process of his pupil; whereas by choosing a kara ablative she will focus the
event description on the learner and on their active involvement.
       The choice offered by narau between ablative and agentive case is displayed by the
two stacked balloons in the central portion of the Graph. Stacked balloons represent
phrases that cannot cooccur. In a Balloon Graph I link these balloons with a vertical
continuous line. Again, using role labels in combination with case names is of great help in
explaining the occurrence of the two oblique phrases, their difference in meaning and the
reasons of their choice.

       Now, the balloon graphs of oshieru and narau display in two dimensions the
originally linear structure of the clauses of these two predicates, and also allow to show the
choice between kara or ni offered by narau. But since these two predicates are conversive
and describe the same event, it would be very interesting to add a third dimension to the
graph, and display the Balloon Graphs of the two predicates so that they lay on two
separate but contiguous planes – or layers. Then we would see the Balloon Graph of
oshieru on the higher, more superficial plane, and underneath it, on the second plane, the
graph of narau. In each Graph, the balloon representing a given semantic role would be
placed in the same position relative to the predicate. In the 3D display then, the balloons
displaying a given role would be seen superimposed right over each other. However, there
is no way to actually draw a 3D Graph of this kind. Even if I could find a system for building
such a display and giving it a meaningful shape, it would be too difficult to use in the
classroom, and probably confusing.

      The "Table Graph" (non-benefactive conversive pairs of "transfer of
      In order to show together the two or more case structures that characterize
conversive predicates, I have developed another graphic implement: the "Table Graph". I
am discussing it now for the first time.

      Table graph of non-benefactive predicate pairs of transfer of possession

       (1)    田中先生が二年生に英語を教える。
              Tanaka-sensei-ga ninensei-ni eigo-o oshieru.
              Tanaka-professor-NOM second grade-DAT English-ACC teach
              Prof. Tanaka teaches English to the second grade.

       (2)    田中先生が二年生を教える。
              Tanaka-sensei-ga ninensei-o oshieru.
              Tanaka-professor-NOM second grade-ACC teach
              Prof. Tanaka teaches the second grade.

       (3)    二年生が田中先生に英語を教わる(習う)。
              Ninensei-ga Tanaka-sensei-ni eigo-o osowaru (narau).
              second grade-NOM Tanaka-professor-AGE English-ACC learn
              The second grade are taught English by Prof. Tanaka.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 29 -
       (4)     二年生が田中先生から英語を教わる(習う)。
               Ninensei-ga Tanaka-sensei-kara eigo-o osowaru (narau).
               second grade-NOM Tanaka-professor-ABL English-ACC learn
               The second grade are taught English by Prof. Tanaka.

       (5)     サツキがトトロに傘を貸した。
               Satsuki-ga Totoro-ni kasa-o kashi-ta.
               Satsuki-NOM Totoro-DAT umbrella-ACC lend-PAST
               Satsuki lent an umbrella to Totoro.

       (6)     トトロがサツキに傘を借りた。
               Totoro-ga Satsuki-ni kasa-o kari-ta
               Totoro-NOM Satsuki-AGE umbrella-ACC borrow-PAST
               Totoro borrowed an umbrella from Satsuki.

       (7)     トトロがサツキから傘を借りた。
               Totoro-ga Satsuki-kara kasa-o kari-ta
               Totoro-NOM Satsuki-ABL umbrella-ACC borrow-PAST
               Totoro borrowed an umbrella from Satsuki.

       (8)     恵子が正男にお金を預けた。
               Keiko-ga Masao-ni okane-o azuke-ta.
               Keiko-NOM Masao-DAT money-ACC give in custody-PAST
               Keiko left her money with Masao.

       (9)     正男が恵子にお金を預かった。
               Masao-ga Keiko-ni okane-o azukat-ta.
               Masao-NOM Keiko-AGE money-ACC receive in custody-PAST
               Masao received some money in custody from Keiko.

       (10)    正男が恵子からお金を預かった。
               Masao-ga Keiko-kara okane-o azukat-ta.
               Masao-NOM Keiko-ABL money-ACC receive in custody-PAST
               Masao received some money in custody from Keiko.

                      (1) 教える                (2) 教える             (3) 教わる・習う          (4) 教わる・習う
Non-benefactive         oshieru                oshieru             osowaru, narau      osowaru, narau
pairs of transfer
 of possession
                       (5) 貸す                                         (6) 借りる           (7) 借りる
                         kasu                                            kariru            kariru

                      (8) 預ける                                        (9) 預かる           (10) 預かる
                        azukeru                                        azukaru           azukaru
   AGENT            nominative が ga       nominative が ga           agentive に ni
   SOURCE                                                                              ablative から
   P.LOCATUM        accusative を o                                 accusative を o     accusative を o
   GOAL               dative に ni
   AGENT                                                           nominative が ga    nominative が ga
   PATIENT                                 accusative を o

                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
                                                          - 30 -
        Those with some linguistic training would immediately recognize a strong similarity
of this Table Graph with the Theta-Role Grid of Generative Grammar. Indeed, the Table
and the Grid display a very similar kind of information (such as the semantic roles of the
entities involved in the event), but are oriented differently – their axes are actually inverted.
However, Tables and the Grids have different purposes and differ in a crucial point: I build
each Table so that it can display together in a single organized space not only all the
alternate case structures of individual predicates, but also the several constructions of
entire classes of predicates, and those of conversive pairs as well. Refer to the sequence
of open boxes at the top of the graph above. This Table plots the constructions of the
whole class of the verbs of giving to which oshieru, kasu and azukeru belong, and can
even accommodate the two-place alternate construction exhibited by oshieru alone. In
short, by this Table Graph one can display together all the alternate case structures used
to describe an entire event class, that of transfer of possession. This the reason of the title
I gave to this Table. I am going to explain later what "non-benefactive" means.

        In a Table Graph, the thicker horizontal lines separate the participants involved in
the event. Within each major horizontal band, the thinner, dashed lines keep distinct the
semantic roles borne by these entities. The labels of these roles are listed in the leftmost
column. The topmost open row of boxes displays the predicates under analysis. Each
predicate or group of predicates is at the top of a column of boxes, which show all the
relevant phrases that can occur in the clauses of that predicate, so that constructions are
read vertically. Each single box displays one phrase, and contains the case and particle
that mark that phrase. In turn, each boxed phrase belongs to a horizontal row of boxes
which show all the possible syntactic realizations of the semantic role(s) displayed in the
leftmost header box.
        What my Tables do not contain are a reference to Grammatical Relations, a
systematic reference to the type of the phrases displayed (NP or DP, PP, CP), and indices
to relate each box to specific phrases occurring in the examples. This is because, as I
have already remarked, Grammatical Relations are not relevant to my way of teaching
Japanese verbs. The type of phrases is also irrelevant, since I consider all phrases to be
related to the verb and bear meaning and case. Finally, the indexing function is substituted
by displaying case particles, which allow an easy quick reference to the examples.

        In the case structures of all predicates, including oshieru (2), an entity bearing the
two roles of AGENT and SOURCE is realized. This entity is the subject only with the
predicates of giving. With those of receiving, this entity is realized as an oblique; either by
means of the agentive case ni, when the speaker wants to underscore its agentivity, or by
the ablative kara, when she rather wants to underscore the act of receiving. The LOCATUM
entity is realized as the direct object in all the constructions of the conversive pairs (i.e., all
constructions but that of oshieru (2)), ubiquitously marked by the accusative case. Another
receiving entity is also realized in all constructions. The conversive pair oshieru/narau lets
the speaker choose between two alternate constructions: one in which this third entity is
the GOAL of the transfer of knowledge, and is obviously marked by means of the dative,
and one in which it is the AGENT and initiator of the act of receiving, and is realized as the
subject. In the header on the left of the Table this entity is labelled as both GOAL and AGENT.

      The two-place construction of oshieru of Column (2) is of particular interest, and the
possibility of displaying it along with the case structures of the other predicates is one of
the merits of the Table Graph. What happened here is that the three-place predicate
oshieru (1) lost one argument, its direct object. This caused a new two-place predicate,
oshieru (2), to be born. The meaning root of this new verb does still entail that a transfer of

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
                                                     - 31 -
information is happening, but its actual contents do not matter, are not entailed in any way
and are not subcategorized. Having the LOCATUM entity so disappear from the clause, the
GOAL entity now comes to be realized as the object. This change of valence is possible
because the GOAL entity – the ninensei – is also an affected participant, as I have
discussed above. The event description of oshieru (2) is focussed on the special relation
binding a teacher and her pupils, rather than on the transfer of knowledge between them.
        Now, loosing an argument in this way also causes a loss of information – a listener
will not know any more what is taught. This is why oshieru (1) and (2) might well describe
different event tokens (actually oshieru (2), being less specific, includes oshieru (1)), and
the new predicate oshieru (2) cannot be considered to be conversive with respect to
osowaru and narau.
        For as much as I know, oshieru is the only verb of giving that alternates in this
fashion. This is probably due to the fact that the affectedness displayed by the learner in
an event of teaching/learning is of a much higher degree than that exhibited by the GOAL
entities in all other the events of giving/receiving. (I have already discussed why and how a
learner is affected by the event of learning.) Anyway, this problem deserves further

Benefactive conversive pairs of "transfer of possession"
       Having so discussed the general basis of my approach and how my two graphic
implements are build, from now on I will apply these heuristic tools to the main classes of
Japanese predicates. Since I have already treated the case structure of the predicates of
transfer of possession, I will stick to the same topic and deal now with the benefactive
predicates of giving and receiving.

      Balloon Graph of benefactive predicate of giving ageru

       (1)   サツキがトトロに傘をあげた。
             Satsuki-ga Totoro-ni kasa-o age-ta.
             Satsuki-NOM Totoro-DAT umbrella-ACC give<BEN>-PAST
             Satsuki gave an umbrella to Totoro.

      (2)    サツキがトトロに傘を貸してあげた。
             Satsuki-ga [Totoro-ni kasa-o kashi-te] age-ta.
             Satsuki-NOM [Totoro-DAT umbrella-ACC lend-COMP] give<BEN>-PAST
             Satsuki lent an umbrella to Totoro.

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
                                                      - 32 -

        が ga
      Nominative                 に ni
         AGENT                   Dative
        SOURCE                   GOAL
                                                                               PATIENT LOCATUM
                                                                                Affected thing

                                                                               Clause v. て te
                                                                                 Effected act

        In this Graph there is a dashed line connecting two balloons. I use a horizontal line
like that to indicate that the markers displayed in the two balloons are typically used jointly.
In this graph, the two balloons joined in this way show how arguments are realized when
the verb refers to the transfer of possession of a physical object. There is no need of
discussing again in detail the roles and interactions of the participants in this type of event,
since they are nearly identical to those discussed for non-benefactive predicates. The only
important difference is that the GOAL entity now bears one more additional role, that of
BENEFICIARY. This role refers to the entity that is advantaged or disadvantaged by an event.

        BENEFICIARY     The entity advantaged or disadvantaged by an event. So labelled to
                        distinguish it from the benefactive case. When the BENEFICIARY and the GOAL
                        (DATIVE) roles are borne by a same entity, its NP is marked by a dative and
                        the two roles cannot be syntactically or semantically distinguished. A distinct
                        benefactive marking appears only when the BENEFACTIVE and GOAL roles are
                        borne by different entities.

        The GOAL entity bears such an additional role because of the particular lexical
nature of the verbs of this group. Their name, "benefactive predicates", comes from the
fact that, besides being ordinary verbs of giving for as much as the dynamics of the
transfer is concerned, they also entail that the receiving entity beneficiates from the event.
It is important to notice that the other verbs of giving discussed earlier do not entail such a
thing: It can surely be that teaching English to a group of children would end up in an
advantage for them, but this is not suggested neither by oshieru, by any other verb of
giving such as kasu, uru, azukeru and so on, or by their conversive companions. Only the
verbs of this particular subset do entail that the goal entities are somehow advantaged by
the transfer. These verbs are commonly called "yarimorai verbs", and include both the

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
                                                     - 33 -
verbs of giving yaru, ageru, sashiageru, kureru, kudasaru and their conversive
companions of receiving, morau and itadaku.18
        However, in the clauses of yarimorai predicates there is no way of giving any
syntactical expression to this beneficiary role by means of marking the GOAL/BENEFICIARY
phrase other than with the dative. We could say that the BENEFICIARY is a sort of
transparent role, invisible to syntax, that bears no influence on the structuring of the clause.
        What is very important for syntax is that yarimorai verbs allow for an additional kind
of object phrase. This phrase is displayed on the right of the graph, stacked under the
balloon of the PATIENT LOCATUM phrase. As it happens, the object of the benefactive verbs
of giving and receiving does not need to be an affected physical entity. It can be an act
that the agentive SOURCE does, or effects, in behalf of the BENEFICIARY. As a performance,
this effected act must be expressed by a clause, and there ought to be a morphological
device for substituting the accusative phrase with a predicate or a whole clause. The
balloon stacked under the o NP balloon shows precisely that this device is the –te form of
the verb: it the complementizer used to mark a clause as an argument and assign it to
ageru. As I said, rather than an affected patient dislocated in space, this argument is an
effected act, in other words an act entirely originated by the agent. As such it should be
considered to have an ordinary PATIENT role.

       Of course a similar Balloon Graph can be build for the conversive predicate morau
as well. Here it is.

        Balloon Graph of benefactive predicate of receiving morau

         (3)    トトロがサツキに傘をもらった。
                Totoro-ga Satsuki-ni kasa-o morat-ta.
                Totoro-NOM Satsuki-AGE umbrella-ACC receive<BEN>-PAST
                Totoro got an umbrella from Satsuki.

        (4)     トトロがサツキから傘をもらった。
                Totoro-ga Satsuki-kara kasa-o morat-ta.
                Totoro-NOM Satsuki-ABL umbrella-ACC receive<BEN>-PAST
                Totoro got an umbrella from Satsuki.

        (5)     トトロがサツキに傘を貸してもらった。
                Totoro-ga Satsuki-ni [kasa-o kashi-te] morat-ta.
                Totoro-NOM Satsuki-AGE [umbrella-ACC lend-COMP] receive<BEN>-PAST
                Totoro borrowed an umbrella from Satsuki.

        (6)     トトロがサツキから傘を貸してもらった。
                Totoro-ga Satsuki-kara [kasa-o kashi-te] morat-ta.
                Totoro-NOM Satsuki-ABL [umbrella-ACC lend-COMP] receive<BEN>-PAST
                Totoro borrowed an umbrella from Satsuki.

  The BENEFICIARY role, then, is not assigned by the event, in the sense that is not a property of the goal
entity which can be inferred from the objectively observable unfolding of the event. Rather, it is a role
assigned by the verb, or, better, entailed by its meaning. This suggests that the BENEFICIARY role should not
be considered to be a "proper" role.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 34 -

        が ga                      から kara
      Nominative                  Ablative
         GOAL                      SOURCE
                                                                            PATIENT LOCATUM
                                                                              Affected thing

                                   に ni                                     Clause v. て te
                                  Agentive                                  Complementizer
                                    AGENT                                           PATIENT
                                                                                  Effected act

       There is no need to comment on this Graph, since it is identical to the Balloon
Graph of the ordinary non-benefactive predicates of receiving. The only difference is that a
–te argument is now stacked under the accusative phrase.
       Again, it would be nice to manage a three dimension Balloon Graph displaying both
conversive companions. I have to resort to build a Table instead. The numbers refer to the
previous examples.

       Table Graph of benefactive predicates of transfer of possession (yarimorai)

Benefactive pairs        上げる・やる                                       もらう・いただく
  of transfer of            ageru, yaru                                 morau, itadaku
   possession           (1)             (2)            (3)             (4)           (5)             (6)
  AGENT 1            nominative nominative           agentive                     agentive
                       が ga           が ga            に ni                          に ni
  SOURCE                                                             ablative                      ablative
                                                                    から kara                       から kara
  P.LOCATUM          accusative                     accusative      accusative
  affected thing        をo                             をo              をo
  PATIENT                            COMP                                              COMP        COMP
  effected act                       clause                                            clause      clause
                                      て te                                              て te        て te
  GOAL                 dative
                       に ni
  AGENT 2                                          nominative       nominative       nominative   nominative
                                                     が ga             が ga             が ga         が ga

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 35 -
        I want to remark two things about this Table. First, all reference to the BENEFICIARY
role borne by the GOAL entity is omitted. This is because this role is semantically entailed
by the verb, is not explicitly marked in any way, and therefore does not have any influence
on case structure. Second, the PATIENT role is split in two: a LOCATUM physical thing,
marked by the accusative case, and an effected act, marked by the –te complementizer.
        Again, a Table Graph allows the teacher to display in one large, well organized grid
all the phrases that characterize the case structure of yarimorai verbs, thus showing a
clear connection between the domain of meaning and that of grammar.

Transitive predicates of change of location
       Now, just to complete the discussion of this particular set of three-place predicates,
I want to deal with the predicate pairs of sending/putting and taking.

       Ani-ga haha-ni shorui-o Amerika-kara Aomori-no ie-e okut-ta.
       my elder brother-NOM mother-DAT documents-ACC America-ABL Aomori-GEN
       house-TRA send-PAST
       My elder brother sent the documents from the USA to my mother at the house in

      Balloon graph of predicate of sending okuru


                                                                        から kara       へe
        が ga                                                            Ablative   Translative
      Nominative                                                          SOURCE      GOAL
         AGENT         Accusative               に ni
        SOURCE            PATIENT

        After this lengthy discussion, it may seem trivial to analyze one more predicate of
the same category, especially a verb such okuru, which describe the transfer in space of a
physical thing in a very concrete way. In fact, the construction of this predicate is not that
simple. Basically, predicate okuru describes a displacement, in other words the physical
relocation of an object in space, and requires three arguments. The first two correspond to
the initiator of the displacement and to the displaced thing. These arguments bear the
roles of AGENT and LOCATUM respectively; it is assumed that the AGENT entity bears the role
of SOURCE too. The third argument corresponds to the GOAL entity. In the event the GOAL is
an animate entity, it is marked by a dative case. If it is an inanimate entity, its marker
should be interpreted as a translative.

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
                                                      - 36 -
       This distinction is crucial because when SOURCE and GOAL are both animate, there
are instances in which two more SOURCE and GOAL phrases can be added to the clause.
When this happens, the added phrases must specify concrete, spatial locations, and are
therefore marked by the local, concrete cases ablative and translative respectively. In such
constructions, the event described turns into an event of both relocation in space and
transfer of possession The animate SOURCE now represents the original possessor, the
animate GOAL represents the new possessor, and the adjunct phrases specifies the
physical locations of the two animate participants.
       Again, a Table manages to capture these complex relations quite nicely.

       Table Graph of predicate of sending okuru

                        送る okuru                     送る okuru                          送る okuru
  Predicate of          displacement                displacement                      displacement
    sending                                                                       transfer of possession

   AGENT              nominative が ga             nominative が ga                   nominative が ga
   P.LOCATUM          accusative を o               accusative を o                    accusative を o
   GOAL                 dative に ni                                                    dative に ni
   GOAL                                            translative に ni
   SOURCE                                                                          ablative から kara
   GOAL                                                                              translative へ e

        I decided not to add to the Table the label, and the corresponding band, of the
additional SOURCE role borne by the AGENT. This is because it does not exert any influence
on the surface structure of the sentence. The distinction between an animate GOAL marked
by a dative and an inanimate GOAL marked by a translative case in the central band is due
to the fact that both GOAL entities are syntactically realized when okuru takes the additional
function of describing a transfer of possession, and their cases must bear different labels.

       The predicates of putting and taking are much simpler in comparison.

       Balloon graph of the predicate of putting oku

       Sensei-ga hon-o tsukue-no ue-ni oi-ta.
       teacher-NOM book-ACC desk-GEN upper part-TRA put-PAST
       A teacher put a book on the table.

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 37 -

        が ga
      Nominative                                                   に ni
         AGENT                     をo                           Translative
        SOURCE                  Accusative                           GOAL

       Oku is a simple verb of relocation in space. The structure of a oku clause is simple
because the GOAL is always an inanimate entity, represented by a place noun, and the
verb does not take the additional meaning of transfer of possession. There is no
cooccurring of animate and inanimate GOAL roles, and neither of animate and inanimate
SOURCE roles. Slightly more complex is the representation of a predicate of taking such as
toru (which, with some effort, might be considered to be conversive with respect to oku).

      Balloon graph of predicate of taking toru

      Keiko-ga tana-kara hon-o te-ni tot-ta.
      Keiko-NOM shelf-ABL book-ACC hand-TRA take-PAST
      Keiko took a book from the shelf (and put it) in her hand.


                                                                                    に ni
        が ga                                                                     Translative
      Nominative           をo                                                       GOAL
        AGENT           Accusative            から kara
        GOAL              PATIENT

       Okuru and oku could be considered to be three-place predicates, in the sense that
they require a GOAL phrase in order to describe the event in its complete form. This holds
even if, as I have noticed, such GOAL phrase is not considered to be an argument NP, but
                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 38 -
rather an adjunct PP. Once we accept this classification, then a predicate of taking such as
toru should also be considered basically trivalent, since it merely describe the inverse act –
in other word, the displacing of an object following an opposite movement vector, from a
physical SOURCE place to the AGENT. I composed the Balloon Graph above following this
line of reasoning. In this event subtype, an AGENT transfers a LOCATUM from a place to
herself, and by doing so also takes the role of GOAL. However, since this AGENT/GOAL entity
is animate, the event could be also read as a transfer of possession, or, more specifically,
as a "taking of possession". Indeed this possible interpretation is confirmed by the
possibility of adding to the construction an adjunct representing a physical goal, namely
the body part of the AGENT where the LOCATUM ends its movement. When this adjunct is
realized, the AGENT stands as a sort of abstract GOAL of the transfer of possession. From a
purely semantic point of view, the event described in this way is fairly composite: it is an
event of transfer of possession, of displacing, and also of putting ("taking possession of a
thing by removing it from a place and putting it in one's hand").
        The Table plotting these relations is surprisingly simple.

       Table Graph of the predicates of putting and taking

 Predicate pair         置く oku                      取る toru
 of displacing             "put"                      "take"
   AGENT            nominative が ga            nominative が ga
   P.LOCATUM         accusative を o             accusative を o
   SOURCE                                      ablative から kara

   GOAL              translative に ni           translative に ni
   inanimate                                       (body part)

       Again, since the GOAL role borne by the agent has no influence on the case
structure of either predicate, there is no reason to specify it in the Table.

Intransitive predicates of change of state and of change of place
        Until now I have extensively discussed a number of predicates that describe a
change of place. There is another important group of predicates that describe a change,
but of state instead of physical place. The most representative predicates of this subclass
are the intransitive verbs naru and kawaru and their transitive "causative" companions
suru and kaeru. While discussing these predicates I am also going to deal with the
subclass of the verbs of movement iku and kuru, that along with the predicates of change
of state form the large class of "translative predicates".
        The first predicate I want to discuss is kawaru, because its case structure is
relatively simple.

       Balloon Graph of the intransitive predicate of change of state kawaru

       Parii no fasshon no eikyō de, josei-no kamigata-ga chōhatsu-kara tanpatsu-ni
       Paris-GEN fashion-INS woman-GEN hairstyle-NOM long hair-ABL short hair-TRA
       Under the influence of Paris' fashion, women's hairstyle changed from long to short.

                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 39 -
       Parii no fasshon no eikyō ni yotte, josei-no kamigata-ga chōhatsu-kara tanpatsu-ni
       Paris-GEN fashion-INS woman-GEN hairstyle-NOM long hair-ABL short hair-TRA
       Under the influence of Paris' fashion, women's hairstyle changed from long to short.


        が ga
      Nominative                     から kara                     に ni           によって
        PATIENT                      Ablative                 Translative        niyotte
                                       SOURCE                      GOAL        Instrumental

                                                                 へe               で de
                                                              Translative      Instrumental
                                                                   GOAL          FORCE

        Verb kawaru describes an event of change of state. The main participant in this
event is the entity that changes state, which is realized as the subject. Even if this
participant is animate, and actively tries to originate all the conditions that will ultimately
cause the change, she has no actual control on the change itself, which happens
spontaneously. This is why the subject entity bears the role of PATIENT rather than that of
AGENT. This entity does not move in physical space, but between two conditions, and I
believe it should be considered to be an ordinary PATIENT rather than a LOCATUM. Two
more "entities" can be mentioned in a kawaru clause: the original and the final state of the
change. (I treat them as all them "entities" but they are not independent participant in the
event.) The original state bears the role of SOURCE, and kawaru assigns to its phrase the
ablative case. The final state bears the role of GOAL, and to its phrase a translative case is
assigned that can be marked either by the particle ni or, under certain syntactical
conditions, by e, with no difference in meaning. These two phrases are not necessary, as
kawaru is strongly focussed on the subject of the change. This is why they are adjuncts,
and I have placed them on the right of the Graph.
        Intransitive predicates like kawaru have no need to specify the entity causing the
change. This entity, however, can be mentioned in their clauses, but as an adjunct. The
balloons referring to this entity are stacked on the right of the Graph. It should be noticed
that there is no way of adding to the clause of kawaru a phrase referring to an external,
animate AGENT. To be mentioned, a causing entity must be inanimate, so bearing the role
FORCE. It is the first time I mention such role, which I define as follows:

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
                                                     - 40 -
       FORCE             An inanimate entity initiating an event. An entity is to be considered to bear
                         the role of FORCE when initiates an event as a mechanical or natural cause.
                         If causing an event of perception can be distinguished as STIMULUS.

       Whenever a FORCE adjunct is mentioned, its phrase is assigned an instrumental
marker. Two are the instrumental markers in Japanese: particle de (the most common
one), and particle niyotte, which is more limited in use. There is no difference in meaning
between them, and kawaru can assign either. The case structure of predicate naru is more
complex and troublesome to represent by a Balloon Graph.

      Balloon Graph of the predicate of change of state naru

      Musuko-ga isha-ni nat-ta.
      son-NOM doctor-TRA became-PAST
      My son became a medical doctor.

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
                                                      - 41 -

                                     に ni
        が ga                      Translative                                から kara
      Nominative                       GOAL                                  Ablative
        PATIENT                                                               SOURCE


                                な Adj.: に ni

                               い Adj.: く ku

                             Clause v. B4 ように

       Basically, the event described by naru can be represented in the same way as that
of kawaru, and there is no need to deal with it separately. But there are very important
differences too, as can be guessed from this peculiar Balloon Graph. First of all, it should
be noticed that by using naru the speaker focusses very heavily on the resulting state of
the change. This is why the phrase describing such a new state is absolutely necessary in
a naru sentence, and has the status of argument. For example, you will never hear
something like Masao-wa dōshitanda (What happened to Masao?)– *Masao-wa natta
(*"Masao became"), with no specification of the new condition taken by Masao. You
should say Masao-wa byōki-ni natta ("Masao got sick") or something like that. Such stress
on the new condition is so strong that the SOURCE phrase referring to the original condition
is absolutely superfluous. With some effort it could be added to a naru clause, but is very
seldom used. I displayed it as an adjunct on the extreme right of the Graph. To show its
status of "unusual" phrase, I connect it to the verb with a dashed line.
                             Dalla Chiesa -The Semantics of Japanese verbs
                                        Helsinki 4th-7th May 2009
                                                   - 42 -
        But probably the most striking difference between naru and kawaru is the large
choice of GOAL phrases allowed by naru. In the Graph, these phrases are displayed as an
impressively high tower of stacked balloons right under the verb. In this tower, each oval
displays the realization of a different lexical category, from nouns in the two topmost
positions to an embedded clause at the bottom. Of course these phrases cannot cooccur –
that is why their balloons are stacked in this way.
        The first two top ovals display the instance when the GOAL condition is referred to by
a noun. Again, the predicate assigns to this noun a translative case marker, usually
realized by the particle ni. The e case marker of the second balloon is not very common
with naru, and students hardly meet it in classes. It is possible to use it, however, and it is
important to display it here because it helps show the strong similarity of the case structure
of naru with that of a predicate of change of place like iku.
        But conditions can also be expressed by adjectives, Actually, the true function of an
adjective is exactly that of describing a condition. The third and fourth ovals in the stack
show what happens in this instance. The third balloon shows what happens when the
adjective belongs to the –na class. In this instance, the –na morpheme (actually, a copula
in adnominal form) is substituted by the –ni adverbial morpheme. This is the very same ni
that is assigned to nouns in order to mark the translative, dative and several other cases.
This is not strange, given that a –na adjective without –na is just a noun. It so happens that
in Japanese the translative case and a large class of adverbs are marked by the same
particle, ni.19 The fourth balloon shows that when the new GOAL condition is described by
an adjective of the –i class, the final morpheme –i of the adjective is substituted by the –ku
adverbial morpheme. This means that the adverbial particle –ku of –i adjectives actually
functions as a translative case marker.20
        But conditions can be expressed by verbs too. There are examples of this in all
languages: in Italian, for example, the verb gioire means "be happy", just as the verb
yorokobu in Japanese. Actually the Japanese language is particularly rich of such verbs,
which are named stative. Therefore, there must be a way of marking a verb, or a whole
clause, as the GOAL argument of naru. With the benefactive predicates of yarimorai, the
morphological device used to do that was the –te form. Naru does not assign a –te form,
but the complementizer yōni, which has the same function and could be analyzed as a
noun yō meaning "condition" followed by the translative marker ni. It should be noticed that
the use of the (no)yōni complementizer is not restricted to stative predicates, but can occur
with predicates describing actions or spontaneous events as well.
        I draw this Graph when I deal with Unit 8 of our ICU textbook, after about 15 or 16
weeks of classes. At this early stage, students still have little knowledge of the plain forms
of verbs, so that they would not know how to use the predicate naru in this fashion. When I
fear that my students would be confused by a full treatment of this topic, I just explain the
function of the bottom balloon and leave it empty.

       What I do is showing how the case structure of kawaru (and, to a lesser degree, of
naru) actually matches with that of the motion verbs of change of place.

        Balloon Graph of motion verb iku


   This fact is not a coincidence, and suggests that most Japanese adverbs (those marked by –ku, –ni and
abstract, subjective –to) are actually translative phrases specifying a new condition originated by the act or
event described by the verb.
   My analytical approach differs from Jacobsen's (1992: 113), who considers ni naru to be a unitary
monovalent predicate.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
                                                          - 43 -
        Tōkyō-eki-kara densha-de Tsukuba-e ik-ō.
        Tōkyō-station-ABL train-INS Tsukuba-TRA go-VOLITIONAL
        Let's go by train from Tokyo Station to Tsukuba.


             が ga                                                                     に ni
           Nominative                                                              Translative
        PATIENT LOCATUM                             から kara                           GOAL
             AGENT                                  Ablative


        The basic case structure of iku is identical to that of kawaru, to the extent that the
two Graphs could be placed one upon the other. The labelling of semantic roles is quite
similar too. The only difference is that the moving subject of iku is agentive – in other
words, is an animate entity that changes place volitionally rather than spontaneously, as it
was the case with the change of state described by kawaru. Hence this participant bears
both the roles of AGENT and of LOCATUM. In a sense, one could say that this entity moves
herself from one location to another. This makes of the verbs of movement a sort of
"reflexive" verbs, as it has been pointed out by several scholars.21 All this means that the
verbs of change of state and those of change of place basically describe events of the
same class, in that they both describe the displacing of an entity from one location to
another. They only differ in that iku and the other verbs of movement describe a concrete
movement between two physical places, whereas the verbs of the kawaru class deal with
a metaphorical movement between two abstract places, or conditions. 22 One can see how
much the use role labels in conjunction with a Localist Approach is of help here. Such an
event metatype can be defined as "moving between points", and the two classes of
predicates can be classified into a larger class or prototype of "translative verbs".

      The term translative
      It is time now to clarify the reason why I decided to use the term "translative" to
name the case assigned to the best part of the GOAL phrases I have been discussing till

  See Jacobsen (1992) for a consistent treatment of the reflexivity of "intentional intransitive verbs".
  The other difference between the two predicates, that concerning the animacy of the S
subject, comes from this: a physical movement such as that described by iku allows for, or even requires, an
animate LOCATUM acting volitionally on herself; while a change of state cannot ultimately be controlled by the
entity undergoing it, hence the possibility for both naru and kawaru of expressing an inanimate entity as the
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 44 -
now. As I pointed out during my first lecture, the notion of GOAL is well expressed by the
particle ni, which is often used in alternation with e to specify the endpoint of a movement.
All dynamic occurrences of ni are very much alike, and I felt the need of having one single
label that could cover all or most of them. In order to name the case marked by these
particles I had a wide choice of terms: dative, dynamic, directional, mutative and the labels
of the dynamic local cases of Finnish allative, illative and translative.
        First of all I decided that I wanted to distinguish between the case assigned to the
animate, agentive, and sometimes benefactive GOAL occurring with the verbs of giving
from all the other kinds of goals. This is because I felt this goal has a very special status.
For one thing, with the verbs of giving, which entail a change of possession, this goal also
becomes a new possessor, so that is somehow affected by the event. Moreover, such
GOAL is also the AGENT and subject of the conversive predicates of receiving. Therefore, I
decided that this case had to go by the traditional label of dative.
        Then, how to name the other instances of ni and e, when these particles specify the
inanimate endpoint of a physical or abstract movement? I discarded dynamic because this
term is better used in relation to the more general distinction between dynamic and static
cases. I also discarded directional, because this term suggests a mere movement in some
direction, and not the reaching of a destination, as the concept of goal implies. And I
discarded mutative, because it makes reference more to a change than to a movement.
Indeed, even the main user of this term, the grammatist Martin,23 has to make a distinction
between mutative and locative mutative, in order to distinguish a change of state from a
change of physical place. I was left then with labels used to name three related cases of
the Finnish language: illative, allative and translative. Now, I do not hold not very much
with the old distinction between an "illative" ni and an "allative" e, since there is evidence
that a distinction between them was blurred even in the speech of Tokyo of the beginning
of the of the 20th Century.24 Since both labels are equally out of the mark, , I could use
either of them, illative or allative, but eventually I decided to discard both, because their
meaning is too specific. Well, only translative was left. I am aware of the fact that in
Finnish it denotes only a change of state. But I particularly like this term because, at least
to me and my fellow Italian language native speakers, it evokes both the image of a
transformation (between conditions) and of a transfer (between places). It also contains
the root lat, which means "to carry" in Latin. And so, translative has been. If Finnish native
speakers feel confused by this improper use of the denomination of a case of their own
language, though, I recommend them the use of the most "neutral" label mutative.

Transitive predicates of change of state
        Let me now make a step back, returning to naru and kawaru. Both these verbs
describe the movement of one entity between two points. This displaced entity is the main
participant in the event, and is realized as the subject. In other words, naru and kawaru are
intransitive. However, they both have a transitive companion that describes the same type
of event (the abstract displacing of a PATIENT) as the effect of the action of a new entity that
was not mentioned in the clauses of either kawaru or naru. This entity causes the change
by acting volitionally and in control, and is realized as the subject.

       Balloon graph of the transitive predicate of change of state kaeru

  Martin 1975: 41.
  This was discussed in an unpublished paper read at the Second Conference on Japanese Language and
Language Teaching held in Naples in 2002. The paper was a report about the discovery of metal drums
containing the recordings of Japanese rakugo dating from the 1906 Paris Expo. Some of the rakugo tellers
were born in Edo during the Tokugawa Era. This find attests to the use of e in all dynamic instances, even
when contemporary Japanese demands ni.
                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
                                                        - 45 -
      Musuko-ga kamigata-o chōhatsu-kara tanpatsu-ni kae-ta.
      son-NOM hairstyle-ACC long hair-ABL short hair-TRA change-PAST
      My son changed his hairstyle from long to short.


                                                                       から kara    に/へ ni/e
        が ga                                                           Ablative   Translative
      Nominative                                                         SOURCE      GOAL
        AGENT           Accusative

        This Balloon Graph shows the role and case structure of kaeru, the transitive
companion of kawaru. One can see that this graph represents an event structure very
similar to that of an ordinary verb of change of place such as okuru. The differences are
important, however. First of all, kaeru requires that the AGENT be an animate entity. Then
the meaning root of kaeru implies that the "movement" happens in an abstract space,
rather than in real space, and that the two roles that represent its path refer to two
subsequent conditions (for instance, two qualities) of the PATIENT.
        Kawaru and kaeru seem to be morphologically related, and indeed they are. But
there is no productive rule to derive one from the other, and both verbs are independent
lexical items – which means they are listed in the dictionary as independent words.

      Balloon graph of the transitive predicate of change of state suru

      Musuko-o isha-ni suru.
      son-NOM doctor-TRA make
      I will make my son a medical doctor.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
                                                       - 46 -
                                                  do, make

                 が ga
               Nominative                            をo                                  に ni
                 PATIENT                          Accusative                          Translative
                                                    PATIENT                              GOAL

                                                                                     な Adj.: に ni

                                                                                    い Adj.: く ku

                                                                                   Clause v. B4 ように

        This is suru, the transitive companion of naru. Suru is morphologically unrelated to
naru. Even if the "Balloon structure" of suru is quite similar to that of kaeru, I have build
this Graph using the Balloon Graph of naru as a template, in order to better show the
similarities between the case structures of two predicates.
        In the Balloon Graph of suru, the balloons of naru are shifted one position to the
right, to leave room for a new subject, the AGENT causing the transformation, which comes
now on stage. The entity undergoing the change, the PATIENT, is therefore moved to the
middle. As the object affected by the agent's action, such PATIENT is marked by the
accusative case. The stack of the several realizations of the goal phrase are basically
unchanged, but shifted to the extreme right of the graph. This does not mean that these
phrases are adjuncts, though. Rather, they are absolutely necessary when suru describes
a change of state, just as it happens with naru. One cannot say, for example *musuko-o
suru "I'll make my son", but has to add the GOAL phrase specifying what she wants to
make her son into – for example, a medical doctor: isha-ni.

  Actually, there is another companion of naru, the verb nasu, which is morphologically related to naru just
as kaeru is related to kawaru. Again, there is no productive rule to produce nasu from naru or vice-versa,
and the two verbs are independent lexical items. I decided to displayed only suru in my graphs because
nasu is used much less frequently and with a more restricted case structure than suru.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
                                                         - 47 -
        Anyway, the only difference between the goal phrases of naru and suru is that suru
can assign to nouns only a translative case marked by ni, not by e. It should be noticed
that, just like with naru, the (no)yōni complementizer can be assigned to any type of clause,
as it often happens, not only to stative phrases. Finally, I must add that this right shifting of
phrases causes the kara ablative phrase to be pushed out of the Graph. (This is an
unforeseen, "emergent property" of my Balloon Graph). Indeed, this phrase cannot occur
in the case structure of suru.

       The relation between the transitive and intransitive companions of these pairs of
predicates can be easily captured and represented by means of a Table Graph. The first
Table I am showing displays kawaru and kaeru.26

        Table Graph of the intransitive/transitive predicate pair of change of state kawaru and naru

     Predicate pair of                 変わる                               変える
     change of state                   kawaru                             kaeru

     AGENT animate                                                 nominative が ga

     PATIENT                      nominative が ga                    accusative を o

     SOURCE inanimate            ablative から kara                  ablative から kara

     GOAL inanimate            translative に・へ ni/e             translative に・へ ni/e

     FORCE                       instrumental で de
                                  によって niyotte

       There is nothing particular to discuss about this Table, save that the phrases
marked by the transitive markers ni and e would actually need two separate columns. But
since these markers have the same meaning, grouping the two phrases together in a
same box is a good idea to save space. The same is true for the two instrumental phrases,
which I placed at the bottom of the grid.

   In a Table Graph it does not really matter I which order the case structures of the intransitive and transitive
predicates are displayed. There is no semantic, syntactic, logical or morphological precedence of one
predicate over the other. The linguist John Lyons (1968:Ch. 8.2) suggested that the event representation of
intransitive events takes the precedence in both the human mind and the grammar over the representation of
the more structured transitive events. Its is tempting to follow this approach, but unfortunately things happen
to seem quite the opposite: it is transitive events that seem to have some cognitive preeminence on
intransitive ones. Since this have not direct bearing on my approach, one can put the columns of the
transitive and intransitive companions in the order she deems more didactically appropriate.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
                                                          - 48 -
       The Table Graph displaying the case structure of suru and naru is not so problematic, but very space-consuming.

       Table Graph of the intransitive/transitive predicate pair of change of state naru and suru

  Predicate pair of          なる               なる               なる                なる                 する             する               する             する
  change of state             naru            naru             naru              naru               suru           suru             suru            suru

   AGENT                                                                                        nominative      nominative       nominative      nominative
                                                                                                  が ga            が ga             が ga            が ga

   PATIENT                 nominative      nominative       nominative       nominative          accusative     accusative       accusative      accusative
                             が ga            が ga             が ga             が ga                 をo             をo               をo              をo

   SOURCE                   ablative         ablative        ablative          ablative
                           から kara         から kara          から kara           から kara

   GOAL                    translative      adverbial        adverbial          COMP             translative     adverbial        adverbial        COMP
                              に ni            に ni            く ku            clause て              に ni           に ni            く ku          clause て

       different lexical     (noun)      ( な adjective)   ( い adjective)        (verb)             (noun)      ( な adjective)   ( い adjective)     (verb)
   categories of GOALS

        The best thing to do would be stacking the Tables Graph of suru and naru on the two planes of a 3D display, so that the role
and phrase boxes of the two predicates would be imposed over each other. But this is impossible, and so the Table must spread like
this in two dimensions. This Graph is so large because the very logic of the Table Graph demands that the several goal phrases of the
two predicates should be plotted into different columns, since they cannot cooccur.

                                                             Dalla Chiesa -The Semantics of Japanese verbs
                                                                        Helsinki 4th-7th May 2009
Transitive/unaccusative predicate pairs
       As I have already said, the predicates of the two pairs kawaru/kaeru and naru/nasu
describe a same type of event, a change of state, but in two different ways. The
intransitive verb has only one argument, which corresponds to the PATIENT entity
undergoing the change. Its transitive companion has two arguments: the PATIENT has been
"demoted" to direct object, and a new participant comes on stage. It is the AGENT causing
the change, which replaces the PATIENT as the subject of the sentence.
       Those are not the only pairs of Japanese predicates that exhibit such a particular
relationship. Actually, the Japanese language is very rich of pairs of this kind. They are the
so called transitive/intransitive pairs, which should be better named transitive /
unaccusative pairs.

      Balloon Graph of the transitive/unaccusative pair aku/akeru

       Mado-ga kaze-de ai-ta.
       window-NOM wind-INS open-PAST
       The window opened because of the wind

       Mado-ga kaze-niyotte ai-ta.
       window-NOM wind-INS open-PAST
       The window opened because of the wind

       Kodomo-ga mado-o ake-ta.
       child-NOM window-ACC open-PAST
       A child opened the window

                     あく                                                      あける
                     aku                                                     akeru

        が ga                    によって                            が ga                    をo
      Nominative                niyotte                       Nominative             Accusative
        PATIENT               Instrumental                       AGENT                PATIENT

                                 で de

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
        These Balloon Graphs show case structures very similar to those of kaeru and
kawaru. Of course there are no GOAL or SOURCE roles.
        This is how I represent the event described by aku and akeru. Aku refers to an
event of "spontaneous" change in which the shutter of a window moves until a space
appears between the shutter itself and the frame. We can describe the same event in
English with the verb open. So The window opens. But the window is inanimate, and
cannot move by itself. Therefore, there must be a force that acts on the window to move it.
This entity could be the wind, an engine of some kind – so that the event is considered
spontaneous, i.e., not volitionally caused – or the hand of a child – which means an
animate AGENT. However, aku does not need a phrase referring to such causing entity in
order to thoroughly express the event. The single argument required by aku is that
referring to the thing that moves, the PATIENT, and is realized as the subject by means of
the nominative. The only way that a mention of the causer can be added to the clause of
aku is by means of an adjunct, marked by one of the two instrumental markers de or
niyotte, but only if such entity is an inanimate FORCE. Here, there is no difference in
meaning or use between de and niyotte.
        The lexicon, however, provides the speaker with another verb to describe the
opening of the window, a verb that makes a specific reference to the actor causing the
event to happen. This verb is akeru, and requires two arguments: a phrase referring to the
AGENT herself, which is realized as the subject and marked by the nominative, and a
phrase that refers to the window, the PATIENT, which is now demoted from subject to direct
Object and marked by the accusative case. This is the Table Graph that captures such

       Table Graph of the transitive/unaccusative pair aku/akeru

unaccusative / transitive       開く aku                    開ける akeru
    predicate pair                "open"                      "open"
   AGENT animate                                        nominative が ga

   PATIENT                  nominative が ga              accusative を o

   FORCE inanimate                 で de
                            によって niyotte

        The particular relation exhibited by aku and akeru can be analysed in several ways.
For example, the fact that the only argument of aku refers to a PATIENT that in several
syntactic instances behaves (specifically, leaves a trace in deep structure) as the object of
a transitive predicate is one of the reasons these particular intransitive verbs are named
unaccusatives or, incorrectly, "ergative". Conversely, that fact a transitive verb such as
akeru actually describes the act of causing something to move is the reason why the
transitive verbs of these pairs are also called lexical causatives. (The adjective lexical is
added in order to distinguish them from the morphological causatives, which can derived
by means of a productive rule from all ordinary verbs).

       Aku and akeru are independent lexical items, but are phonologically related.
However, not all the pairs of unaccusative / transitive verbs exhibit such phonological
relation. Here is an example I like very much.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
          Table Graph of the transitive/unaccusative pair korosu/shinu

          Mufāsa-ga byōki-de shinda.
          Mufasa-NOM disease-INS die-PAST
          Mufasa died of disease.

          Sukā-ga Mufāsa-o koroshita.
          Skar-NOM Mufasa-ACC kill-PAST
          Skar killed Mufasa.

unaccusative / transitive          死ぬ shinu                    殺す korosu
    predicate pair                    "die"                         "kill"
   AGENT animate                                            nominative が ga

      PATIENT                   nominative が ga              accusative を o

      FORCE inanimate          instrumental で de

       Mufasa died, and that might be believed to have been a spontaneous event,
perhaps with some mechanical, immediate cause such as a fall from a cliff or a disease.
This event is described by the intransitive clause. But one might discover that actually the
death of Mufasa was not accidental, but rather volitionally caused and controlled by some
agent, like his brother Skar. This eventuality is described by the transitive verb. Shinu and
korosu are completely unrelated morphologically, but their relation of meaning, and the
relation between the events they describe, is the same as that between aku and akeru.27
       This is the Table Graph displaying the argument and case structure of the
predicates pairs of this kind.

          Table Graph of the case structure of transitive/unaccusative predicate pairs

unaccusative / transitive          intransitive                  transitive
    predicate pairs               unaccusative              (lexical causative)

      AGENT                                                 nominative が ga

      PATIENT                   nominative が ga              accusative を o

       I believe this Table Graph manages to show the relation of meaning and the
difference in syntax and case structure between unaccusative and transitive (or causative)
predicate pairs better than any other graphic device.
       So even if one considers aku and akeru to describe a same event, they only
describe the same type of event (the opening of something), not necessarily one same,
     See however Fodor 1970.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
specific event, or event token. Indeed, while aku makes no mention of the ACTOR, which
can be anything or anybody, akeru does, and specifies a well defined entity as the ACTOR.
This means that the transitive clause kodomo-ga mado-o akeru entails the opening of a
window described by mado-ga aku, but that this latter clause, mado-ga aku, does not
entail the acting of any specific AGENT, as in kodomo-ga mado-o akeru. The two clauses
have a different truth-value, aku includes akeru, and the two predicates are not conversive.
        It is too bad that the Japanese language has no morphologic adjective meaning
"open". (The function of the missing adjective is taken by the resultative forms ai-teiru and
ake-tearu respectively). However, the predicate pair takamaru and takameru actually does
have a parent adjective, takai. I want to show now another Table, plotting this time the set
of predicates related to the adjective takai.

       Predicates of the takai family

       高い takai (high)
            Kono-seihin-wa hinshitsu-ga takai.
            this-product-TOP quality-NOM high
            The quality of this products is high.

       高まる takamaru (raise, improve INTR)
          Kono-seihin-wa hinshitsu ga takamat-ta.
          this-product-TOP quality-NOM improve-PAST
          The quality of this product has improved.

       高める takameru (rise, improve TRANS)
          Mēkā-ga kono-seihin-no hinshitsu-o takameru.
          maker-NOM this-product-GEN quality-ACC improve
          The company will improve the quality of this product.

        The adjective takai means "high". The verb takamaru is an intransitive verb that
means "raise", "increase" or "becoming high". The verb takameru is transitive, and means
"rise", "increase" or "making high". These two verbs are clearly derived from the adjective
takai, but not by a productive rule. This means that the verbs derived from adjectives in
this fashion are fixed by the lexicon and there is no morphological rule that allows a
speaker to freely produce such a verb from a given adjective. Takamaru and takamaru are
in exactly the same relationship as, for instance, aku and akeru, as the following simple
Table is showing.

                                  高まる                          高める
unaccusative / transitive        takamaru                     takameru
    predicate pair               "increase"                   "increase"

   AGENT                                                  nominative が ga

   PATIENT                   nominative が ga               accusative を o

      This is what happens when one wants to plot together in one single table the three
predicates of the takai family.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
       Table Graph of the predicate family takai

                   Adjective                                  Verb                                                                  Verb
                                                          INTRANSITIVE                                                           TRANSITIVE
                       -i                      -maru                         -ku naru                              -meru                         -ku suru

                     高い                       高まる                          高くなる                                   高める                           高くする
                     takai                   takamaru                      takaku naru                            takameru                      takaku suru

  AGENT                                                                                                        nominative が ga                nominative が ga

  THEME         nominative が ga

  PATIENT                                nominative が ga                nominative が ga                        accusative を o                 accusative を o

  GOAL                                       Root taka-            高 taka + adverbial く ku                       Root taka-              高 taka + adverbial く ku
   Aspectual         state               BECOME+state =                BECOME+state =                   CAUSE+BECOME+state =            CAUSE+BECOME+state =
   properties                              achievement                   achievement                     CAUSE+ achievement =            CAUSE+ achievement =
of predicates                                                                                               accomplishment                  accomplishment

       The constructions of the three predicates are displayed in a succession of columns, from left to right. The first column on the left
contains the construction of the stative predicate takai, which means "high". My analysis will start from this column. From the point of
view of lexicon and morphology, takai is an adjective. It does not matter. This predicate describe a state (a "stative eventuality"), and
requires only one argument, that referring to the entity being in the state takai. This entity does not act at all, and indeed is not
considered to be an AGENT, but rather a THEME. I define a THEME is "an entity located in a place or condition that does not act and does
not undergo a change of state".

         THEME               An entity located in a place or condition that does not act nor undergo a
                             change of state. This narrow definition is only possible because I gave a
                             detailed definition of AGENT, EXPERIENCER and PATIENT.

                                                               Dalla Chiesa -The Semantics of Japanese verbs
                                                                          Helsinki 4th-7th May 2009
        The column on the left of that of the adjective takai displays the construction of the
related predicate takamaru. Takamaru describes the event that initiates the state takai,
and can be translated as "becoming high". It is an intransitive event involving only one
participant: the entity undergoing the change. As such, predicate takamaru requires only
one argument, again realized as the subject by means of the nominative case. This
participant is the same entity as the predicate takai describes as being in the condition
"high", but since it undergoes a change it now bears the role of PATIENT rather that of
        But whenever a change happens, there is a path, namely a starting condition and
an endpoint. The endpoint reached by the PATIENT is the state "takai". There is no need to
specify such final state by means of an "independent" phrase, because it is already
expressed by the predicate itself. In other terms, the predicate takamaru communicates
the GOAL of the change (the condition "high") by means of its meaning – and, by the way,
through its phonological "shape". I indicate this fact by referring to the GOAL as to "Root
taka". By root I mean the meaning root of the verb, which corresponds to the condition
expressed by the adjective takai. As it happens, however, taka is also the morphological
root of the adjective taka-i, to which a number of flexional morphemes can be affixed.
        Let us skip the third column, that of takaku naru, for the moment. The fourth column
displays the case structure of predicate takameru. This predicate describes the same type
of event as takamaru, "becoming high", but adding the intervention of a completely new
participant, an external AGENT causing the change. Takamaru and takameru are then a
transitive causative predicate and an intransitive unaccusative predicate respectively, and
exhibit the same relation as, say, akeru and aku, korosu and shinu and so on. The subject
marked by the nominative case is now the AGENT, and the PATIENT becomes the direct
object, marked by the accusative. Again, in a takameru clause there is no syntactic
specification of a GOAL, because the GOAL is entailed by the verb's root.
        Now it is time to take into account the two remaining columns. The dynamic event
"becoming high" expressed by takamaru can be alternatively described by using the
adjective taka-i in conjunction with the verb naru "became", which I have already
discussed. This instance is displayed in the third column. You will notice that the role and
case assignment do not change with respect to those of takamaru, but now the GOAL role
is represented by means of the independent phrase filled by the adjective taka-i in its
adverbial form taka-ku.
        These considerations apply to the fifth column as well, which contains the
construction taka-ku suru. Here the transitive event of "making high" or "increasing" is
described by means of the verb suru, which, as I have discussed earlier, is the transitive
companion of naru. Again, role and case assignments are the same as takameru, with the
exception that the GOAL role is now realized by means of the adverbial taka-ku.

        My analysis of the relation between taka-i and its two daughter verbs could end
here. But having drawn such a large Table Graph, I think it would be a pity if I did not
discuss the aspectual relation between those predicates as well. The Table Graph allows
us to add any number of boxes in any position (to a certain extent, of course), and I
exploited this opportunity by adding at its bottom a box row that specifies the aspectual
properties of the three predicates I analyzed.
        Taka-i is an adjective, and obviously describes a STATE. The intransitive,
unaccusative verbs takamaru and naru describe the dynamic event that ends in the state
taka-i. Such an event can be represented as being formed by a change-plus-a-state, or as
BECOME+STATE. The events composed in this way (and the verbs that describe them) are
defined as telic (which means that they have an endpoint) and classified under the label

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
achievements. The event described by the transitive, causative verbs takameru and suru
adds a further subevent to the achievement "becoming high". It is the causative act that
actually instigates the change. These predicates and the events they describe can
therefore be represented as CAUSE+BECOME+STATE, or as CAUSE+ACHIEVEMENT. The events
composed in this way are also telic, because they have the same endpoint as the
achievements. They, and the verbs that describe them, are classified as accomplishments.
         This classification of events and predicates was originally proposed by linguist Zeno
Vendler,28 and is now widespread and commonly applied to the Japanese language as
well. There is no originality in the use of it I make in my Graphs. Only the idea of integrate
it into a Table Graph and using as an aid for lecturing in Japanese grammar is "new", if
you want.

Predicates of perception
       Before proceeding to discuss stative predicates and alternations, let me finish my
treatment of transitive verbs by showing you a number of Graphs displaying the behavior
of the cluster of predicates of perception centered on miru, "watch".

        Balloon Graph of the transitive predicate of volitional perception miru

        Masao-ga Iwatesan-o Hachimantai-kara mi-ta.
        Masao-NOM Mt.Iwate-ACC Hachiman Plateau-ABL watch-PAST
        Masao watched Mt.Iwate from the Hachiman Plateau.


                    が ga                           をo                               から kara
                  Nominative                    Accusative                          Ablative
                    AGENT                           THEME                            PLACE

       Miru is a verb of volitional perception, and it means " watch", not "see". In the
example above, Masao was not watching the sky, or the landscape, and he just happened
to get a sight of Mount Iwate in his field of vision. Therefore, the example does not mean
"Masao saw Mount Iwate". Rather, Masao was directing his full attention to Mount Iwate
and was looking at it, maybe with a pair of field glasses, in full control of the event.
       Miru takes the ordinary accusative case pattern, with some oddity. Its first argument
represents the AGENT that volitionally performs the act of watching, and is realized as the
subject by means of the nominative case marker. This entity also bears the role of
EXPERIENCER of the perception. It is the first time I mention this role, which I define as the
animate perceiver in events of sensorial perception or feeling.

  Zeno Vendler, "Verbs and Time", in Z. Vendler (ed.), Linguistics in Philosophy, Ithaca, Cornel University
Press, 1967.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
        EXPERIENCER      An animate entity represented as undergoing an experience of perception or

       The second argument refers to the watched entity. This is not a PATIENT, because is
not affected in any way, but rather a THEME. It is realized by the accusative case. As I have
already said, the use of the accusative particle o entails a certain degree of control on the
part of the subject. Here the control is not directed towards the object, which is not affected
and can also be very far, but over the event of watching itself.
       This accounts for the ordinary accusative case structure of miru. Miru, however, can
take an ablative phrase as an adjunct, which is not common with transitive two-place
predicates of this kind. It is not possible to consider this phrase as realizing the role of
SOURCE, as the ablative case marker would seem to suggest, since no movement initiates
from the location marked by kara. Rather, this ablative phrase specifies the location where
the act of perception takes place, and realizes the role PLACE.

        PLACE            The physical location where an event takes place. Often LOCATION; here
                         PLACE to distinguish it from the grammatical case.

        Ordinarily this role would be marked by the locative de, and indeed the place of
action is commonly specified in this fashion, saying, for example, that Masao watches TV
in his room:

       Masao-ga jibun-no heya-de terebi-o mi-teiru.
       Masao-NOM self-GEN room-LOC TV-ACC watch-EXTENSIONAL
       Masao is watching TV in his room.

        Why an ablative case, then? This phrase obviously represents the point of origin of
a line of sight connecting the location where the act of perception takes place with its
object of perception (here, the THEME). The odd fact is that this ablative adjunct occurs with
all the predicates of the miru group.

     Now for mieru, the intransitive companion of miru. It is not so easy to convey its
meaning with just one word. It means something like "be visible" or "be seen".

      Balloon Graph of the intransitive predicate of spontaneous perception mieru

       Hachimantai-kara aozora-ni Iwatesan-ga mie-ta.
       Hachiman Plateau-ABL blue sky-LOC Mt.Iwate-NOM be visible-PAST
       From the Hachiman Plateau Mt.Iwate was visible in the blue sky.
       From the Hachiman Plateau I could see Mt.Iwate in the blue sky.

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009

                                                                                 から kara
                 が ga                                                                 PLACE
                  FORCE                                                    に ni
                (STIMULUS)                                                Locative

                                                             に ni

        Mieru is one of a fairly large group of predicates of spontaneous, involuntary
perception. Like its companions, it is a one-place, intransitive predicate. Its single
argument represents the perceived entity. In the event described by miru, the perceived
entity is bearing the role THEME. Now it rather bears the role of STIMULUS, which I consider
to be a variation of the role FORCE. I represents the event as follows: The speaker (maybe
Masao) was on the Hachiman Plateau; he just happened to glance at the blue sky, and got
the sight of the perfect cone of Mount Iwate. With mieru, then, the subject entity is not a
participant that affects an experiencer by acting volitionally; it does not even need to be
animate. Rather, it initiates the event spontaneously (we can also say casually), by getting
into the visual perception field of some animate entity, activating its organs of perception,
causing an involuntary perception, and so becoming the unaffected object of it. In short, it
stimulates an event of "seeing". It is important to notice that when this happens, the
EXPERIENCER "sees" the stimulus entity, but does not necessarily "watches" or "looks at" it.
Masao did not necessarily stare at Mount Iwate, after getting sight of it. In other words, a
mieru event does not entails a corresponding miru event. Miru and mieru are therefore in a
relation quite similar to that of the paired transitive and unaccusative predicates discussed
early on.
        Usually, the perceiver or EXPERIENCER is not mentioned in a mieru clause, because
it is understood to be the speaker. When the need arises, however, it can be specified by
means of an adjunct phrase marked by the dative case ni . (In the Table above, I display
this phrase in a dashed balloon.) The dative marking is an instance the so called dative of
perception, well attested crosslinguistically. It is such a reference to an EXPERIENCER that
allows a mieru clause to fully express the meaning of "seeing something".
        There can be two other particular adjuncts in a mieru sentence. One specifies the
background field of vision, realized by means of a static locative. The other is an ablative
adjunct specifying the origin of the line of sight.

      This basic meaning and case structure of mieru are strictly related to its use as a
verb meaning "look like, seem".

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
Balloon Graph of the stative predicate of subjective perception mieru

Masao-ni-wa Hachimantai-kara Iwatesan-ga Fujisan-ni mieru.
Masao-DAT-TOP Hachiman Plateau-ABL Mt.Iwate-NOM Mt.Fuji-TRA seem
To Masao, from the Hachiman Plateau Mt. Iwate looks like Mt. Fuji.

Hachimantai-kara Iwatesan-ga taka-ku mieru.
Hachiman Plateau-ABL Mt.Iwate-NOM high-ADVERB seem
From the Hachiman Plateau Mt. Iwate looks high.

Hachimantai-kara Iwatesan-ga chikaku-ni mieru.
Hachiman Plateau-ABL Mt.Fuji-NOM vicinity-TRA seem
From the Hachiman Plateau Mt.Iwate looks close / in the vicinity.

Hachimantai-kara Iwatesan-ga [sora-o tsuku-yōni] mieru.
Hachiman Plateau-ABL Mt.Iwate-NOM [sky-ACC pierce-COMP] seem
From the Hachiman Plateau Mt.Iwate looks like it's piercing the sky.

                           Dalla Chiesa -The Semantics of Japanese verbs
                                      Helsinki 4th-7th May 2009

                                                                                      から kara
                                     に ni                                             Ablative
        が ga                      Translative                                          PLACE
      Nominative                       GOAL
                                                                             に ni
                               な Adj.: に ni

                               い Adj.: く ku

                             Clause v. B4 ように

       One can easily notice that the Balloon Graph of mieru is nearly identical to that of
naru. The reason is quite simple: here verb mieru does not describe an event of perception,
but rather an abstract change that takes place in the mind of the experiencer at the
moment of the perception. To follow the "story" of Masao on the Hachiman Plateau:
Masao is there, turns his head gazing at the horizon, and then Mount Iwate gets into his
perception field and Masao gets sight of it. This in turn starts a cognitive process: in the
mind of Masao, because of its shape, colors, whatever, Mount Iwate takes a new quality –
new because unnoticed or cognitively unassigned before – or abstractly changes into a
different but similar thing. Such new quality can be expressed by the adjective "high", the
noun meaning "close place", the capability of "piercing the sky", or even the shape of
Mount Fuji.
       All the new conditions subjectively taken by the perceived THEME are listed in the
tower of stacked balloons in the center of the Graph. Just as for naru, the ovals are
stacked according to the lexical category of the item they contain. The phrases displayed
in these balloons are all GOAL phrases, and, where applicable, their case is the translative.
       This verb mieru differs from that meaning "be visible" because is a two-place
predicate, just like naru, and requires two arguments. The subject realizes the perceived

                             Dalla Chiesa -The Semantics of Japanese verbs
                                        Helsinki 4th-7th May 2009
entity, which is a THEME but also a PATIENT, since it undergoes a change in the mind of the
experiencer. The second argument is an oblique, the goal phrase discussed before. But
since mieru describes an event of abstract, subjective change, it entails the presence of an
EXPERIENCER. Again, the EXPERIENCER is taken, by default, to be the speaker, but can be
specified by means of an adjunct marked by a dative of perception. Finally, since mieru
belongs to the miru family, it constructions can include an ablative adjunct indicating the
point of origin of the line of sight.
        If the need arises, this is how I discuss the occurrence of the particle wa assigned to
Masao-ni in the first example above. Wa can be a topic or a contrastive marker.
Apparently, this wa is contrastive, contrasting this sentence with other, unexpressed,
"virtual" sentences saying different things about the same event type. For example, wa
might suggests that for another person, maybe Masao's friend Keiko, Mount Iwate would
look completely different when seen from the Hachiman Plateau. But why the need for a
contrastive marker here? My personal opinion is that there might a problem of heaviness
here: the clause is too long and there is too much material in it to easily accommodate one
more phrase. To find some room, the added material must be explicitly marked as being
secondary, quite apart from the focus of the clause. Hence the topicalization of the
        Next I am going to discuss, in the quickest possible way, the case structure of the
third and last member of the miru family, the verb miseru " show".
      Balloon Graph of dative predicate miseru

       Keiko-ga Hachimantai-kara Masao-ni Iwatesan-o mise-ta.
       Keiko-NOM Hachiman Plateau-ABL Masao-DAT Mt.Iwate-ACC show-PAST
       Keiko showed Mt.Iwate to Masao from the Hachiman Plateau.


                                                                                から kara
        が ga                                                                    Ablative
      Nominative                                                                 PLACE
         AGENT         Accusative                に ni
        SOURCE           PATIENT

       The case structure of miseru very closely corresponds to that of the predicates of
giving. Indeed, miseru is not a true causative verb meaning "make somebody see", as it is
often taken to be. This mistake arises from the phonological form of miseru, which
resembles that of the morphological causative, which would be misaseru or misasu. On

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
the contrary, miseru is more a verb of giving, whose transferred entity is visual
information.29 Since I have already treated the verbs of giving, I will not discuss this
Balloon Graph any further. But, just notice the ablative adjunct, popping up once more in a
disquieting way.
       The following is a Table Graph plotting together all the different constructions of the
several predicates related to miru.

         Table Graph of the predicates of visual perception (1)

                            action (durative)    spontaneous event (punctual)          condition (stative)
Predicates of visual           (transitive)              (intransitive)                  (intransitive)
        (1)                      見る                           見える                           見える
                                 miru                        mieru                            mieru
                              "watch, look         "be seen, see" "be visible"          "look like, seem"
     AGENT                 nominative が ga
     EXPERIENCER                                           dative に ni                    dative に ni
     THEME                  accusative を o                                              nominative が ga
     FORCE                                              nominative が ga
     GOAL                                                                               translative に ni

       different lexical                                                                adverbial く ku
             categories                                                                  adverbial に ni
                                                                                       COMP clause て te
     PLACE                 ablative から kara            ablative から kara                ablative から kara

         In the Table above, the perceiver entity bears the roles of AGENT and EXPERIENCER,
It is the Masao of the examples. The perceived entity, THEME or FORCE, is Mnt. Iwate.
These two participants stand in a common accusative relation in the clause of miru, with
the agentive EXPERIENCER taking the relation of subject. With mieru, it is rather the THEME
that is realized as the subject, and in the few, limited instances in which the EXPERIENCER is
to be mentioned, it is marked by a dative phrase.
         The question now arises of how to classify this group of predicates from an
aspectual point of view. It is not really necessary to do so, in fact, but since I have briefly
discussed the aspectual facets of other similar predicates I think it could be interesting to
do it here again. Miru is obviously a transitive verb of action. As for mieru, its first instance
describes a spontaneous punctual event. Its resulting state can be expressed by the
extensive –teiru form, taking the meaning "be visible", or "be in sight". The second
instance of mieru describes instead the property of the observed thing, the FORCE (or
STIMULUS), of looking like something else or of taking a new subjective quality. Now the
predicate is stative.

         The following is the Table Graph of the third predicate of this series, miseru.

   It is not a simple verb of giving, though. Since its causes the EXPERIENCER to experience, is belongs to the
verbs of hatarakikake. These are verbs that describe the act of exerting a certain amount of pressure on an
animate entity in order to cause some event to happen. Quite consistently, the recipient of this pressure is
realized by means of a ni phrase. Again, it can be discussed whether the marking of such oblique phrase is
to be considered to a dative or an agentive case.
                                       Dalla Chiesa -The Semantics of Japanese verbs
                                                  Helsinki 4th-7th May 2009
       Table Graph of the predicates of visual perception (2)

                              action (durative)
  Predicates of visual          (ditransitive)
          (2)                     見せる
                               miseru "show"
  AGENT                       nominative が ga
  THEME                        accusative を o
  GOAL                           dative に ni

  PLACE                      ablative から kara

        Miseru requires a separate Table because it would be too complex to integrate its
participants and role structure with those of miru and miseru. I consider miseru to be verb
of giving assigning dative case even if its companion predicate miru takes no agentive ni
phrase, as the other verbs of receiving do. For example, one cannot say something * Kare-
ni mita, *"I saw (it) from him", whereas one can use in this way the other predicate of
voluntary perception kiku "listen to" and say Kare-ni kiita "I heard (it) from him". I also
exclude from the Table the other role of EXPERIENCER borne by the goal entity, since it finds
no expression in the surface case structure. No further comment on this Table but this:
that ablative adjunct is still there.

      The stative predicate aru "be"
      It is time now to move on to a stranger and more controversial ground. This is the
ground on which I believe my approach is more original and powerful. I am introducing
now the several uses of stative predicates aru and iru.

       Balloon Graph of locative aru

       Niwa-ni ki-ga aru.
       Garden-LOC tree-NOM be
       There are trees in the garden.

       Shiminkaikan-de konsāto-ga aru.
       Community Hall-LOC concert-NOM be
       There is a concert in the Community Hall.

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009

          が ga
        Nominative                                        に ni
                                                     Static locative

                                                       で de
                                                   Dynamic locative

       This Balloon Graph displays the most common use of aru, that as a locative verb.
Locative aru expresses a location eventuality, in other words, specifies the position in
space of things or events. It requires two arguments. One refers to the entity that is
described as being located somewhere, bearing the role THEME. This argument is realized
as the subject, and its phrase is marked by the nominative particle ga. The other argument
represents the location where the THEME is. It obviously bears the role PLACE. There are
two ways of realizing this PLACE phrase. The first one is a ni static locative. It is compulsory
when the THEME is a physical object. All students of Japanese become accustomed to this
case structure quite early in their study lives. The second way is by means of a de
dynamic locative. Such locative is to be used when the THEME is an event. This is a slightly
less usual construction, but still quite ordinary. I know of only few textbooks that introduce
this pattern during their Basic language course. It should be noticed that these two
locatives are probably the purest example of local cases, but they do not mark
"background functions" as I have named them. Rather, they mark true arguments,
because their phrases are required in order to thoroughly describe an eventuality of
       A similar graph can be drawn for iru. The Table Graph of iru is simpler because iru
cannot assign a dynamic locative.

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
          Balloon Graph of locative iru


              が ga
            Nominative                                          に ni
                                                           Static locative

        I make no commentary on this Graph. Rather I want to proceed to the uses of aru
an iru to specify possession.30

          Balloon Graph of possessive aru (iru)

          Tsuma-ni-wa ginkō-ni chōkin-ga hyakuman'en aru.
          My wife-DAT-TOP bank-LOC savings-NOM one million yen(QUANT) be
          My wife has got savings for one million yen in the bank.

          Kare-ni-wa futari-no kodomo-ga aru.
          He-DAT-TOP two-GEN child-NOM be
          He's got two children.

          Kare-wa Karuizawa-ni bessō-ga aru.
          He-TOP Karuizawa-LOC cottage-NOM be
          He's got a cottage in Karuizawa.

          彼にはアメリカに 90 歳の祖父がいる。
          Kare-ni-wa Amerika-ni kyūjussai-no sofu-ga iru
          He-LOC-TOP America-LOC 90 years' old(QUANT)-GEN grandfather-NOM be
          He's got a grandfather of 90 in the States.

          彼には子どもが 3 人いる。
          Kare-ni-wa kodomo-ga sannin iru
          He-DAT-TOP children-NOM three(QUANT) be
          He's got three children.

     The following examples are taken and adapted from the Nihongo kihon dōshi yōhō jiten 1989: 31, 48.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
                         ある iru
                         いる aru

             に ni                          が ga                                          に ni
             Dative                      Nominative                                 Static locative
              PLACE                          THEME                                      PLACE

          には niwa

             は wa

        When aru or iru is used to express possession, it takes two arguments. The first one
is the possessor, and must be an animate entity. Since I follow the Localist Approach, I
consider this entity be a PLACE. This entity is realized as the "logical subject"31 of the
clause, and its phrase is displayed in the left stack, in the position always taken by the
subject. This phrase is marked by a ni oblique case to which the topic marker wa can be
added or substituted. (This process of substitution is represented by means of the middle
balloon. I will be discussing the implications of this in a short while). The second argument,
drawn in the middle, corresponds to the possessed entity, which bears the role THEME. Its
phrase is marked by the nominative particle ga.32 This construction can accommodate a

   This distinction was originally formulated by Jespersen 1924.
   Kuno considers this second argument to be marked as a "object" (1976: Ch. 4). See my treatment of
hoshii below. It should be noticed that such an unexpected subject realization might be interpreted as truly
ergative (as briefly discussed by Jacobsen [1992: 23-4], quoting Kuroda [1978]), due to its marking pattern
(ergative case form also functioning as locative and generalized oblique, Dixon 1994: 57) and to the difficulty
of assigning grammatical relations to its arguments (see Manning [1996: 4] on the traditional analysis of
ergativity). The Japanese language would therefore exhibit a double ergative split: a split conditioned by the
semantic (aspectual) nature of the verb and a "main/subordinate" clause split (as described by Dixon 1994:
102). This conclusion is however prevented by the fact that ergative split systems always involve a
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
third phrase, which I draw on the right side of the graph. It is an adjunct, and refers to the
PLACE in which the THEME is physically located. Obviously it is marked by the locative
particle ni. Since there cannot be two identical surface cases in a sentence, I deduce that
the oblique case assigned to the subject phrase is not another locative, but rather a dative.
The Japanese language would not be first language in the world to express possession by
means of such a "dative of possession".
        This pattern seems to be derived from that of locative aru (or iru) by addition of the
EXPERIENCER argument. Once this new phrase is inserted into the left of the graph, the
other two arguments of aru get displaced to the left. But since aru remains a two-place
predicate, and can only take two arguments, the locative phrase on the extreme left of the
graph becomes an adjunct.
        I am aware that the presence of two PLACE roles is a bit disorienting. When I
composed my inventory of semantic roles I decided to keep their number at a minimum,
and therefore to do without a specific POSESSOR role. After all, the occurrence of this role
would be mostly limited to the predicates of possession like this one. To accept such
double-PLACE construction takes some effort, but there should not be much confusion once
one realizes that the PLACE role of the possessor can be borne only by an animate entity,
when in fact that of the physical place can be represented only by a place noun.

      And at long least, the time to behead the eel has come. The following are the
sentences that I am going to use in my discussion.

        Watashi-wa gakusei-desu.
        I-TOP student-COPULA(POLITE)
        I'm a student.

        Watashi-wa nihonjin-dewaarimasen.
        I'm not Japanese.

        Watakushi-wa gakusei-de ari-masu.
        I(formal)-TOP student-ESS be-POLITE
        I'm a student.

        I-wa unagi-ni shi-masu.
        I-TOP eel-TRA do-POLITE
        I'm for an eel.

        Watashi-wa unagi-desu.
        I-TOP eel-COPULA(POLITE)
        An eel, for me.

differential treatment of the subject S of intransitive verbs, not of the two core relations A and O of transitive
verbs, as it appears to be the case with Japanese.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
      Wagahai-wa neko-de aru.
      I-TOP cat-ESS be
      I am a cat.

        I have displayed these examples in all notations, but later on I am going to use only
their transcript in the Latin alphabet. In these examples, the topic marker wa has replaced
the nominative marker ga, as it is common in plain declarative clauses, so that the
examples would sound more natural and closer to those in the textbooks. This does not
change the overall case assignment pattern, however, whose relevant part is the right area
of the clause. The following is the Balloon Graph that synthesizes all the examples above.

      Balloon graph of identificative aru


          が ga                                             で de
        Nominative                                         Essive
          THEME                                           MANNER

         For as much as I know, a clause like watashi-wa gakusei-desu appears in the very
first unit of all the Japanese language textbooks around the world. It is usually explained
by the teacher as being composed by the subject and by a nominal predicate, with desu
being a "sort" of verb "to be". A little more detailed explanation would say that desu is a
copula, meaning "to be". The same happens to its negative form, with dewaarimasen
standing as the negative verb "to be" or the "negative copula". This is indeed the analysis I
used in the notations above. This kind of explanation usually does stick, because, I must
admit, it is sound and easy to understand, at least at the level of the students it is offered
to. It sticks, but after a while, usually only few weeks of classes, it stinks. What happens is
that once students learn that the affirmative and negative verb forms of Japanese verbs
are morphologically related, and one can be derived from each other, they would start to
secretly look for a mechanism for converting desu in dewaarimasen. This will turn in all
kinds of problems afterward, with students having difficulties in understanding why
particles wa and mo appear or disappear from verbal or adjectival phrases. I am thinking
of "words" like denai, or phrases like sō demonai ("it's not really so"), mazu-ku-wa nai
kedo... ("it's not bad, but...") and so on.
         In order to cope with such a heuristic problem, I just tackle it as soon as possible,
and I draw the above graph on the blackboard no later than my second lecture to the first
grade, right after explaining – something I usually do during my first lecture – that a
Japanese clause is made by a sequence of phrases composed by groups of noun+particle
and headed by a verb, and that the verb does not change according to the person. It works
                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
nicely, and imprints my students with a certain way of seeing Japanese grammar that lasts
        In short, my analysis is based upon singling out the nominal part of the identificative
predicate as an independent argument phrase marked by particle de. This particle I
identify as the marker of the essive case. This is how I go – I am using here the exact
wording I employ in classes.

        (1) Desu sentences are not composed by the ordinary sequence of phrases made
by a noun followed by a particle;
        (2) on the contrary, negative "be" sentences (identificative sentences), ending in a
noun followed by dewa (or ja) arimasen, do follow such a basic pattern;
        (3) no morphologic relation exists between desu and dewa (or ja) arimasen, and
there is no rule for generating dewa arimasen from desu or vice-versa;
        (4) desu sentences are the result of a two-stage substitution process that consists in
first "beheading" an affirmative sentence by cutting away its verbal head and rightmost
particle, and then in replacing this lost material with a "dummy verb";33
        (5) a "dummy verb" is a word shaped like a verb which displays some of the
morphological behavior of a verb, but which is not a verb and has no meaning. Its only
function is restoring the courtesy level and tense of the missing verb, because the actual
communication just cannot do without these two pieces of information.
        (6) Desu is a dummy verb of such kind, and the only information it carries is
"present tense", "polite speech". Other members of this family are da, which is desu's
brother and marks the present tense in plain speech, deshita, which is desu's cousin and
carries the information "past tense", "polite speech", and datta, the sister of deshita,
denoting "past tense", "plain speech". Cousin brothers deshō and darō come later.
        (7) So what actually happened to the sentence watashi-wa gakusei-desu? Its
original form was actually

          Watashi-wa gakusei-de ari-masu.
          I-TOP student-ESS be-POLITE

       This sentence just plainly follows the basic pattern. The pronoun watashi "I" is
followed by the particle wa, which replaces ga but still identifies watashi as the subject.
The noun gakusei "student" is also followed by a particle, de, which specifies the condition
under which the entity watashi exists. (More about this later). The verb "to be" is arimasu,
which is the present tense, affirmative, polite form of verb aru. To this verb I give the
meaning of "be" throughout all my lectures.
       You can find this pattern in the negative sentence. Its ending is gakusei-de-wa

          Watashi-wa nihonjin-de-wa arimasen.
          Watashi-TOP Japanese-ESS-TOP be+ POLITE+NEGATIVE

        Here, gakusei is ordinarily followed by de, and then comes the head arimasen,
which is the present tense, negative, polite form of arimasu. In between comes wa. This is
the same wa of watashi-wa, and obviously does not mark the subject. What's there for?
Let it suffice to say that negative forms often call wa into their clause. (This is indeed the
topic marker, but used in a contrastive function).

     Lyons 1968: 324.
                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009
        How do we come to the desu clause, then? It's so simple:

        Watashi-wa gakusei-de arimasu.
        Watashi-wa gakusei-de arimasu.
        Watashi-wa gakusei-desu.

       Watashi-wa gakusei-de arimasu was indeed the original sentence. The beheading
took away the predicate, arimasu, and the particle at its right, de. Now, without its head,
the clause would just collapse. The problem is not really that we have lost the information
delivered by the meaning of the verb (which was "be"). That we can do without; several
languages in the world actually do without (I am thinking of Russian, for instance). What
we cannot do without is the information concerning tense and level of courtesy. After all,
there are only two tenses in Japanese, past and non-past, and if a speaker starts ignoring
them, she will never able to put events in time. Moreover, using the correct level of
politeness is absolutely crucial for interpersonal communication in Japan. So the lost
material must be replaced somehow. That is exactly the function of desu and its family,
which only carry that relevant information, and nothing more.

       This surgical operation of beheading can be done to most affirmative clauses,34 no
matter the meaning or behavior of their final, heading predicate. But it is most common
with identificative "be" clauses. Actually, it is so common with these clauses that one is
lead to think that desu is the verb "to be". However, example of "be" clauses that do not
undergo the beheading are not so unusual too. One can hear them in very formal, official
speeches, or, quite ironically, in the commentary to baseball and wrestling matches.
       But then, how can one really understand what watashi-wa gakusei-desu means, if
the semantic information carried by the missing verb is lost? So we come to the infamous
example of the eel.

        Watashi-wa unagi-desu.

        If we take desu to mean "to be", the sentence means that I am an eel. But obviously
this is not the case. Indeed, such a sentence comes from a number of other possible
sentences, like, for example:

        Watashi-wa unagi-ni shimasu.

      I have already discussed the case pattern of predicate suru "do". When this
predicate assigns translative case, it can convey the abstract meaning of "decide for
something" or "choose something". This sentence means "I choose eel", and is used at the
restaurant, when you plan your meal. The beheading went like:

        Watashi-wa unagi-ni shimasu.
        Watashi-wa unagi-desu.

   An exception are kara and made phrases, which cannot deprived of their particle. This is due to the fact
that these phrases specify the boundary of segments (in either time or space), so that without them one
cannot tell whether the surviving noun would specify a start-or an end-point. Alternatively (but it amounts
very much to the same thing), one can say that since kara/made phrase pairs specify segments and
therefore quantities of time or space, they are adverbials just like nijikan ("(for) two hours) or nikiro ("(for) two
kilometers"), and thus they form a whole which cannot be mutilated of any of its parts.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
        The final part -ni shimasu was cut away, and replaced by desu. This reveals one
basic facet of the beheading: in a desu transform much information is lost. It is the
semantic information carried by the verb. As I said, one can indeed do without it, but only
in a situation when it is clear what the lost material was, and its meaning is retained
somehow. So it is just fine if I say watashi-wa unagi-desu at a restaurant, for everybody
will understand me. But if I say it in the classroom, when a teacher is supposed to
introduce himself, that would just be disorienting, even disturbing. At best, it would
communicate nothing. But since the beheading happens with most of the "be" sentences,
and we are prone to take this as the default situation, in a setting like that watashi-wa
unagi-desu would be most probably taken as meaning "I am an eel".

        But what if one wants to say that she really is an eel? What if eels could speak and
tell us their story? How could one specify that watashi-wa unagi-desu really means "I am
an eel"? Writer Natsume Sōseki did something like that in his famous novel Wagahai wa
neko de aru, published in 1905-1906.

       Wagahai-wa neko-de aru.

       In this novel, Sōseki described the life of a cat. Or, better, the lives of human beings
as seen from the eyes of a cat. Wagahai is an old male pronoun for "I". In his title, Sōseki
wanted to specify exactly that the speaker was a cat. To do that, he used the whole, un-
beheaded clause. Of course the stylistic requirements of his time demanded for a formal,
refined prose style, and this dearu ending was somehow mandatory. But he could have
never, never done with something like Wagahai-wa neko-da. He couldn't behead his own
       When I come to this, I usually conclude my analysis by discussing the essive de. I
comment on it saying that this phrase represents here an argument that is absolutely
necessary to the two-place predicate aru. (It is so necessary, in fact, that the particle de is
usually taken as a part of aru, and the verb "to be" is often indicated to be dearu. The
phrase marked by de would then be the nominal part of the predicate). There are many
instances, however, in which similar de phrases occur as adjuncts, with predicates of all
kinds. For example, such are the adverbials hitori-de "alone, by myself" or jibun-de "by
myself". The "entity" specified by these de phrases bears the role of MANNER, which I
defined as:

        MANNER          The condition under which the process of a dynamic event is carried out or a
                        stative event takes place.

       The essive de phrase specifies the "manner of existence", namely the condition
under which the eventuality expressed by the verb aru unfolds. In an identificative event
such as that described by wagahai-wa neko-de aru, the condition of neko puts a constraint
on the existence of the subject. In more direct, clear words, the fact of being cat conditions
the beliefs, behaviors, ecology of the subject to such an extent that the subject comes to
be classified, or identified, by the very condition of "cat". There is a complete fusion of "I"
and "cat". Wagahai-wa neko-de aru then means "I exist as a cat" in the sense of "I exist
being subject to the condition of 'cat'" If you have read the novel you know that a large part
of being a cat is living in the constant fear of being eaten by the humans.

      Having reached this point, it would be a pity not to draw the Table Graph that plots
together all these uses of aru.

                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
         Table Graph of predicate aru "be"

        Location events               locative               possessive               partitive    identificative

    Predicate(s) of being            ある aru                  ある aru                   ある aru        ある aru
                                 (non-motile theme)         (inanimate,            (human theme)
                                                          animate theme)

                                     いる iru                 [ いる iru ]               [ いる iru ]
                                   (motile theme)            (modified,            (human theme)
                                                          animate theme)

Thing        THEME                  nominative               nominative             nominative     nominative
                                      が ga                     が ga                   が ga           が ga
             PLACE                 locative に ni
             (physical             (thing, being)
Reference    field)                locative で de
locations                              (event)
             PLACE                                          dative に ni
             possessive field)
             MANNER                                                                                essive で de
             (condition of
             PLACE Adjunct                                 locative に ni              locative

        In this Table I consider aru to be one and the same predicate across all
constructions, no matter the kanji used to write it. I also loaded into this Table a fourth use
of aru, that of partitive aru.
        The mean feature of this Table is the grouping of the three oblique arguments of
locative, possessive and identificative aru under the general denomination of "Reference
locations". I believe this is some kind of requirement, if one wants to display in just one
table all the uses of aru. This is concretely done under the provisions of the Localist
Approach, which I have already mentioned. This Approach considers all event to be
involving one "thing" and one "place" (properly a "reference thing"). Events are then
divided into location events, involving one thing and one location, and motion events,
involving one thing and one path. Events can be further classified into several "semantic
fields": a positional field, a possessional field, and an identificational field, according to the
type of place they involve. (A fourth field, the temporal field, is of no relevance here). In
this Table are represented the Location events described by verb aru, which involve a
THEME and three types of locations. These three types of locations are listed at the header
of the large central band of the Graph. Let us consider now each predicate individually.

       Locative aru and iru involve a located thing, the THEME, and a physical place. When
the THEME is an entity that cannot move spontaneously (or a happening), the event is
described by aru. When the theme can move spontaneously, the verb is iru. (I took the
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
term motile from biology. It defines an animal capable of spontaneous movement). This
argument is the subject. The second argument is an oblique locative, bearing the role
PLACE, and I have already discussed its two alternate markings.
       Possessive aru involves a possessed thing, that bears the role THEME, and the
abstract place where this THEME entity is located – in other words, its possessor. This
possessing entity bears the role PLACE. Officially, this event of possession should be
described only by aru. It is common, however, to substitute aru with iru when the
possessed thing is animate. Furthermore, contemporary native speakers feel that this
substitution is mandatory when such animate theme is modified by a non-quantifier
adjective or by a relative clause. I have already discussed the occurrence of the locative
adjunct and the marking of the oblique phrases. In the Table above, the locative adjunct
requires a horizontal band all for itself.
       The intruder in this Table is partitive aru. This is an example of it.

        Gakusei-no naka-ni [hantaisuru] hito-ga at-ta.
        Gakusei-GEN inside-LOC [oppose+ADNOMINAL] people-NOM be+PAST
        Among the students there was someone against (it).

       Apparently, iru would work just as well here, but I believe this is a recent corruption.
According to my reference texts,35 the THEME entity must be a person, and there is no need
to mention a place. When one does it, however, the locative phrase is an adjunct – that is
why I displayed it in the bottom band of the Table.
       Last comes identificative aru. This is perhaps the most interesting verb to discuss in
accordance with the Localist Approach. The THEME is the entity described as being in a
certain condition. The condition it is in, bearing the role MANNER, is marked by the essive
case marker de. But de is also the marker of the dynamic locative. Thus, one can say that
aru describes the identificative event of being in a certain state as if it were a location
event: aru places an entity in an abstract location – a condition.36 This is very consistent
with the way I analyzed the predicates of change, when I treated change as an abstract
movement between two conditions. One step further, and one should re-name the de
essive marker as locative. Indeed, in the following example ni actually functions as an
essive case marker, just like de:

        Kare-ga kurabu-no kaichō-no za-ni aru.
        he-NOM club-GEN president-GEN seat-LOC be
        He is in the seat of President of the club.
        He is the club's president.

       Both ni and de hence mark a location, and are essive case markers whenever the
location is a condition of existence.

   The Nihongo kihon dōshi yōhō jiten (1989).
   The particle de has a verbal origin and actually comes from the contraction of ni and the verbal form te. It
is interesting to notice that while ni marks the new condition assumed as the result of an event of change of
state, if followed by the extensional marker te it rather takes the function of marking a condition as it were
extended in the linear, abstract space of time.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
      In the next session, before proceeding to a discussion of alternations, I am going to
deal with some more stative predicate. The first one will be hoshii.

       Balloon Graph of the stative predicate of desire hoshii


          が ga                                              が ga
        Nominative                                        Accusative
       EXPERIENCER                                            THEME


                                                       Clause v. て te
                                                          Effected act

        Hoshii is a stative predicate describing a feeling, namely a desire. Lexically, it is an
adjective, but it is a true predicate since it assigns case just like a verb. It requires two
arguments, one referring to the entity that feels a desire, the other to the object of her
        The first argument is obviously the EXPERIENCER, and is realized as the subject by
means of the nominative case marker. In ordinary conditions. hoshii cannot occur with any
other subject than the first person "I". The most common exception is a direct question, in
which case the experiencer is the listener. All this means that the EXPERIENCER argument is
very seldom realized, a fact that I indicate in my Table by drawing its balloons with a
dashed contour. It should be noticed also that ga is most commonly to replaced by the
topic marker wa.
        Let us consider then how to mark the second argument, that describing the THEME,
that is, the object of desire. There are two basic marking patterns, displayed in the stacked
balloons on the right. The first pattern is used when the object is a physical thing. Its

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
argument is then filled by a noun, and is marked by ga. But what is the case that this
particle ga is marking? There can be only one nominative case in any given sentence. So
obviously this is not a nominative. Following Kuno and Teramura,37 who consistently
consider this argument to be an object, I classify this marking as an accusative. There are
two basic reasons for doing that.
        The first one is that a large number of languages exhibit instances in which the
morphologic realization of the accusative case is identical in form to that of the nominative.
(Just like the unmarked nominatiivi and akkusatiivi of Finnish). The second reason is that
the a double-ga construction exhibited by hoshii is not peculiar to hoshii, but shared with
several other predicates of feeling, such as suki-da (appreciate, like) or kirai-da (dislike).
Now, in rare instances, these predicates can substitute ga with the ordinary accusative
case marker o. This happens with hoshii as well, but it’s a very rare instance. Martin report
the following:38

        Kore-o hoshii ka.
        this-ACC want Q
        Do you want this?

        Sorekara [utsukushi-i onnagata-o dasu-yō-na] mono-o hoshi-i desu ne
        then [beautiful-PRESENT onnagata-ACC exhibit-PRESENT-LIKE-
        And then, (on the program) I'd like to have something that would present a beautiful
        onnagata, you see.39

       I have displayed such a rare alternate marking in the central balloon of the object
stack. Of course it is possible that this be a case alternation, but I believe it is rather the
evidence that this second argument exhibits a double, alternate marking of the same case,
the accusative.
       The second marking pattern is realized when the object is a performance, more
specifically, an action that the subject wants to be performed (effected) in her behalf by
someone else. Since acts are described by verbs, this argument is to be filled by a verb, or,
more precisely, by a clause. This is indeed the same situation I discussed for the
benefactives predicates of giving. The solution is also the same: a clause is inserted into
the sentence by means of the complementizer –te.

       Correctly identifying the case assigned to the object is the key to explain the odd
alternate constructions exhibited by a group of related predicates of feeling.

        Balloon Graph of the predicate of feeling urayamashii

        Boku-ga Masao-ga urayamashii.
        I-NOM he-ACC envious
        I envy Masao.
   Kuno 1976: Ch. 4; Teramura 1973: 22, 1982: 64, 1988: 146.
   Martin (1988) discusses extensively these instances, and makes the following examples on pages 196
and 199 respectively.
   Martin explains the latter structure as derived from the pattern –o mochitai "I want to have" (1988: 199).
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
          Boku-ni Masao-ga urayamashii.
          I-DAT Masao-NOM envious
          Masao is enviable to me.
          I envy Masao.


             が ga                                             が ga
           Nominative                                       Accusative
           EXPERIENCER                                         THEME

              に ni                                            が ga
              Dative                                        Nominative
           EXPERIENCER                                         THEME

         This group or subclass of predicates includes ureshii (happy), osoroshii (terrifying),
yorokobashii (pleasant) and so on. They all show the pattern above. In the case structure
displayed in the first pair of horizontally linked balloons, the EXPERIENCER and the THEME
stand in the same relationship as those of hoshii. As in hoshii, marking the EXPERIENCER
with the topic marker wa is far more common, but still a ga marking is possible, and it must
be accounted for. 40
         In the second pair of balloons, the particle marking the THEME phrase does not
change, but the EXPERIENCER is now realized by means of the oblique particle ni. This
causes two problems.
         The first problem is how to classify the case marking of the THEME phrase. Every
sentence must have both a subject and a nominative case, but they do not necessarily
have to be realized by the same phrase. This is what happens here. The EXPERIENCER is
still the subject, because all events of perception and feeling are centered on the
perceiving entity, and also because this ni phrase does exhibits the syntactic behavior of a

     The examples above and their general discussion are taken from Teramura 1973: 22, 1982: 64, 1988: 146.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
Subject.41 But is not marked by the nominative case. The nominative is rather assigned to
the second entity, the THEME.
       The second problem is how to classify the ni case marker assigned to the
EXPERIENCER phrase. I have decided for the most obvious solution, that is, to consider this
case to be a dative. This is consistent with the behavior exhibited crosslinguistically by the
large number of predicates that express possession and feeling by means of the oblique
pattern "to me is" (in Italian: A me piace "I like"). In one word, this dative marking is part of
the large class of the datives of feeling, possession and perception. But there are other
possibilities. For instance, according to a strong Localist Approach, this could be a locative,
marking the place where the desire is felt, as the "possessor" of a feeling can be
       That exhibited by urayamashii and its companions is a radical change of case
structure, and there is a huge distance between the two case patterns. The two structures
do not have to be necessarily related – there might be no syntactic or semantic rule to
generate one from the other. Nevertheless, there are several theories that try to determine
which construction is the parent and which one is derived, via the processes of
subjectivization, thematization and so on. However, this question does not really matter
when it comes to teach these predicates in the classroom.

        I want to discuss now a completely different kind of predicates. They are called
“symmetric” because they describe events that involve the equal or "symmetric"
participation of two entities. The first group of the following examples concerns the
wedding and divorce of Keiko an Masao, the second group displays the behavior of the
stative predicates meaning "same" and "different".

        Masao-ga Keiko-to kekkonshita.
        Masao-NOM Keiko-REC marry-PAST
        Masao married Keiko

        Keiko-ga Masao-to kekkonshita.
        Keiko-NOM Masao-REC marry-PAST
        Keiko married Masao.

        Keiko-to Masao-ga kekkonshita.
        Keiko-CONJ Masao-NOM marry-PAST
        Keiko and Masao married.

        Masao-ga Keiko-to rikonsuru.
        Masao-NOM Keiko-REC divorce
        Masao will divorce Keiko.


 For instance, predicates receive honorific marking according to the social status of the subject. Here the
marking would be determined by the ni EXPERIENCER phrase. See Tsujimura 1996: 232 from Harada 1976.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
      Keiko-ga Masao-to rikonsuru.
      Keiko-NOM Masao-REC divorce
      Keiko will divorce Masao.

      Keiko-to Masao-ga rikonsuru.
      Keiko-CONJ Masao-NOM divorce
      Keiko and Masao will divorce.

      Kono kasa-ga sono kasa-to onaji-da.
      This umbrella-NOM that umbrella-REC same-COPULA
      This umbrella is the same as that.

      Kono kasa -ga sono kasa-to chigau.
      This umbrella-NOM that umbrella-REC different
      This umbrella is different from that.

      Symmetric predicates come as action verbs (kekkonsuru or rikonsuru), stative verbs
(chigau) or adjectives (onaji). No matter their lexical classification, they all behave in the
same way syntactically.

      Balloon Graph of symmetric predicates of action kekkonsuru and rikonsuru

                 結婚する                                                      離婚する
                kekkonsuru                                                 rikonsuru

      が ga                   と to                               が ga                     と to
    Nominative             Reciprocal                         Nominative               Reciprocal
       AGENT                  AGENT                              AGENT                  AGENT
       (GOAL)                (GOAL)                             ABLATIVE               ABLATIVE

                                Dalla Chiesa -The Semantics of Japanese verbs
                                           Helsinki 4th-7th May 2009
          Balloon Graph of symmetric stative predicates onaji and chigau

                      同じ                                                              違う
                      onaji                                                          chigau

          が ga                     と to
        Nominative               Reciprocal                         が ga                        と to
           THEME                   THEME                          Nominative                  Reciprocal
                                                                      AGENT                     AGENT
                                                                   (SOURCE)                   (SOURCE)

        The eventuality described by a symmetric predicate involves only two participants.
Their mode of involvement is identical, hence they bear the same role(s); moreover, their
participation is equally necessary for the event to happen. This means that neither entity is
more important for the realization of the event, and either can be realized as the subject,
with no difference whatsoever in meaning. Which participant is to be realized as the
subject and marked by nominative case is up to the speaker. The other argument takes
the reciprocal case.
        Let us discuss the first three examples. They concern the wedding of two persons,
Keiko and Masao. When a speaker decides to describe this event by means of the
predicate kekkonsuru, she can fill the position of subject with Keiko, Masao, or even with
both, joining their names in one single phrase by means of the conjunctive particle to. This
reflects the fact that, according to "cultural necessity",42 an event of such kind obviously
involves two participants in equal control of the event.
        Why chose Keiko as the Subject, then? Why not Masao, instead? Why not both? In
terms of meaning, there would be no difference. Basically, the choice is made by the
speaker for reasons of discourse. For example she might choose Masao because she is
speaking about him and he is already the subject of the preceding sentence.
        Once the subject is chosen, however, the problem arises of how to mark the second
argument. Neither an accusative or an oblique case would do, because this would mean
that the entity represented by such second argument participates into the event in a
different way than the subject, and with less control and importance. Therefore, Japanese
symmetric predicates do not assign accusative or oblique case; instead they assign the
very particular case named reciprocal, whose marker is to. This marker transmits a clear
message. It says that the entity filling that phrase possesses the same role and, virtually,
the same case as the subject.
        The same is true for rikonsuru. The fact that the two entities separate instead of
unite is of no importance. The reciprocal case underscores the fact that a divorce too
requires the equal participation of both actors.

     See Lyons (1995: 121-124) for a discussion of semantic entailments of this kind.
                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
        Again, the same is true for stative symmetric predicates. For the assignment of the
reciprocal case, whether the predicate describes an action or a state is absolutely
irrelevant. Indeed, it should be noticed that the reciprocal case is totally indifferent to the
role(s) borne by the entity it marks. It just duplicates all the roles borne by the subject. The
two arguments of rikonsuru bear both the roles of AGENT (a divorce is voluntary) and of
SOURCE (the two actors separate from each other, both in a physical and in an abstract
way), those of kekkonsuru are AGENTs (but could be said to also bear the role GOAL, since
the two actors abstractedly move toward each other), and the arguments of onaji and
chigau are just THEMEs.
        I have already remarked that the reciprocal case is assigned only by symmetric
predicates. It should not be confused with the comitative case. This case has a function
quite similar to that of the reciprocal, for it also serves to duplicate a set of roles and other
"virtual" cases. However, it realizes a participant that is not necessary to the event. As
such, the comitative marks an adjunct phrase. Moreover, it can duplicate any other
argument or adjunct, non only the subject. For instance, the sentence

       Keiko-ga Masao-toisshoni kekkonshita.
       Keiko-NOM Masao-COM wed-PAST

        may only mean that Keiko and Masao held their wedding together, but each got
married to a different person. The joining of Masao in the event is not necessary for the
wedding of Keiko to happen. The comitative case can be marked by particle to as well, just
like the reciprocal. Since the latter is assigned only by symmetric predicates, there are
very few instances in which an ambiguity can arise.

       I have reached now the last section of my lectures. In this section I want to discuss
a number of predicates that alternate, that is, that can describe a same event type,
involving just one and the same set of roles, by means of two different case structures.
The Table Graph is particularly good in capturing alternations and analyzing the case
patterns of these predicates. This is why, in the next section, I will skip the Balloon Graph
representations of the alternating predicates and use their Table Graphs only.

       The most famous and discussed example of alternation is the so called “spray/
paint" alternation”, which in Japanese is exhibited by predicate nuru.

              (1a)   父が壁に赤いペンキを塗った。
                     Chichi-ga kabe-ni aka-i penki-o nut-ta.
                     father-NOM wall-TRA red-ADNOMINAL paint-ACC smear-PAST
                     My father applied red paint onto the wall.

              (1b)   父が壁にペンキを赤く塗った。
                     Chichi-ga kabe-ni penki-o aka-ku nut-ta.
                     father-NOM wall-TRA paint-ACC red-ADVERB smear-PAST
                     My father applied red paint onto the wall.

              (2a)   父が壁を赤いペンキで塗った。
                     Chichi-ga kabe-o aka-i penki-de nut-ta.
                     father-NOM wall-ACC red-ADNOMINAL paint-INS paint-PAST
                     My father painted the wall with red paint.
                               Dalla Chiesa -The Semantics of Japanese verbs
                                          Helsinki 4th-7th May 2009
                 (2b)    父が壁をペンキで赤く塗った。
                         Chichi-ga kabe-o penki-de aka-ku nut-ta.
                         father-NOM wall-ACC paint-INS red-ADVERB paint-PAST
                         My father painted the wall red.

       Table Graph of the alternating predicate nuru

    alternation               (1a)                   (1b)                    (2a)              (2b)
    塗る nuru
    AGENT               nominative が ga      nominative が ga          nominative が ga    nominative が ga
    P.LOCATUM           accusative を o         accusative を
    INSTRUMENT                                                       instrumental で de   instrumental で de
    PATIENT                                                           accusative を o      accusative を o
    GOAL                translative に ni     translative に ni
    GOAL                                         translative                                translative
    (abstract)                                     / adverb                                   / adverb

         The spray/paint alternation is one example of locative alternation. This
denomination comes from the fact that the argument whose case marking alternates as
the objects refers to the physical place where something is put. (Actually one could debate
is this is not also an instrumental alternation).
         This is how I represent this alternation. In all the events described by nuru an AGENT
participates that initiates the event as its ultimate cause. This entity (in the examples,
chichi) physically transfers a stuff (penki), bearing the role of P.LOCATUM, on to a surface
(kabe), which represents the concrete, spatial GOAL of the event. (The LOCATUM possesses
the physical property of being liquid or soft enough to be smeared, which is semantically
important but not relevant here.) This interacting of roles is captured by the construction
plotted in columns (1a) and (1b). In these case patterns, the P.LOCATUM is realized by
means of an accusative phrase and the spatial GOAL by a translative ni phrase.
         In so doing, however, the AGENT also changes the aspect of the GOAL, because the
wall changes color. This means that the wall can also be considered to be the PATIENT of
an event of change of state. In this event, the stuff that represents the effector or
immediate cause of the change is the INSTRUMENT of the act performed by the AGENT. This
additional mode of interaction is captured by the constructions in Columns (2a) and (2b).
These constructions represent a case structure significantly different than that of Column
(1), since now the argument representing the PATIENT of the change of state (the wall
kabe) receives accusative case, whereas the phrase denoting the stuff effecting the
change (the paint penki) is realized by means of the instrumental marker de.
         In other words, nuru alternate between two semantic and syntactic behaviors: that
of a predicate of putting, which is of course transitive, and that of a transitive predicate of
change of state.
         The line at the foot of the Graph plots the two akaku adverbials of the (b) examples.
I added these lines to the Graph because these phrases must be accounted for, since they
obviously realize an additional, abstract GOAL role, namely the new condition assumed by
the PATIENT kabe as an effect of the act. (This is the only situation in which there might be
a need of distinguishing between spatial and abstract goals by labelling the latter
FACTITIVE). When the filler is an adjective, this goal phrase is morphologically realized by

                                     Dalla Chiesa -The Semantics of Japanese verbs
                                                Helsinki 4th-7th May 2009
an adverb (a –ku or –ni adverb, according to the class of the adjective). When the filler is a
noun, it is marked by the translative case (particle ni). Another interesting fact is that the
adjective akai denotes a quality which is originally a property of the stuff penki. This is
made clear in the (a) examples, where the adjective akai appears as a modifier of penki.
The act nuru then transfers this quality to the wall by way of transferring the penki itself
onto it. This shift of head is syntactically realized by displacing the adjective akai to the
adverbial position of (b) and thus changing its function to that of an object predicative
        The spray/paint alternation is the focus of much research because the two
alternating marking patterns are not semantically equivalent (and therefore do not
necessarily describe the same event token). The second case pattern has a holistic
interpretation – the wall is considered to be entirely painted. This difference of meaning
appears across a large number of languages in the world.43 In my opinion this difference
stems from the different semantic properties of the two arguments realized as direct
objects. Kabe refers to a limited surface, and the painting of it is a telic, bound act. This act
is necessarily accomplished by covering all the wall's surface with paint, hence the sense
of completeness of kabe-o nuru.44

       I am discussing next the alternation exhibited by the verb of searching sagasu. I
believe this alternation should be named "locative" too, for one of the alternating
arguments expresses the physical location of the event.

        Table Graph of the alternating predicate sagasu

        (1)     警察が犯人のマンションでナイフをさがした。
                Keikan-ga hannin-no manshon-de naifu-o sagashi-ta.
                policeman-NOM criminal-GEN house-LOC knife-ACC search-PAST
                The police looked for the knife in the criminal's house.

                Poketto-no- naka-de kagi-o sagasu.
                pocket-GEN inside-LOC key-ACC search
                I look in my pocket for the key.

        (2)     地図で大島をさがす。
                Chizu-de Ōshima-o sagasu.
                map-INS Ōshima-ACC search
                (I) search Oshima on the map.

        (3)     警察が犯人のマンションをさがした。
                Keikan-ga hannin-no manshon-o sagashi-ta.
                policeman-NOM criminal-GEN house-ACC search-PAST
                The police searched the criminal's house.

   For Japanese, see its discussion in Jacobsen 1992: 44.
   Similarly, the stuff "paint" does not exist in a limited quantity, and the act of spreading it on the wall does
not necessarily consume all of it. This is why penki-o nuru does not entails the use of all the paint and has no
corresponding holistic interpretation. However according to Croft (1994: 44), They sprayed paint on the wall
"implies that the paint is completely affected (all of it was used)". The Japanese penki-o nuru allows, but not
requires, such a holistic interpretation.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
                Poketto-no naka-o sagasu.
                Pocket-GEN inside-ACC search
                I search my pocket.

      Locative alternation                  (1)                           (2)               (3)
     さがす sagasu "search"
        AGENT                       nominative が ga              nominative が ga     nominative が ga
        THEME                        accusative を o               accusative を o
        THEME                                                                        accusative を o
        PLACE                        locative で de
        INSTRUMENT                                               instrumental で de

       In this event, an entity acts within a limited place in order to get sight of a specific
object.45 Neither the searched place or the searched object are affected by the search.
The animate entity initiating the event is the AGENT in all three constructions. The object
targeted by the search appears only in Columns (1) and (2), where it is realized by an
accusative phrase. This entity bears the role THEME because is actually unaffected by the
agent's search. Indeed, searching for it does not mean finding it, and, even if found, this
object could just be left where it is, without changing place or possessor.
       The third participant is the location where the search takes place. In Column (1) it is
a physical space that bodily contains both agent (wholly or in parts) and patient. It
therefore bears the role PLACE, realized by the dynamic locative de, the locative of action.
Column (2) represents the situation in which this entity is a "space" only in an abstract
sense, since it is a container of information rather than a physical place. Interestingly, the
oblique surface marker does not change from (1), but now cannot be considered to be a
locative. I assign to this entity the role of INSTRUMENT because it is a physical object
exploited and possibly manipulated by the AGENT to carry out her search. Column (3) plots
an alternate construction which captures the control that the agent has over the searched
place and the volitionality of his act.46 As an additional THEME, the searched place is now
realized as the second argument of the predicate by means of an accusative o phrase.

       Next comes another example of alternation, the instrumental alternation exhibited
by the two predicates utsu meaning "beat" and "shot". These verbs have the same
phonological realization, but are written with two different kanji.

        Table Graph of the alternating predicate utsu "beat"

         (1)    事務員がワープロで手紙を打っている。
                Keiko-ga wāpuro-de tegami-o ut-teiru.
                Keiko-NOM word processor-INS letter-ACC hit-EXTENSIONAL
                Keiko is writing a letter with a word processor.

        (2)     事務員が一本の指でワープロを打っている。
                Keiko-ga ippon-no yubi-de wāpuro-o ut-teiru.

   See Jacobsen (1992: 31), who considers the purpose of the AGENT's act to be "the appearance of some
entity into the field of vision of the perceiving subject".
   See Teramura 1982: 107-108 for the use of o phrases as a mean to express the AGENT's control.
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
               Keiko-NOM one(QUANT)-GEN finger-INS word processor-ACC hit-
               Keiko is typing on the word processor with one finger.

   Instrumental alternation               (1)                          (2)
        打つ utsu "hit"            as a verb of creation          as an action verb
       AGENT                       nominative が ga              nominative が ga
       PATIENT                      accusative を o
       PATIENT                                                   accusative を o
       INSTRUMENT                 instrumental で de
       INSTRUMENT                                               instrumental で de

        In this first Table I am plotting the alternate case structures of the predicate utsu
meaning "beat". This is the verb utsu that students always meet during their basic
language training. I discuss its constructions as follows.
        The case structure of utsu plotted in Column (1) belongs to utsu as a verb of
creation. A verb of this class has an object that does not exist before the accomplishment
of the event it describes, but comes into being (or is effected) in the course of the act itself.
An object of this kind is called a resultative object. The predicates of creation are transitive
and require two arguments. The case pattern of Column (1) focuses on the initiator of the
event, which is the AGENT and is realized as the subject by means of a nominative phrase,
and on the entity that embodies the main purpose of her act, the object that she creates.
As such, this entity bears the role PATIENT and is realized by an accusative phrase. In
Column (1) a third entity is expressed which is the object that, controlled by the agent,
physically puts the letter into being. This entity bears the role of INSTRUMENT and is referred
to by an instrumental adjunct phrase.
        By means of the case structure displayed in Column (2), utsu merely describes the
AGENT's physical act of hitting a board's keys. The effect of this act, the coming into being
of a new entity, if any, is not relevant in this particular event description and so no phrase
referring to a resultative object appears. Instead, this construction focuses on the keyboard
as the entity that is most affected by the act. This second argument is therefore associated
with the PATIENT role and realized by an accusative phrase. This construction also allows
to express an immediate cause of the AGENT's act, standing as an effector of her
manipulation of the keyboard. For example, one of her fingers. Again, this entity is the
INSTRUMENT and is referred to by an instrumental adjunct phrase.

      At this point I sometimes add a discussion of the alternation exhibited by the utsu
predicate meaning "shoot". This is not necessary for my classroom work, but it adds a nice
touch and is quite useful.

       Table Graph of the alternating predicate utsu “shoot”

       (1)     警官がピストルで犯人を撃った。
               Keikan-ga pisutoru-de hannin-o ut-ta.
               policeman-NOM pistol-INS criminal-ACC shoot-PAST
               A policeman shot the criminal with a pistol.

       (2)     警官が犯人にピストルを撃った。
               Keikan-ga hannin-ni pistoru-o ut-ta.

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
               policemen-NOM criminal-TRA pistol-ACC shoot-PAST
               A policeman shot/fired a gun at the criminal.

  Instrumental alternation               (1)                            (2)
     撃つ utsu "shoot"
    AGENT                        nominative が ga               nominative が ga
    PATIENT                       accusative を o
    GOAL                                                        translative に ni

    PATIENT                                                     accusative を o
    INSTRUMENT                   instrumental で o

        "Shoot" utsu also requires two arguments, an agentive subject and a direct object.
The object is an affected PATIENT in both constructions (1) and (2). A strong affectedness is
very evident in the case pattern displayed in Column (1). Here a police officer shoots a
criminal by means of an effector, a pistol, realized as an instrumental adjunct. This
construction has a telic interpretation because it conveys the meaning that the police
officer does indeed reach the criminal with a bullet. In the case pattern displayed in
Column (2), on the other hand, the argument realized as the direct object is that
representing the less affected INSTRUMENT. Now the sense of achievement of the first
construction is lacking, and the clause does not entail that the bullet reaches its target.
This is made evident by the fact that in (2) the entity representing the criminal, the former
PATIENT, does not correspond to any argument phrase, but can only be expressed by a
translative adjunct phrase as the GOAL in which direction the bullet is shot and where it is
supposed to end its run. (The aspectual classification of utsu also changes, and the
predicate is now considered to be semelfactive, that is, describing an instantaneous action
which causes no resulting state).

       Table Graph alternating predicate tsukuru

       (1)     父が竹で笛を作った。
               Chichi-ga take-de fue-o tsukut-ta.
               father-NOM bamboo-INS flute-ACC make-PAST
               My father made a flute of bamboo.

       (2)     日本人は米からお酒を造る。
               Nihonjin-wa kome-kara osake-o tsukuru.
               Japanese-TOP rice grain-ABL sake-ACC make
               The Japanese make/brew sake from rice.

       (3)     正男がサバを刺身に作った。
               Masao-ga tai-o sashimi-ni tsukut-ta.
               Masao-NOM mackerel-ACC sashimi-TRA make-PAST
               Masao made the mackerel into sashimi.

       (4)     我が社がアメリカに工場を作る。
               waga-sha-ga Amerika-ni kōjō-o tsukuru
               our-company-NOM America-TRA factory-ACC make
               Our company will build a factory in the USA.

                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
       (5)       我が社がアメリカで工場を作る。
                 waga-sha-ga Amerika-de kōjō-o tsukuru
                 our-company-NOM America-LOC factory-ACC make
                 Our company will build a factory in the USA.

       Table Graph alternating predicate tsukuru

   Alternating            (1)                  (2)                  (3)                  (4)                (5)
   predicate        as a predicate       as a predicate       as a predicate       as a predicate     as a predicate
    つくる               of creation          of creation         of change of          of creation        of creation
 tsukuru "make"                           & of taking              state            & of putting
   AGENT              nominative           nominative           nominative           nominative        nominative
                        が ga                 が ga                 が ga                 が ga              が ga
  PATIENT           accusative を o      accusative を o                             accusative を o accusative を o
  INSTRUMENT instrumental で de
  SOURCE                                     ablative
                                           から kara
  PATIENT                                                       accusative
  GOAL                                                          translative
                                                                   に ni
  PLACE                                                                            translative に ni   locative で de

         Basically, tsukuru is a verb of creation, which put its object into being. This is indeed
how it behaves in the first two examples, which are also the most important ones for the
classroom use. Instances of these two constructions are found in most of the textbooks I
         The first example (Column 1) describes an event in which the AGENT uses a stuff to
create a new object. Emphasis is given to the PATIENT created in this way, which is marked
by the accusative. The stuff used is not so important in the unfolding of the event, and
indeed could be mentioned only by means of an instrumental adjunct. The semantic
relation between the act and its instrument is quite important, though. The stuff bearing the
role of INSTRUMENT is not much changed in appearance, and can still be recognized as
such in the effected thing. The bamboo does not disappear because of the act tsukuru, but
rather takes a new form and, more important, a new function, as a flute.
         Many of these considerations apply to the second example too (Column 2). Its
adjunct, however, is marked by the ablative case. This conveys the meaning that the stuff
used to produce a new object is completely changed in form and substance, and cannot
be recognized any more in the effected thing. One cannot discern the original rice in a cup
of sake. This departure from the shape and use of the original stuff is well specified by the
fact that the argument referring to it is marked by the ablative case. The verb tsukuru can
be understood to mean something close to "extract" or "take". And indeed, this
construction is close to that of the predicate toru, "take".
         The real alternation happens between the case patterns of Columns (1)-(2) and that
of (3), because tsukuru now turns from a verb of creation into one of change of state.
Therefore, there is a shift of direct object, and the case structure is altered. In (3), the
second most important entity on which the description of the event is focusing is the stuff
itself. The predicate is now a full three-place transitive predicate of change of state that
describes the altering of a stuff and its taking a new appearance. The affected thing is the
                                   Dalla Chiesa -The Semantics of Japanese verbs
                                              Helsinki 4th-7th May 2009
PATIENT   argument, and is realized by means of the accusative. The new shape or function
it takes, or, better, into which is put by the agent, represents the GOAL argument phrase,
and is marked by the translative. In the example above, Masao takes a rough substance,
dead fish, and skillfully changes it into a ryōri, a dish with a name ("sashimi"), a cultural
history, a precise algorithm of preparation.
        The examples displayed in Columns (4) and (5) are instances in which tsukuru is
used with the meaning of tateru, "build". It is still a verb of creation. In the both examples, a
company builds a (new) factory in the USA. The alternation between the ni and de marking
of the PLACE adjunct is semantically relevant. When a de locative is used, the clause
implies that the company does already possess other factories in the USA, and is now
building an additional one. On the contrary, when a ni phrase is used, the entailment is
that this would be that company's first factory ever built in the USA. The de adjunct is
obviously a dynamic locative, since the company building the factory is already active
inside of the USA. But the ni case marking cannot be considered to be a static locative,
because what is taking place in the USA, the building of a factory, is a very dynamic event.
I believe this is a translative case, since it conveys the meaning that the company is not
active in the USA but it is getting into the USA now for the first time. This ni marks a
translative even if you judge that the ni marking is “imported” from the case structure of the
related predicate tateru, which always assigns translative case to its place adjunct. With
tateru, the ni translative conveys the meaning of "sticking into the ground the foundation
poles of a building".

       A similar alternation between a case pattern of "creating" and one of "putting" is
also exhibited by kaku "write".

       Table Graph of kaku "write"

       (1)    正男が紙に名前を書いた。
              Masao-ga kami-ni namae-o kai-ta.
              Masao-NOM paper-TRA name-ACC write-PAST
              Masao wrote his name on a piece of paper.

       (2)    母がケーキの作り方をメモに書いた。
              Haha-ga kēki-no tsukuri-kata-o memo-ni kai-ta.
              mother-NOM cake-GEN make-way-ACC memo-TRA write-PAST
              My mother wrote how to bake the cake on a memo.

       (3)    恵子が紙に絵を書く。
              Keiko-ga kami-ni e-o kaku.
              Keiko-NOM paper-TRA picture-ACC write-PAST
              Keiko drew a picture on a piece of paper.

       (4)    誰かが黒板にいたずら書きを書いた。
              Dareka-ga kokuban-ni itazuragaki-o kai-ta.
              someone-NOM blackboard-TRA doodle-ACC write-PAST
              Someone traced some doodle on the blackboard.

       (5)    正男が先生へ手紙を書いた。
              Masao-ga sensei-e tegami-o kai-ta.
              Masao-NOM professor-TRA letter-ACC write-PAST
                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
              Masao wrote a letter to the teacher.

       (6)    学生にお知らせを書く。
              Gakusei-ni oshirase-o kaku.
              student-DAT notice-ACC write
              Writing a notice for the students.

    Predicate                       (1, 2)                         (3, 4)               (5, 6)
  書く kaku "write"            as a verb of putting               as a verb of     as a verb of creation
                                                                  creation           and sending

   AGENT                      nominative が ga                nominative が ga      nominative が ga
   P.LOCATUM                  accusative を o                                       accusative を o
   PATIENT                                                    accusative を o
   GOAL                         "transfers in"                   "creates in"         "sends to"
                              translative に ni                translative に ni     translative へ e
                                                                                     dative   に ni

        Examples (1) and (2), are instances of kaku as a verb of putting, since it describes
the act of transferring an existing information onto a new support: for example, Masao's
name from his long term memory onto paper, the algorithm of preparation of a dish from
visual or auditory memory onto a pad. Column (3) displays the use of kaku as a (telic) verb
of creation, whose effected object is incrementally created by the act and comes into full
being only at the end of it. It should be noticed that Example (2) (memo-ni kai-ta) can be
effectively considered to be both a verb of putting and a verb of creation, as a "memo" is
both the physical support of the transferred information and a created document. The two
instantiations of kaku as a verb of sending displayed in Columns 5, 6 entail that a new
physical object is effected by the act, but also that it is moved in physical space to reach
an animate goal (hence the label "translative" of the e phrase of 5) or that its informational
content has an animate goal as its destination (hence the label "dative" of the ni phrase of

      Table Graph of elative predicate of leaving deru

       (1)    煙が家から出る。
              Kemuri-ga ie-kara deru.
              smoke-NOM house-ABL exit
              Smoke gets out from the house.

       (2)    子どもがブランコから落ちる。
              Kodomo-ga buranko-kara ochi-ta.
              child-NOM swing-ABL fall-PAST
              A child fell from the swing.

       (3)    子どもが家から出る。
              Kodomo-ga ie-kara de-ta.
              child-NOM house-ABL exit-PAST
              A child got out from the house.
                                 Dalla Chiesa -The Semantics of Japanese verbs
                                            Helsinki 4th-7th May 2009
       (4)     息子が家を出た。
               Musuko-ga ie-o de-ta.
               son-NOM house-ACC exit-PAST
               My son exited from the house.
               My son left home.

       (5)     息子が高校を出た。
               Musuko-ga gakkō-o de-ta.
               son-NOM school-ACC exit-PAST
               My son left school.
               My son graduated from school.

       (6)     息子が会社をやめた。
               Musugo-ga shigoto-o yame-ta.
               son-NOM job-ACC quit-PAST
               My son quitted his job.

                        (1) 出る                       (3) 出る                       (4) 出る            (5) 出る
Elative predicates
    of leaving         (2) 落ちる                                                                    (6) 卒業する
                       spontaneous                                                volitional         volitional
    AGENT                                                                 nominative      が ga   nominative   が ga
    P.LOCATUM        nominativeが ga             nominative    が ga
                     ablative から kara          ablative から kara
                       physical place            physical place
                                                                           accusative を o        accusative を o
                                                                            physical place        abstract place

       The "ablative" alternation manifested by deru is one of a number of alternations that
are caused by the semantic properties of the nominals filling the semantic roles and by the
meaning of the clause as a whole, rather than by the semantic roles themselves. The
Table Graph is nevertheless capable to plot the two alternate realizations of the SOURCE
role and to show the difference of meaning associated with them.
       Deru alternates the realization of the SOURCE role between a kara ablative phrase
and an o accusative phrase. As pointed out by Teramura (1982: 107-8), the former
marking is used when describing a spontaneous event uncontrolled by the patient-like
subject; while on the contrary the latter is used in order to describe an event that is
considered to be under the volitional control of the agentive subject. The choice of a o
phrase thus reveals that in an event of volitional separation from a place, the entity bearing
the SOURCE role is cognitively represented as being closer to an affected object than to an
unaffected place. I have applied similar conclusions to a number of other o phrases during
my lectures (for instance, that realizing the searched place with sagasu or that realizing
the THEME with miru). Here, this strong component of meaning possessed by the o phrase
is consistent with its syntactic classification as a "postpositional phrase" (PP) rather than
as a "noun phrase" (NP).
       In constructions (1) and (2), the participants realized as subjects are respectively an
inanimate entity and an animate, affected entity. These non-acting subjects do not control
the event, and are therefore associated only with the P.LOCATUM role. The SOURCE role is

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
consequently realized by a kara ablative phrase. On the other hand, the meaning
conveyed by (4)-(6), in which the SOURCE role is realized by an accusative phrase, is that
the animate subject volitionally leaves a physical place, as in (4), or a place denoting an
activity or a condition by metonymy, as in (5) (6).
        Clauses (4) to (6) are strongly marked for agency, by means of the o accusative
phrase. On the other hand, both the nature of the nominals filling the subject position and
the use of the ablative case mark clauses (1) and (2) for non-agency. In (3), however, the
predicate admits volition, the subject is animate, but the kara ablative conveys a meaning
of spontaneous movement. In absence of an adverb such as wazato ("intentionally), such
a case pattern leaves it undetermined whether the subject is acting volitionally and in
control. This is the reason why I fused the two subject boxes of (3) together and I did not
classify the event for either spontaneity or volition.

       Table Graph of predicates keru "kick" and naguru "strike"

       (1)     キャプテン翼がボールを蹴った。
               Kyaputen Tsubasa-ga bōru-o ket-ta.
               captain Tsubasa-NOM ball-ACC kick-PAST
               Captain Tsubasa kicked the ball.

       (2)     正男がブーツでキャンを蹴った。
               Masao-ga būtsu-de kyan-o ket-ta.
               Masao-NOM boots-INS can-o kick-PAST
               Masao kicked a can with his boot.

       (3)     息子が隣の子を拳骨で殴った。
               Musuko-ga tonari-no ko-o genkotsu-de nagut-ta.
               son-NOM neighborhood-GEN child-ACC fist-INS strike-PAST
               My son stroke the child of the neighbors with his fists.

                         (1) 蹴る keru             (2) 蹴る keru            (3) 殴る naguru
     AGENT             nominative が ga         nominative が ga          nominative が ga
     PATIENT            accusative を o          accusative を o           accusative を o
                           one's foot
                                              instrumental   で de      instrumental   で de

       In (1), the phrase between angular brackets represents a rooted meaning. In other
words, in (1) the body part that serves as the effector or INSTRUMENT of the act is specified
by the meaning of the verb, not by some instrumental phrase occurring in its sentence (the
verb kick entails that you act with your foot, not with your fist). If necessary, this instrument
can be explicitly instantiated as an instrumental adjunct, as shown in (2). Predicate naguru,
instead, conveys only the general meaning of "strike", and one should specify if with fists,
a stick, or something else, by means of an instrumental adjunct.

                                  Dalla Chiesa -The Semantics of Japanese verbs
                                             Helsinki 4th-7th May 2009
         What I want to in this last chapter, just to round things up, is not to trace an outline
of all the things I have been talking about during the past four lectures. Rather, I want to
discuss my theoretical bases and some of the problems that are inherent in my approach,
as I came to understand them myself. On the one hand it is true that my Graphs are not
meant to be an attempt at rewriting whole cognitive processes and mapping algorithms. As
I said, I only want them to be a field aid for teaching Japanese verbs to students. On the
other hand, however, I did have to resort to a significant amount of theorizing in order to
build them. Hence, I think I should honestly discuss this process before I conclude.

         Broadly speaking, at the background of my graphs lays what could be called an
"event-based" or "entity-centred" approach. The syntactic treatment of a phrase, in other
words, its case, is determined by the role borne by the entity that is referred to by that
phrase. This is nothing special, I have been saying it all the time.
         But where do semantic roles come from? Roles are not assigned by the verbs, but
rather come from events. Roles are not a property of phrases, but rather a property of the
entities to which those phrases refer to – are determined on the basis of how entities
participate into events. This is to say that roles are event-dependent or, in other words,
that events define roles. Events come first; they happen independently of the verbs that
describe them. They are an ontology. Crisply put, semantic roles derive from an ontology
of events and define event-based properties of event-participants.47
         This is what lays at the background of my Balloon Graphs and Tables. I am
perfectly aware that this is not how semantic roles are commonly understood according to
linguistic theory. Usually, roles are assumed to be assigned by predicates, just as cases
are. But such an ontological precedence of events over predicates is what is required for
my Graphs to work.
         Such a direct relation of roles and events allows any given entity to bear more than
one role. This quality of semantic roles (and of thematic relations, which are the same
thing) is also called non-indexing function. In other word, an entity participating into an
event is not defined or labelled (indexed) by one single role. Giving roles such an indexing-
function would limit the extent of the participation of a given entity in any event to just one
modality (just like thematic roles do). Rather, an event participant should be defined by an
open set of roles, as many as the event requires. This is also crucial for my approach, for it
allows an event participant to be distinguished by a plurality of roles that can be
syntactically realized by means of different cases.
         However, this approach leaves unsolved the problem of how surface case and its
marker (particle or flexional morpheme) are determined. In Japanese, one solution could
be that of giving a meaning to cases, by means of a semantic analysis of its particle (like in
the semantic approach discussed above). This would make case a function of the event
itself, since the ultimate source of meaning is the ontology of the event. But how to
distinguish surface case from role, then? Surface case is obviously assigned by the
predicate, as represented by the Painkiller examples and implied by the very logic of
Teramura's grammar. A given predicate (or predicate class) selects for certain cases

   No matter how many entities are involved in a given event token, however, only a finite, small number of
them bears roles that are cognitively salient and necessary for an event of that type to happen. Therefore,
roles are finite and events can be defined and classified with respect to the number and type of their relevant
roles. This in turn entails that event types are also finite in number. In other words, at the background of my
approach an analysis of events is assumed which consists in identifying a set of formally (semantically)
salient functions and actants, in a way reminding of Propp's classical analysis of the morphology of the
folktale (Propp 1928), rather than in decomposing events in a number of temporally connected subevents.
                                    Dalla Chiesa -The Semantics of Japanese verbs
                                               Helsinki 4th-7th May 2009
because the meanings of their markers are consistent with the meaning of the predicate
itself. In turn, a speaker chooses a predicate and one of its accompanying case patterns (if
there is a choice) because they best describe the event she wants to represent (or her
perception of it). In short, my graphs are implicitly assuming that the realization of entities
by means of surface cases is the result of two, or possibly three, separate processes: an
objective semantic process of representing entities as roles, a morphosyntactic process of
surface case marking controlled by the verb, plus a third cognitive process, performed by
the speaker, of selecting a predicate and case structure among those provided by the
lexicon. I use semantic roles to explain why nominals are marked by certain cases and not
by others because all these processes seem to be governed by similar rules, and an
entity's role appears to have some influence on how that entity's syntactic realization is
marked. My graphs do not want to provide a consistent theory unifying the three processes.
Their only purpose is showing the regularities in the coupling of certain roles to other roles,
and of roles to cases.
         At the very end of my conclusion let me now mention three other problems that are
inherent in my line of reasoning. All stem from the fact that an event-centred approach like
mine requires that roles be assigned to entities over an objective observation of the event.
         Firstl, from a truly objective point of view, all animate entities participating in an
event are experiencers of some kind. I believe the reason is quite obvious: all animate
entities perceive, and react emotionally and cognitively to, the flow of events around them.
This is true no matter how the event is linguistically represented. Strictly speaking, then, all
animate entities bear a default EXPERIENCER role. But to formally assign such role to all
animate event participants would be useless and quite meaningless, for in the best part of
event descriptions such role is not explicitly realized. One could say that it remains in a
virtual state, transparent to syntax. Therefore, I decided to formally label as EXPERIENCERs
only those participants that are explicitly represented as such in event descriptions that
make a reference to a sensorial perception or to an emotional experience, as it happens
with the predicates of feeling, visual perception and so on. This means that not the whole
role structure of an event – as it can be ascertained over an objective observation of the
event itself – is necessarily realized in the syntax. This is consistent with my overall
approach, and does not represent a real problem.
         Second, my approach entails that all event participants must be observable and
provided with physical existence. How to consider places and times, then? Moreover,
several predicate require phrases that do have a "role", but instead of a physical entity
they merely denote a property of a physical entity, which therefore should not be
considered to bear a proper role. This is particularly evident with verbs describing
conditions and manners of existence. For example, the much discussed stative event
Wagahai wa neko de aru only involves one entity, that indicated by the pronoun wagahai.
The noun neko does refer to a physical being or category of beings, but functions only as a
reference to an identificative property of wagahai, rather than to a second, independent
participant in the event. This problem might be solved by developing a classification of the
entities cognitively relevant for syntactic realization in, for instance, animate actants,
inanimate actants, nonactant physical entities (places) and nonactant nonphysical entities
(as represented by nominals referring to qualities, resultative states, time etc.).
         Third, if events are assumed to have ontological priority over predicates and roles
are consequently defined over the objectively observable behaviors and interactions
exhibited by the participants, no BENEFICIARY role should exist, as the judgment of whether
any entity is advantaged or disadvantaged by an event is subjective and the selecting of a
benefactive verb or marker properly pertains to the realm of modality. Presently I have no
solution for this problem.

                              Dalla Chiesa -The Semantics of Japanese verbs
                                         Helsinki 4th-7th May 2009

To top