1The typology proposed by Talmy (2000) involves searching for the interface between spatial concepts and their morphosyntactic instantiations. Concepts relating to a motion event are divided into five components: “Motion”, “Figure”, “Ground”, “Path”, which constitute a motion event, and a “Co-Event”, which is associated with the motion event. The notion of motion pertains to the fact that the figure or moving entity changes its location in space and is localized with respect to the ground, and the path is the trajectory that a figure traces in motion. A Co-event includes various events, which are associated with the motion event through various relations such as “Manner”, “Enablement”, “Cause”, “Concomitance”, and so on.
- 1 Satellite is defined as follows: “it is the grammatical category of any constituent other than a n (...)
2At the morphosyntactic level, figure and ground are typically instantiated by nouns. The other concepts are candidates to be lexicalized in verbs, and for most languages, the main verb encodes the notion of motion and another component: the path or manner. When the main verb contains a path, the manner is expressed using adverbials or subordinate verbs. If the main verb expresses a manner, the path will be encoded by verb affixes or adverbal elements, the so-called ‘satellite’.1
(1)
|
a.
|
The bottle floated into the cave.
|
|
|
|
b.
|
La
|
botella
|
entró
|
a
|
la
|
cueva
|
flotando
|
|
|
|
the
|
bottle
|
enter.PST
|
to
|
the
|
cave
|
float.GER
|
(Talmy 2000: 49)
|
3In satellite-framed languages like English (1a), the main verb float represents manner information, while path information is encoded by the verb particle into. In verb-framed languages like Spanish (1b), the main verb is used to encode path, and manner is put into the gerund. Thus, in the framework of the current typology, languages are grouped into two types according to the distribution of path information on morphosyntactic elements.
- 2 For the event described in (1), French does not allow a similar construction as Spanish in (1b): ? (...)
4Our target languages, Japanese and French, are both classified by Talmy among the verb-framed languages. For example, we can see the same distribution pattern of path and manner information on the morphosyntaxtic elements.2
- 3 For glossing Japanese examples, segmentable morphemes are segmented by hyphens (‘-’), and clitic b (...)
(2)
|
a.
|
Je
|
suis
|
allé
|
en courant
|
à
|
la
|
bibliothèque
|
de
|
l’institut
|
français
|
|
|
I
|
AUX
|
go.PP
|
run.GER
|
to
|
the
|
library
|
of
|
the-institute
|
French
|
|
|
‘I ran to the Library of the French Institute’
|
(Olivier Emile, Mille eaux)
|
|
b
|
Watashi=wa
|
toshokan=ni
|
hashit-te
|
it-ta3
|
|
|
I=TOP
|
library=DAT
|
run-CVB
|
go-PST
|
|
|
‘I ran to the Library’
|
5The distribution of spatial information is identical in these languages; figure and ground are instantiated by nouns: je and bibliothèque in (2a), and watashi and toshokan in (2b). The path or arrival meaning is encoded in the main verbs aller and iku respectively, and manner information is expressed by the gerundive form of courir (en courant) and the converb form of hashiru (hashit-te). This comparison indicates that Japanese and French can be grouped into the same typological category.
6However, some differences can be observed between French and Japanese. In particular, the differences between these languages seem remarkable with respect to manner and deictic expressions. As for the manner expression, Slobin (2000, 2004, 2006) has proposed the saliency or facility of manner expression as another criterion for cross-linguistic comparison. With regard to French, Slobin (2000) says, for the situation in which a person ran into a house, that it is possible to say il est entré dans la maison en courant (he entered the house running), but the manner element is not frequently expressed. Instead, only a path element appears so that il est entré (dans la maison) is the most preferred expression. In contrast to French where only the path is preferably expressed without manner information, serial verb languages like Thai can express path and manner information in syntactically equivalent verbs, which leads to a third type of the typology: equipollently-framed language (Slobin 2004, Zlatev and Yangklang 2004).
7In Japanese, a manner expression seems to appear more readily than in French; for the same situation cited in the above paragraph where someone ran into a house, Japanese will allow the following construction.
(3)
|
a.
|
Kare=wa
|
ie=ni
|
kake-kon-da
|
|
|
he=TOP
|
house=DAT
|
run.CVB-enter-PST
|
|
|
‘He ran into the house. (lit. He entered the house running)’
|
8It seems easier to use the converb form of kakeru (kake=‘run’) than the French gerundive en courant, but the syntactic status of the path verb and that of the manner verb is not equivalent; it is possible to identify the path verb as the main verb and the converb as a subordinate. The Japanese converb form seems to differ from both serial verbs and the French gerundive.
9Moreover, some authors have already indicated another characteristic of French; viz. that constructions analogous to the satellite-framed languages are sometimes possible in French (cf. Kopecka 2004, 2006).
(4)
|
a.
|
Il
|
a
|
écouté
|
un
|
coup
|
de
|
fil
|
et
|
il
|
a
|
couru
|
dans
|
son
|
bureau
|
|
|
he
|
AUX
|
hear.PP
|
a
|
hit
|
of
|
line
|
and
|
he
|
AUX
|
run.PP
|
in
|
his
|
office
|
(4)
|
b.
|
??
|
Denwa=ga
|
naru=no=o
|
kii=te
|
kare=wa
|
office=ni
|
hashit-ta
|
|
|
|
phone=NOM
|
ring=NMR=ACC
|
hear-CVB
|
he=TOP
|
office=DAT
|
run-PST
|
- 4 The boundary-crossing or telic interpretation with a manner main verb is conditioned by the size o (...)
10It appears that the French manner verb courir (run) allows the construction of a satellite-framed type as in (4a), where the preposition dans (in) introduces a boundary-crossing path. In a strict sense, the French preposition is not a satellite but an adnominal element (prepositional phrase) because it must be followed by a noun phrase (*Il a couru dans). However, in (4a), the prepositional phrase dans son bureau, an element other than the main verb, conveys the meaning of translational motion from outside to inside of the office. This construction is not always realized by all manner verbs (#marcher dans son bureau ‘walk in the office’) and the understanding of the boundary-crossing meaning owes much to the size or enclosing property of the ground object (courir dans le bureau vs. courir dans la mer),4 but the prepositional phrases in this type of construction can be treated as an element comparable to the satellite (cf. Croft et al. 2000). In Japanese, on the other hand, the same construction is not as acceptable, and other complex predicates or complex verbs such as isoi-de it-ta (hurried to go) or kake-konda (enter running) are preferred.
11Thus, when it comes to the expression of manner, its realization does not seem the same, and Japanese and French exhibit different patterns of motion expressions; French can sometimes accept constructions analogous to the satellite-framed ones, expressing path by prepositional phrases as in (4a) (hereafter, we treat this construction as a type of satellite-framing), and Japanese has more subordinate verb forms to express manner.
12The second apparent difference comes from the deictic expressions. In Japanese, the combination of a path verb and a deictic verb is possible, while this does not seem possible in French.
(5)
|
a.
|
Dareka=ga
|
hai-te
|
ki-ta
|
|
|
someone=NOM
|
enter-CVB
|
come-PST
|
|
b.
|
*
|
Quelqu’un
|
est
|
venu
|
en entrant
|
|
|
|
someone
|
AUX
|
come.PP
|
enter.GER
|
|
c.
|
*
|
Quelqu’un
|
est
|
entré
|
en venant
|
|
|
|
someone
|
AUX
|
enter.PP
|
come.GER
|
|
|
|
‘Someone came in.’
|
13For example, when the speaker is located in an enclosure such as a room or house and describes someone else entering the room, s/he can say ‘dareka=ga hait-te ki-ta’ (‘someone came in’, lit. ‘came entering’) as in (5a). In French, the speaker cannot use two verbs entrer and venir in the same clause as shown in (5b), and venir may be preferred: ‘il y a quelqu’un qui est venu’ (there is someone who came).
14So, although the same structure is sometimes kept even with the deictic verbs as in (2), the deictic verbs also show a different distribution between Japanese and French. Deixis is currently considered as a component of the notion of path, but the above example (5) prompts us to investigate whether deixis really is a component of the path or not.
15The preliminary observation above suggests the existence of certain recurrent variations between French and Japanese, in particular involving the expression of manner and deixis. This paper presents a frequency analysis to verify if these variations are really typologically significant, and to elucidate the lexical, semantic, or syntactic factors bringing about these variations. For that, this study comprises two parts: collection of data and frequency analysis (section 2), and semantic and syntactic analysis on the factors triggering the recurrent variations (section 3).
16Certain previous studies have been centered on the intratypological variations. For example, Ibarretxe-Antuñano (2004, 2009) analyzes fine-grained path expressions in Basque by postpositional phrases and adverbials, which leads to a variation in the verb-framed languages and a similarity with the satellite-framed languages. Berthele (2004) considers a dialectal variation in German (satellite-framed), and Kita (1997) or Sugiyama (2005) show the frequent use of mimetics in Japanese to express manner with respect to Spanish or French. Also, Huang & Tanangkingsing (2005) demonstrate frequent co-expressions of manner and path in Tsou.
17Our study shares a common objective with these studies that analyze intratypological variations, and our target is centered on the intratypological variations between French and Japanese. In particular, in the following sections, we argue that the different frequency of manner expressions concerns the facility of the event integration itself (integration of a co-event into the motion event), and certain uses of the deictic verbs should be treated as non-path verbs. The intratypological variations do not constitute mere counterexamples against the current typology, but new parameters for cross-linguistic comparison helping us to understand the real states of languages.
- 5 For example, moguru in Japanese (to which correspond many English expressions such as dive, swim u (...)
18Before going into the analyses, we propose a categorization of motion verbs. As understood within the current framework, the notion of path is semantically defined as the route which a figure follows during a motion event, and it is constructed from three sub-constituents: “vector”, “conformation”, and “deixis” (cf. Talmy 2000: 53-57). Manner is also a conceptual element among several co-events such as cause, concomitance, or enablement. These semantic and conceptual definitions are applied to the categorization of motion verbs in accordance with intuitive comprehension, sometimes involving certain ambiguities.5 Instead, we propose another way of defining the path and manner verbs based on their aspectual properties.
- 6 Japanese =made and French jusqu’à can be used to mark the goal when the main verb is a manner verb (...)
19A correlation between lexical aspects and motion events has already been indicated by, for example, Cappelle & Declerck (2005), Levin & Rappaport Hovav (1992), Morita (2009), Sarda (1999), or Tenny (1994, 1995). In a word, the lexical aspect in terms of telicity correlates with the absence/presence of a boundary in motion events. For example, in the use of telic verbs such as entrer (enter), sortir (exit), partir (leave), or arriver (arrive), a boundary such as a source or goal is obligatory. Even with certain verbs representing a traversal path like franchir (get over), a boundary is present in space and the verb is telic. By contrast, atelic verbs like avancer (move forward), marcher (walk), suivre (follow), or courir (run) do not introduce any boundary except in the cases of active-accomplishment with quantified objects or internal objects (courir 100 mètres, ‘run 100 meters’), or of the delimitation of a traversal path with adpositions of the up to-type (jusqu’à in French, =made in Japanese).6
20Among atelic verbs, the notion of linearity serves to distinguish directional verbs from manner verbs. For example, the above-cited atelic verbs, avancer and marcher, are distinguished by their linearity. The verb avancer shows an affinity for linear or straight objects like rue (street) or chaussée (roadway), while marcher is indifferent to the form of ground objects. Likewise, the motion trajectory traced by suivre (follow) or longer (go along) is always determined by the form of a ground object, while courir (run) or nager (swim), which are exempt from linearity, can represent motion trajectories that are non-bounded by the form of ground objects. Other tests are available for Japanese, e.g. susumu (move forward) cannot construct a complex verb with mawaru (go around, *susumi=mawaru) and cannot appear with the locative particle =de instead of =o which is used to mark the traversal path (shadoo {=o/*=de} susumu, ‘move forward in the roadway’). Other verbs like hashiru (run) or aruku (walk) are not subject to these constraints; kooen {=o/=de} aruku ‘walk in the park’ and aruki=mawaru ‘walk around’ are acceptable. Verbs of the avancer type are categorized as directional in light of their lexicalized linearity, and other verbs without linearity are considered to express the manner of motion. The notion of manner or subclasses of the manner verbs are elaborated by Cifuentes (2010) or Ozcaliskan and Slobin (2003). For our purpose, a frequency analysis of the use of manner verbs and their combination with other verbs, we shall limit ourselves to distinguishing the manner verbs from the other path or directional verbs.
21Typical verbs are listed below.
(7)
|
a.
|
Path verbs
entrer (enter), sortir (exit), partir (leave), arriver (arrive), franchir (go over).
hairu (enter), deru (exit), shuppatsu-suru (leave), tsuku (arrive), koeru (go over).
|
|
b.
|
Intermediate class of path/directional
monter (ascend), descendre (descend), tomber (fall), passer (pass).
agaru (ascend), oriru (descend), kaeru (return), ochiru (fall), tooru (pass).
|
|
c.
|
Directional verbs
avancer (move forward), s’orienter (head for), suivre (follow), longer (go along).
susumu (move forward), mukau (head for), tsutau (follow).
|
|
d.
|
Manner verbs
marcher (walk), courir (run), nager (swim), voler (fly), glisser (slide).
aruku (walk), hashiru (run), oyogu (swim), tobu (fly), suberu (slide).
|
- 7 Slobin (2006) says that manner verbs are almost activity verbs; however, activity verbs are not al (...)
22In short, following the notional categorization proposed by, for example, Matsumoto (1997), path verbs range from (7a) to (7c), and manner verbs correspond to (7d). The category of path verbs is an amalgam of a variety of verbs, from telic (7a) to atelic directional verbs (7c) via an intermediate class (7b) that can be telic or atelic depending on the argument structure. Manner verbs are atelic verbs without linearity.7
- 8 The venitive verbs kuru and venir (come) show a difference between Japanese and French: venir can (...)
23As for the deictic verbs that are not listed in (7), they show different aspectual properties between French and Japanese.8 The French deictic verbs aller (go) and venir (come) require a goal, implicit or explicit, and appear always in telic events (il est venu/il est venu en France, *il est allé/il est allé en France). Consequently, they correspond to the category (7a) and they can be treated as path verbs. The Japanese deictic verbs iku (go) and kuru (come) frequently appear in the telic events with a goal as well as in French, but they have an atelic use with the marker of a traversal path =o. For example, phrases such as hosoi toori=o iku (go through a narrow street) or yamamichi=o kuru (come through a mountain road) can accept 30 pun=kan (for 30 minutes). Hence, the Japanese deictic verbs have a path verb property and a directional verb property. As long as the deictic verbs express telicity or linearity, they can be treated as path verbs in the same ways as the other path verbs listed in (7a) to (7c) (further analysis will be given in 3.2). The frequency analyses in the following sections follow this categorization.
- 9 For data sources, see the list of references at the end of this article. The parenthetical abbrevi (...)
24Our data is constructed from modern literary works and their translations (cf. Slobin 1996, 1997).9 We choose these works for the sake of availability and collection of various complex path, manner, and deictic expressions. The original Japanese texts comprise 2062 examples, and their translations in 1586 French examples. The original French data accounts for 1984 examples and their counterparts in 1777 Japanese examples. The different numbers of occurrences between the original texts and their translations arise from several reasons; for example, not all motion expressions are translated into the target language, and a motion event can be collapsed with the expression of another event.
25For the frequency analysis, first, all examples are sorted by the types of main verb in each language. Next, the corresponding translated examples are also examined according to the same criterion, that is, the information lexicalized in the main verb. Then, in order to take into account the path and manner expressed by subordinate verb forms, Japanese converbs (-i/-eform and -te form), French gerunds, or all the other verbal expressions (e.g., the concomitance marker =nagara, participles, and so on) are also identified. Path verbs are divided into those with deictic and non-deictic paths. In order to verify the patterns of expression found in each original text, the translations are also examined for the types of main verbs, frequency of subordinate elements, and combination of verbal means.
26Tables 1 and 2 show the distribution of motion information encoded in the main verb, in the original texts and their corresponding translations.
|
Japanese original (%)
|
French translation (%)
|
Path
|
49.4 (N=1018)
|
64.0 (N=1015)
|
Manner
|
13.0 (N=268)
|
14.8 (N=235)
|
Deixis
|
37.6 (N=776)
|
21.2 (N=336)
|
Total
|
2062
|
1586
|
Table 1: Types of main verbs: Japanese to French
|
French original (%)
|
Japanese translation (%)
|
Path
|
64.9 (N=1287)
|
56.8 (N=1010)
|
Manner
|
16.1 (N=319)
|
14.5 (N=258)
|
Deixis
|
19.0 (N=378)
|
28.6 (N=509)
|
Total
|
1984
|
1777
|
Table 2: Types of main verbs: French to Japanese
27As we see in these tables, the main verbs generally convey the notion of path in both Japanese and French, accounting for about 50% in Japanese and 65% in the French originals. In this respect the two languages are considered to have the same pattern of expression, as the current typology predicts.
- 10 Significance is calculated by the chi-square test.
28However, these tables also reveal two major differences: on the one hand, the use of path verbs is more frequent in French than in Japanese and on the other hand, the number of occurrences of deictic verbs also indicates a remarkable difference. As Tables 1 and 2 show, the use of deictic verbs is more frequent in Japanese (comparison between Japanese original and French original, X2=170.32, df=1, p<0.001),10 and this tendency is clearly retained in translation. The high frequency of French path verbs can also be observed in both tables (comparison between French original and Japanese original, X2=98.46, df=1, p<0.001). Given that the use of manner verbs is not frequent, French sentences including path verbs should be translated as deictic verbs or the complex predicates containing them in Japanese. In contrast, deictic verbs or the complex predicates containing them in Japanese would be lost in translations to French and reproduced only by path verbs, which leads to a high frequency of path verbs in French.
29Next, consider the distribution of path and manner information in subordinate verb forms. The tables below show the use of path and manner verbs in the subordinate positions when the main verb represents path, manner, and deixis respectively.
Main verb
|
Subordinate verb
|
Japanese original
|
French translation
|
|
Path
|
57
|
5
|
Path
|
Manner
|
161
|
27
|
|
Total
|
218
|
32
|
Table 3-1: Use of subordinate verbs with the main path verb in Japanese
Main verb
|
Subordinate verb
|
Japanese original
|
French translation
|
|
Path
|
1
|
0
|
Manner
|
Manner
|
22
|
6
|
|
Total
|
23
|
6
|
Table 3-2: Use of subordinate verbs with the main manner verb in Japanese
Main verb
|
Subordinate verb
|
Japanese original
|
French translation
|
|
Path
|
218
|
0
|
Deixis
|
Manner
|
75
|
0
|
|
Total
|
293
|
0
|
Table 3-3: Use of subordinate verbs with the main deictic verb in Japanese
Main verb
|
Subordinate verb
|
French original
|
Japanese translation
|
|
Path
|
3
|
51
|
Path
|
Manner
|
40
|
87
|
|
Total
|
43
|
138
|
Table 4-1: Use of subordinate verbs with the main path verb in French
Main verb
|
Subordinate verb
|
French original
|
Japanese translation
|
|
Path
|
3
|
5
|
Manner
|
Manner
|
23
|
30
|
|
Total
|
26
|
35
|
Table 4-2: Use of subordinate verbs with the main manner verb in French
Main verb
|
Subordinate verb
|
French original
|
Japanese Translation
|
|
Path
|
0
|
160
|
Deixis
|
Manner
|
4
|
58
|
|
Total
|
4
|
218
|
Table 4-3: Use of subordinate verbs with the main deictic verb in French
30At a glance, subordinate verbs are used more frequently in Japanese. Table 3-1 shows that when path verbs are used in the main verb position, many subordinate verbs are used to express manner, and Table 3-3 shows that when deictic verbs are used in the main verb position, path verbs appear frequently in the subordinate position. Corresponding subordinate elements do not appear on a number of occasions in the translated French texts.
- 11 When a manner verb is used in the main verb position in French, the majority of the subordinate ma (...)
31Tables 4-1 to 4-3 show the use of subordinate verbs in French: when the main verb is used to express path (Table 4-1), manner may be represented by a subordinate verb form,11 which is nevertheless less frequent than in Japanese. The starkest contrast between French and Japanese is observed when the main verbs are deictic verbs; the association of the deictic verbs with path verbs is very frequent in Japanese (Table 3-3), but non-existent in French (Table 4-3). In general, subordinate verbs are hardly used for path information in French, and they are limited to a few expressions such as en suivant (following). Subordinate manner verbs are not as limited as subordinate path verbs, but they are less frequent than Japanese subordinate manner verbs. These distribution patterns are all retained in each translated text; that is to say, Japanese subordinate elements are omitted or translated into the main verbs in French, and Japanese translations add path or manner information in subordinate verbs that are absent in the original French texts.
32Even in Japanese, the deictic verbs do not appear in subordinate positions, and subordinate path verbs are not used when the main verbs are manner verbs (see Table 3-2). This is because the complex predicates are always constructed in the order of manner, path, and deixis (e.g. hashiri yot-te kuru ‘run-approach-come’; cf. Matsumoto 1997). Few examples in Table 3-2 and Table 4-2 (translation) correspond to expressions containing a converb form of sou (sot-te, going along) or tsutau (tsutat-te, following) appearing for example as sot-te aruku (walk along).
33Possible combinations of verbs are summed up in Table 5.
|
Main verb
|
Subordinate verb
|
Examples
|
|
Path
|
Manner
|
traverser en courant (cross running)
|
French
|
Deixis
|
Manner
|
aller en courant (go running)
|
|
Manner
|
Manner
|
nager en ondulant (swim swaying)
|
|
Path
|
Path, Manner
|
toori-nukeru (go through and exit)
kake-komu (enter running)
|
Japanese
|
Deixis
|
Path, Manner
|
hai-te kuru (come in, lit. come entering)
arui-te iku (go walking)
|
|
Manner
|
Manner
|
tobi-haneru (jump and bound)
|
Table 5: Different combinations of main and subordinate verbs in French and Japanese
34Another difference arises from path information expressed by adpositions. A few prepositions functionally analogous to the satellite are observed in French, but any equivalent postpositions are not found in the Japanese data. The result is shown in Table 6.
|
Original (%)
|
Translation (%)
|
Total
|
Japanese
|
0
|
0
|
0
|
French
|
6.58 (N=21)
|
9. 36 (N=22)
|
16.94 (N=43)
|
Table 6: Occurrences of telic path expression by adpositions with manner main verb
35Given the occurrences of manner verbs in the main verb position in French (319 in the original and 235 in translation), the satellite-framed construction is not so frequent both in the original and translation data. The rate in translation is greater than that in the original, but there is no significant difference between them (X2=1.10, df=1, p=0.0295).
36The frequency analysis conducted so far has revealed three principal differences between Japanese and French: (i) the existence of the telic path expression by adpositions in combination with manner main verbs in French and its complete absence in Japanese, (ii) the higher frequency of the combination of manner verbs in the subordinate position with main path verbs in Japanese, and (iii) the more frequent use of deictic verbs in the main verb position with subordinate path verbs in Japanese. These differences conflict with the predicted distribution patterns of the verb-framed languages in the current typology. The analyses in the following sections of this paper elucidate the morphosyntactic or semantic factors triggering these differences.
37Firstly, our discussion is focused on (i) a French construction resembling that found in satellite-framed languages and (ii) the use of manner verbs in the subordinate position in Japanese. From a perspective on the co-expression of manner and path, the use of manner verbs pertains to a different degree of aptitude for the event integration. Then, from the same perspective, we shall explore the possibility of the lexical integration of manner and path in French.
38In examining the use of manner verbs, the first difference found between Japanese and French is the satellite-framed construction in French. As mentioned earlier, some authors have already noted this type of construction (cf. Aurnague 2008, Cummins 1996, Fong & Poulin 1998, Kopecka 2004). A few examples of the same type also appear in our data. For example:
(8)
|
a.
|
La
|
brume
|
bleue
|
qui
|
descend
|
des
|
montagnes
|
vers
|
|
|
the
|
mist
|
blue
|
REL
|
descend.PRES
|
of‑the
|
mountains
|
around
|
(8)
|
a.
|
le
|
soir,
|
à
|
la
|
façon
|
d’un
|
châle
|
qui
|
glisse
|
doucement
|
|
|
the
|
evening
|
at
|
the
|
way
|
of-a
|
shawl
|
REL
|
slide.PRES
|
softly
|
|
|
sur
|
des
|
épaules
|
|
(Linh)
|
|
|
|
|
on
|
INDEF
|
shoulders
|
|
|
|
|
(8)
|
b.
|
Yuugata=ni
|
naru=to
|
yama=kara
|
ori‑te
|
kuru
|
aoi
|
kiri
|
|
|
evening=DAT
|
become=as
|
mountain=ABL
|
descend‑CVB
|
come
|
blue
|
mist
|
|
|
kata=ni
|
yasashiku
|
suber‑i‑ochiru
|
shooru=no
|
yoona
|
(Linh‑tr)
|
|
|
shoulder=DAT
|
softly
|
slide‑CVB‑fall
|
shawl=GEN
|
like.as
|
|
|
|
‘Blue mist that descends from mountains around evening as if it were a shawl falling softly on the shoulder’
|
(9)
|
a.
|
J’ai
|
couru
|
dans
|
ses
|
phares
|
et
|
j’ai
|
réussi
|
à
|
|
|
I‑AUX
|
run.PP
|
in
|
its
|
headlights
|
and
|
I‑AUX
|
succeed.PP
|
to
|
|
|
compter
|
les
|
premiers
|
billets
|
(Neige)
|
|
|
count.INF
|
the
|
first
|
bills
|
|
|
|
‘I ran into the headlights and I could count the first bills.’
|
|
b.
|
Boku=wa
|
raito=ni
|
mukat‑te
|
hashir‑i
|
[…]
|
(Neige‑tr)
|
|
|
I=TOP
|
light=DAT
|
seek‑CVB
|
run‑CVB
|
[…]
|
|
|
|
‘I ran in the direction of the headlights, and […].’
|
(10)
|
a.
|
Akiko=wa
|
batabatato
|
daidokoro=kara
|
de‑te
|
[…]
|
(Crépuscule)
|
|
|
Akiko=TOP
|
MIMETIC
|
kitchen=ABL
|
exit-CVB
|
[…]
|
|
|
|
‘Akiko exited the kitchen clattering.’
|
(10)
|
b.
|
Elle
|
se
|
précipita
|
en dehors
|
de
|
la
|
cuisine
|
[…]
|
(Crépuscule-tr)
|
|
|
she
|
REFL
|
rush.PST
|
outside
|
of
|
the
|
kitchen
|
[…]
|
|
|
|
‘She rushed out of the kitchen.’
|
39The main verbs used in (8) to (10) are all manner verbs in French, and the prepositions introduce the source or goal of motion. In (8), the manner and path information is distributed between the main verb glisser(slide) and the preposition sur (on) respectively, and the figure châle (shawl) comes to have contact (the meaning of onto) with the ground épaules (shoulder). In the corresponding Japanese translation, the path verb ochiru (fall) is indispensable to express the same meaning of goal-directedness. Example (9) illustrates a gap between the meaning of an arrival involving a boundary crossing in French and a mere directed motion in Japanese (=ni mukat-te hashiru ‘run in direction of’). The figure ran into the zone illuminated by car headlights in French (metonymic expression), while Japanese needs a path verb such as hairu (enter) or complex verb with -komu (enter) in order to represent the same situation. In Japanese-French translation (10), the Japanese mimetic adverb batabata (clattering) and the path verb deru (exit) are translated into the manner verb se précipiter and the prepositional phrase en dehors de in French. Japanese is a language where mimetics are very rich to express manner of motion (cf. Kita 1997, among others) and other mimetics are also found in our data, but the mimetics are syntactically adverbs and the predicted verb-framed construction is surely kept. In French translation, the main verb is a manner verb, and the meaning of translational motion or the telicity of the event is due to the prepositional phrase.
40These examples clearly show that Japanese keeps the distribution pattern of motion information of the verb-framed languages, which French does not always follow. As mentioned earlier, the understanding of French prepositional phrases as the satellites is conditioned by the size or the enclosed property of ground objects: the two-dimensional limited surface of shoulder in (8a), the small enclosed area illuminated by headlights in (9a), and the enclosed kitchen in (10b). Nevertheless, the function of the prepositional phrases cannot be ignored. They certainly introduce a goal or a source, and make events telic that would be otherwise atelic. They are not satellites in a strict sense, but they certainly express the path meaning comparable to onto, into, or out in English.
41This construction analogous to the satellite-framed construction is, however, limited to a very small number of examples in French (see Table 5). We have shown elsewhere (cf. Morita 2008), based on data from Frantext,12 that this construction is possible with certain manner verbs such as glisser (slide), sauter (jump), rouler (roll), courir (run) or se précipiter (rush), but it is not, or hardly, possible with other verbs such as déambuler (wander), errer (roam), flâner (stroll), flotter (float), or nager (swim). More precisely, among manner verbs, the satellite-framed construction is concerned with ballistic motion (cf. Slobin 2004, 2006) or manner hyperonyms. For example, marcher (walk) can be used with the preposition sur (on) or dans (in) so that a boundary-crossing meaning could be expressed as in marcher dans une flaque d’eau ‘walk into a puddle’ in spite of the lack of force dynamics compared to courir (run). However, another verb trotter (trot), which is more dynamic than marcher, cannot be used to represent the meaning of arrival without jusqu’à (up to) ‘*trotter à la gare’ (trot to the station). With hyponyms that describe more detailed manner of motion compared to marcher such as déambuler, errer, flâner (which, in addition to the manner of marcher, express purposeless, non-directed, time-consuming motion), prepositions cannot convey the dynamic meaning involving arrival or boundary-crossing.13 Conforming to the prior observation,our data includes glisser, courir, sauter, se précipiter, se presser, foncer (dash), and se ruer(rush). Consequently, compared to English manner verbs allowing the satellite-framed construction even with wander, stroll, or roam, the characteristics of French as a satellite-framed language are more limited.
42Concerning the difference between Japanese and French with regard to the “satellite-construction” (French) and the dynamic meaning of adpositional phrases, we conclude that although it is not so frequent, this construction is partially possible in French while it is completely excluded in Japanese. As a result, French cannot be unambiguously grouped with verb-framed or satellite-framed languages, but is a language that accepts, albeit partially, both types of constructions. Although the verb-framed construction is clearly dominant in French and prepositional phrases are not satellites in a strict sense, the possibility of the construction analogous to the satellite-framed construction constitutes a parameter by which Japanese and French are typologically distinguished.
43The satellite-framed construction allows co-expression of manner and path because manner is expressed by the main verb and path by other elements than the main verb. The converb construction frequently used in Japanese is another way for expressing path and manner information at the same time. In addition, Table 3-1 and Table 4-1 confirm that the use of subordinate verbs is much more frequent in Japanese than in French. This suggests that Japanese is a language suited for the co-expression of manner and path, and the following examples demonstrate this difference between French and Japanese.
(11)
|
a.
|
Otowa=ga
|
dote=o
|
kake‑ori,
|
[…]
|
(Bouddha)
|
|
|
Otowa=TOP
|
bank=ACC
|
run.CVB‑descend.CVB
|
[…]
|
|
(11)
|
b.
|
Otowa
|
descendit
|
le
|
talus
|
en courant
|
(Bouddha‑tr)
|
|
|
Otowa
|
descend.PST
|
the
|
bank
|
run.GER
|
|
|
|
‘Otowa ran down the bank.’
|
(12)
|
a.
|
Minoru=wa
|
sore=ni
|
kake‑yot‑te,
|
[…]
|
(Bouddha)
|
|
|
Minoru=TOP
|
it=DAT
|
run.CVB‑approach‑CVB
|
[…]
|
|
(12)
|
b.
|
Minoru
|
fut
|
incapable
|
de
|
s’en
|
approcher
|
[…]
|
(Bouddha-tr)
|
|
|
Minoru
|
be.PST
|
incapable
|
of
|
REFL‑there
|
approach.INF
|
[…]
|
|
|
|
‘Minoru could not approach there.’
|
(13)
|
a.
|
Quantité
|
d’oiseaux
|
tourbillonnent
|
et
|
plongent
|
parfois
|
dans
|
les
|
|
|
many
|
of birds
|
hover.PRES
|
and
|
dive.PRES
|
sometimes
|
in
|
the
|
(13)
|
a.
|
eaux
|
du
|
port
|
pour
|
en
|
ressortir
|
[…]
|
(Linh)
|
|
|
waters
|
of the
|
port
|
for
|
from there
|
again.exit.INF
|
[…]
|
|
|
|
‘Many birds hovering in the sky, sometimes dive into the port and exit.’
|
(13)
|
b.
|
Takusan=no
|
tori=ga
|
senkai‑shi,
|
tokiori
|
minato=no
|
|
|
Many=GEN
|
birds=NOM
|
hovering‑do.CVB
|
sometimes
|
port=GEN
|
(13)
|
b.
|
mizu=ni
|
mogut‑te=wa
|
[…]
|
tob‑i‑de‑te
|
kuru
|
(Linh‑tr)
|
|
|
water‑DAT
|
dive‑CVB=TOP
|
[…]
|
fly‑CVB‑exit‑CVB
|
come.PRES
|
|
|
|
‘Many birds hovering in the sky, sometimes dive into the port and fly out.’
|
- 14 The function of -i/-e converbs is very diversified. In example (11a), kakeru (run) and oriru (desc (...)
44Japanese has two types of verb forms (indicated as converb in the examples) which are a part of verbal conjugation and which can precede another verb to constitute a complex verb: -i and -e.14 In (11a), the path is encoded in the second head verb oriru (descend), and the manner is represented by the first subordinate verb kakeru (kake= ‘run’). This structure is transposed directly onto the corresponding French translation in (11b); the main verb descendre serves to express path information and the gerundive en courant expresses manner information. This is a case where the same distribution pattern appears in Japanese and French, as predicted.
45However, the same correspondence is not observed for the bulk of the data. Example (12) is a typical case of the deletion of manner information in French; although the complex verb kake-yoru (approach running) is comprised of a manner verb and a path verb, its French counterpart keeps only a path verb s’approcher (approach). To assure the informational equivalence, it is not impossible to use, for example, a prefixed manner verb accourir (courir ‘run’ + prefix ac- representing arrival) or the gerundive form en courant, but the data shows that manner information hardly appears in French. Example (13) is another case of semantic discrepancy; that is, the addition of manner information in Japanese that is absent in the original French text. The birds simply ‘exit’ from the water in French, but they ‘exit flying’ in Japanese.
46This difference is a consequence of the syntactic availability of each language. Japanese has several subordinate verb forms such as -i/-e converbs or the -te converb that constitute complex verbs or complex predicates, or the concomitance form -nagara. In contrast, French has only a gerundive form which is relatively independent or “heavy” in that it corresponds sometimes to a clausal expression such as quand il est sorti du cinema/en sortant du cinema (when he went out of the movie theater/going out of the movie theater). Slobin (2006) mentions that “it is not satellite-framing alone that accounts for the rate of use of manner verbs; morphosyntactic structure and lexical availability also contribute to a language’s ‘rhetorical style’ ” (Slobin 2006: 69). The different frequency of manner verbs in Japanese and French is a clear case of different morphosyntactic availability leading to a variation in verb-framed languages (cf. Huang & Tanangkingsing 2005).
47In examples (11) to (13), all sentences contain a path verb in the main verb position and therefore the verb-framed construction is retained. Typologically, path information is preponderantly important because it pertains to other constructions such as a resultative or change of state (a motion event is a type of macro event). However, with respect to the quantity of motion information or the co-expression of manner and path, Japanese seems closer to English than to French: tobi-deru (=fly out), kake-yoru (=run to), kake-oriru (=run down), etc.
48As mentioned earlier, intratypological variations appear in various constructions of various languages (cf. Ibarretxe-Antuñano 2004, 2009, Berthele 2006, Huang & Tanangkingsing 2005, Sugiyama 2005, etc.). The present analysis provides another perspective on the inter/intratypological variation: aptitude for the event integration. The satellite-framed construction allows the co-expression of manner and path information, and two events, a motion event and a co-event, can readily be integrated into one clause. The converb construction is another way to integrate two events in one clause, and Japanese is a converb abundant verb-framed language that is apt for event integration: at the level of aptitude for event integration, Japanese and English can be grouped into the same type. By contrast, since the expression of manner by a subordinate verb form is rare and the satellite-framed construction is not frequent, the co-expression of manner and path is very difficult in French. This means that in the framework of the event integration, the association of a motion event and a co-event itself is difficult in French. Although the encoding strategy for path information is identical between Japanese and French, the aptitude for the event integration constitutes one important parameter for the distinction between languages in which manner and path are encoded at the same time and languages in which only the path notion is represented (cf. Bohnemeyer et al. 2008).
49In the preceding discussion, we have argued that the syntactic strategies for co-expression of path and manner in French are not commonly used. However, French seems to have another means of co-expression, that is, the integration of path and manner in a single morpheme.
(14)
|
a.
|
On dirait
|
des
|
étoiles
|
tombées
|
à
|
terre
|
et
|
qui
|
cherchent
|
|
|
It seems
|
INDEF
|
stars
|
fallen
|
at
|
ground
|
and
|
REL
|
Search.PRES
|
(14)
|
a.
|
à
|
s’envoler
|
de nouveau
|
vers
|
le
|
ciel
|
(Linh)
|
|
|
at
|
REFL‑off.fly.INF
|
again
|
for
|
the
|
sky
|
|
(14)
|
b.
|
Chijoo=ni
|
ochi‑ta
|
hoshikuzu=ga
|
mata
|
sora=ni
|
mukat‑te
|
|
|
ground=DAT
|
fall‑PST
|
stardust=NOM
|
again
|
sky=DAT
|
seek‑CVB
|
(14)
|
b.
|
tob‑i‑tato-o=to
|
shi‑te
|
iru
|
yoo=ni=mo
|
mieru
|
(Linh‑tr)
|
|
|
fly‑CVB‑leave‑INT=QOT
|
do‑CVB
|
PROG
|
aspect=DAT=EMPH
|
seem.PRES
|
|
|
|
‘It seems fallen stars try to fly off again for the sky’
|
(15)
|
a.
|
J’ai
|
grimpé
|
sur
|
la
|
congère
|
[…]
|
(Neige)
|
|
|
I-AUX
|
Climb.PP
|
on
|
the
|
snowdrift
|
[…]
|
|
(15)
|
b.
|
Boku=wa
|
[…]
|
yukidamari=o
|
yoji‑nobor‑i
|
[…]
|
(Neige‑tr)
|
|
|
I=TOP
|
[…]
|
snowdrift=ACC
|
cling‑ascend‑CVB
|
[…]
|
|
|
|
‘I climbed up the snowdrift.’
|
(16)
|
a.
|
Les
|
torrents
|
qui
|
dévalent
|
la
|
montagne
|
[…]
|
(Linh)
|
|
|
the
|
torrents
|
REL
|
Hurtle.PRES
|
the
|
mountain
|
[…]
|
|
|
b.
|
Kyuuryuu=ga
|
yama=o
|
kake‑or‑i
|
[…]
|
(Linh‑tr)
|
|
|
torrent=NOM
|
mountain=ACC
|
run.CVB‑descend‑CVB
|
[…]
|
|
|
|
‘The torrents quickly run down in the mountain.’
|
50The verb used in (14a) is the prefixed verb s’envoler (fly off) in which the division of labor can be observed; the prefix en corresponds to the path ‘off’, and the root voler to the manner ‘fly’. Most French prefixes are no longer productive (e.g. accourir, emporter/apporter ‘take/bring’); however, it is possible to identify their meaning analytically, by comparing for example the prefixed verb s’envoler with the corresponding simple verb voler. In this case, manner and path information are not superimposed on a single morpheme, but distributed separately on prefix and root. Here, the “manner/result complementarity” (Rappaport Hovav & Levin 2010) is maintained.
51The situation is not the same for (15a) and (16a). The main verb of (15a), grimper, exhibits an upward motion in the same way as monter (ascend). These two verbs share the same aspectual properties (capable of being telic or atelic depending on the argument structure) so that they can be classified as path verbs. However, French speakers feel a sense of the effort for the motion of grimper, while a neutral upward motion for monter. In addition, when grimper is followed by prepositions, the motion involves the action of grasping a ground object with the hands (grimper à la corde, shin up the rope). In this case, grimper and monter are clearly different in their manner of motion (cf. Hickmann & Hendricks 2010). French has other verbs that represent an upward motion such as gravir or escalader, and they are distinguished by either their ground configuration or manner of motion.
52The same type of superimposition of manner and path on a single morpheme appears in (16a). The main verb dévaler represents a downward motion and potentially competes with descendre (descend) or dégringoler (fall down, tumble down), and they all have the same aspectual properties. From a point of view of taking speed as a component of manner (cf. Slobin 2006), dévaler expresses a quicker downward motion than descendre, involving other manner information such as violence or roughness. If the intentionality is a part of manner information, dégringoler also lexicalizes a path meaning and a manner, expressing the same direction as descendre and conveying the notion of “fall” or “tumble” which implies uncontrolledness or unintentionality. None of these verbs follow the principle of “manner/result complementarity”. In the same way that the English climb constitutes an exception to this hypothesis, the above-cited French upward and downward motion verbs are not in accord with this hypothesis.
53In order to reproduce the same motion described by these verbs, Japanese uses a more productive form of composition: yoji-noboru (cling-ascend), hai-agaru (creep-ascend), kake-oriru (run-descend), koroge-ochiru (roll-fall), etc. Here, the hypothesis of manner/result complementarity is confirmed perfectly. Probably, this hypothesis is more efficient in languages where the division of labor of manner/result is realized by means of converb constructions or verb root/satellite distribution. In other languages like French where converb construction is not very developed and a preposition is not so powerful as to introduce the result of motion, the path and manner information may be concentrated in one morpheme.
54We shall now examine deixis. According to our criterion for the categorization of motion verbs, the deictic verbs can be treated as path verbs as long as they introduce telicity or linearity in motion events. However, the Japanese deictic verbs iku (go) and kuru (come) sometimes seem to convey deictic direction only when they are composed with other telic or directional verbs. The possibility of the combination of a telic or directional path verb with the deictic verbs in Japanese and the impossibility of such combination in French will shed light on the analysis of the deictic verbs as path verbs.
55The deictic verbs orient motion events from the perspective of the speaker, and at the same time confer the telicity to motion events. The latter function is a property of path verbs, so that the deictic verbs can be treated as path verbs.
56This double function of the deictic verbs is observed in both Japanese and French.
(17)
|
a.
|
Taroo=ga
|
eki=ni
|
it‑ta
|
/
|
ki‑ta
|
|
|
Taroo=NOM
|
station=DAT
|
go‑PST
|
/
|
come-PST
|
|
|
‘Taro went/came to the station.’
|
(17)
|
b.
|
Paul
|
est
|
allé
|
/
|
venu
|
à
|
la
|
gare
|
|
|
Paul
|
AUX
|
go.PP
|
/
|
come.PP
|
at
|
the
|
station
|
|
|
‘Paul went/came to the station.’
|
57Example (17) shows the arrival of the figure at the station. At the same time, each verb represents deictic direction. When the speaker utters iku or aller, (s)he must not be at the station. If the speaker uses kuru or venir, (s)he must be at the station. This is a general use of the deictic verbs in Japanese and French, and the same use is found in the bulk of our data.
58It is well known that manner verbs cannot be used with the goal marker =ni in Japanese and à in French (cf. Cummins 1996, Kageyama 1996, Matsumoto 1997). These goal markers turn out to be possible by adding a deictic verb.
(18)
|
a.
|
*
|
Taroo=ga
|
eki=ni
|
arui‑ta
|
|
|
|
Taroo=NOM
|
station=DAT
|
walk‑PST
|
|
b.
|
|
Taroo=ga
|
eki=ni
|
arui‑te
|
{
|
it‑ta
|
/
|
ki‑ta
|
}
|
|
|
|
Taroo=NOM
|
station=DAT
|
walk‑CVB
|
{
|
go‑PST
|
/
|
come‑PST
|
}
|
|
|
|
‘Taroo went/came to the station on foot.’
|
59Example (18a) shows that when the main verb is the manner verb aruku (walk), the sentence is not so acceptable. With a deictic verb that makes the motion event telic, the sentence is quite natural, as in example (18b).
60In French, the combination of marcher and aller/venir is not used because a prepositional phrase à pied (on foot) is preferred, but the same phenomenon occurs, for example, with other manner verbs such as courir (run).
(19)
|
a.
|
??
|
Il
|
a
|
couru
|
à
|
la
|
gare
|
|
|
|
he
|
AUX
|
run.PP
|
to
|
the
|
station
|
(19)
|
b.
|
??
|
Il
|
est
|
{
|
allé
|
/
|
venu
|
}
|
à
|
la
|
gare
|
en courant
|
|
|
|
he
|
AUX
|
{
|
go.PP
|
/
|
come.PP
|
}
|
to
|
the
|
station
|
run.GER
|
|
|
|
‘He went/came to the station running.’
|
|
|
|
61Although courir is one of the verbs that allow a construction analogous to the satellite-framed one, (19a) is not so plausible out of context (compare with (4)). The addition of the deictic verbs settles this oddity, which proves that deictic verbs include the property of path verbs.
62What is to be emphasized here is that the deictic verbs introduce telicity in addition to deictic direction. That is to say, as long as they exhibit the same aspectual property as the other path verbs, deictic verbs can be considered as path verbs.
63As mentioned earlier, in Japanese, the deictic verbs appear in the main verb position, frequently following other directional or path verbs in the -te converb.
(20)
|
a.
|
Inu=ga
|
[…]
|
shibaraku
|
wareware=no
|
ato=o
|
|
|
dog=NOM
|
[…]
|
briefly
|
us=GEN
|
behind=ACC
|
(20)
|
a.
|
tsui-te
|
ki-ta
|
ga, […]
|
(Norway)
|
|
|
follow-CVB
|
come-PST
|
but
|
|
(20)
|
b.
|
Le
|
chien
|
[…]
|
nous
|
suivit
|
un
|
moment,
|
mais […]
|
(Norway-tr)
|
|
|
the
|
dog
|
[…]
|
us
|
follow.PST
|
a
|
moment
|
but
|
|
|
|
‘The dog followed us for a time, but […]’
|
- 15 [yuku] is an allophone of the deictic verb iku.
(21)
|
a.
|
Akiko=ga
|
kaidan=o
|
ori-te
|
yuki,15 […]
|
(Crépuscule)
|
|
|
Akiko=NOM
|
stairs-ACC
|
descend-CVB
|
go.CVB
|
|
(21)
|
b.
|
Elle
|
descendit,
|
(Crépuscule-tr)
|
|
|
she
|
descend.PST
|
|
|
|
‘Akiko descended the stairs.’
|
64Unlike the examples in (18b) or (19b), here the event remains atelic in spite of the deictic verb in (20). In this example, the relation between the two entities wareware (us) and inu (dog) is expressed and the motion of inu is deictically directed to wareware. In Japanese, as long as the scene is described from the perspective of wareware, the use of kuru is obligatory. If the motion is described by the viewpoint of “dog” or neutral perspective, tsui-te iku (follow-cvb go) is to be used. The precedent verb tsuku (follow) cannot be used autonomously and is always followed by iku or kuru, but what is important here is that the physical relation between the follower and the one being followed is the same even though the deictic verb changed, and the telicity remains invariant: atelic. The deictic verbs do not have an effect on the physical linear relation or on the telicity, but direct the motion from the viewpoint of a speaker or narrator. In French, the viewpoint will be differentiated by the alternation of figure/ground and voice: le chien nous suivit (the dog followed us) or nous fûmes suivis par le chien (we were followed by the dog).
65In the same way, the motion event in example (21a) is expressed from the perspective of someone other than Akiko: someone upstairs or from other neutral perspectives, and the viewpoint must not be located downstairs. The subordinate verb oriru expresses a downward motion, and the main verb yuku (go) orients this downward direction deictically. In this case again, French deictic verbs do not appear and the combination of descendre/aller provokes strangeness in (21b): *elle alla en descendant (she went descending), *elle descendit en allant (she descended going). As a result, French uses only path verbs in the main verb position, as shown in Tables 3 and 4, in contrast to the high frequency of subordinate path verbs in Japanese.
66What is noteworthy in (20) and (21) is that the deictic verbs do not describe physical motion in space. Indeed, deixis is a kind of direction, but it does not arise from the physical relation between a figure and a ground; it is rather determined by the viewpoint of a speaker or narrator. Therefore, the functional level of a deictic direction is distinct from that of a physical direction in the sense of linearity determined by the relation between a figure and a ground.
67More striking differences are observed for Japanese complex predicates comprising telic path verbs and their French counterparts. In Japanese, the deictic verbs behave purely as an indicator of deictic direction and do not concern the achievement of physical motion.
(22)
|
a.
|
Boku=ra=ga
|
shokuji=o
|
shi-te
|
iru
|
aida=ni=mo
|
nan’ninka=ga
|
|
|
I=PL=NOM
|
meel=ACC
|
do-CVB
|
PROG
|
during=DAT=EMPH
|
some.people=NOM
|
(22)
|
a.
|
hait-te
|
ki-te,
|
nan’ninka=ga
|
de-te
|
it-ta
|
(Norway)
|
|
|
enter-CVB
|
come-CVB
|
some.people.NOM
|
exit-CVB
|
go-PST
|
|
(22)
|
b.
|
Pendant
|
que
|
nous
|
mangions,
|
des
|
gens
|
entrèrent,
|
d’autres
|
sortirent
|
|
|
during
|
that
|
we
|
eat.PST
|
INDEF
|
people
|
enter.PST
|
others
|
exit.PST
|
|
|
‘Even while we were eating, some people entred, others exited.’
|
68In (22a), the motion event is described from the perspective of boku=ra (us) who are positioned in the dining room. The deictic verbs play no role in terms of the expression of physical motion because they do not change telicity or linearity, but introduce purely deictic direction. Even removing the deictic verbs, the same physical motion event can be described.
(23)
|
nan’ninka=ga
|
hait-te,
|
nan’ninka=ga
|
de-ta
|
|
|
some.people=NOM
|
enter-CVB
|
some.people=NOM
|
exit-CVB
|
|
|
‘Some people entered, others exited.’
|
69The difference between (22a) and (23) is whether the speaker’s viewpoint is explicit or not. In other words, the event in terms of the physical motion of a figure is accomplished by the subordinate path verbs in (22a), and the deictic verbs have a relation only with the viewpoint of the speaker or principal character in the novel. Once again, the function level of the deictic verbs, based on the cognition of direction by the speaker, is different from that of the other path verbs.
70This combination of path and deixis is completely absent from the French data. The use of venir and partir (leave) allow us to grasp the same deictic direction as example (22a), in exchange for the boundary-crossing meaning conveyed by entrer and sortir.
(24)
|
a.
|
*
|
Des
|
gens
|
entrèrent
|
en venant,
|
d’autres
|
sortirent
|
en allant/partant
|
|
|
|
INDEF
|
people
|
enter.PST
|
come.GER
|
others
|
exit.PST
|
go/leave.GER
|
(24)
|
b.
|
*
|
Des
|
gens
|
vinrent
|
en entrant,
|
d’autres
|
allèrent/partirent
|
en sortant
|
|
|
|
INDEF
|
people
|
come.PST
|
enter.GER
|
others
|
go.PST/leave.PST
|
exit.GER
|
(24)
|
c.
|
*
|
Des
|
gens
|
vinrent,
|
d’autres
|
partirent
|
|
|
|
INDEF
|
people
|
come.PST
|
others
|
leave.PST
|
|
|
|
‘Some people came, others left.’
|
71French deictic verbs are always path verbs introducing telicity. In other words, once the telicity is introduced by path verbs, another telic element cannot be added into the same clause. Therefore, French does not have an example of the co-expression of path and deixis, as shown in Table 4-3.
72This difference can be attributed to the varying degree of grammaticalization of deictic verbs in the two languages. According to Shibatani (2007), Japanese deictic verbs show a diverse degree of grammaticalization in accordance with verbs that precede them. Shitabani argues that the deictic verbs retain spatial meaning when they are combined with the manner of motion verbs, and that they are more grammaticalized when they follow other motion verbs. His analysis accords with our discussion based on the telicity of events and the linearity exhibited by the deictic verbs. The French deictic verbs are also grammaticalized to express near future/past or intention, but they are not pure indicators of deictic direction as for the expressions of motion events.
73The distinct characteristics of the deictic verbs in Japanese and French have two consequences for our understanding of the current typology. First, Japanese deictic verbs are not always path verbs. Although deixis is treated as one of the path components in Talmy’s framework, Japanese deictic verbs sometimes function exclusively at the discursive or cognitive level to orient the motion from the viewpoint of a speaker. This characteristic is clearly different from the other path components, the vector and conformation that have physical properties in space: departure, traversal, arrival, two/three dimensional ground object, surface support or the containing/contained relation between figure and ground, etc. Deixis appearing in complex predicates in Japanese does not pertain to these spatial characteristics, and it is no longer a component of path in the sense of being a spatial property.
74Second, deixis, exempt from spatial property, is encoded in the main verb position, and the path in the sense of telicity or linearity is represented by the subordinate verb in Japanese. This distribution of information does not accord with the typical instantiation of motion events in verb-framed languages. In French, the deictic verbs always serve to introduce the telicity in motion events and appear in the main verb position, which conforms to the prediction in the current typology.
75This is, needless to say, an understanding of Japanese deictic verbs limited to the case of complex predicates with path verbs; the deictic verbs can be used without other subordinate verbs and retain, for the most part, the properties of path verbs as shown in examples (17) and (18). Our data, however, shows a non-negligible difference of frequency between Japanese and French for the frequency of use of deictic verbs and subordinate forms of path verbs, which does not accord with the predicted distribution in the current typology. In short, Japanese complex predicates comprising a path verb and a deictic verb represent a peculiar construction; Japanese can be classified neither in the verb-framed languages nor in the satellite-framed languages when the deictic verbs follow other directional or path verbs because the main verb represents the speaker’s viewpoint and the subordinate verb encodes path notion. On the other hand, French deictic verbs always keep the path verbs’ property and fit with the verb-framed construction.
76From the beginning, research on the linguistic expressions of motion events has been based on, or focused on, the interface between the perception of extra-linguistic motion events and their linguistic representations, and the motion events have a model comprising of a figure, ground, motion, path, and co-event, which omits other social participants such as a speaker or listener. Looking at it in this way, the extra-linguistic notion and the purely linguistic or social notions such as a speaker, listener, or deixis, must be distinguished. To speak of deixis, the current model of a motion event is not sufficient and the linguistic or social dimensions must be accounted for (cf. Gathercole 1978).
77By analyzing the co-expression of manner/path and path/deixis, we have shown significant differences between Japanese and French. What are the implications of these differences, which deviate more or less from the current typology?
78Briefly, these differences do not constitute simple counterexamples that do not accord with the current typology, but provide parameters by which Japanese and French can be correctly compared and compensate intratypologocal variations. In previous research on the motion events, the syntactic realization of path information has been the principal parameter for comparison between languages, and languages have been divided into two types. However, when this criterion is treated as only one parameter among others, a more accurate account is possible. Verb-framing and satellite-framing are not language types, but construction types (cf. Croft et al. 2010, Huang & Tanangkingsing 2005); one of them may exclusively appear in one language, and both of them may co-exist in another language. In French and Japanese, except complex predicates comprising deictic verbs, Japanese motion events follow almost exclusively the verb-framed construction, and French partially allows the two types of framing patterns.
79Other differences between Japanese and French are expressed in a similar parametric way. The co-expression of manner and path, or path and deixis, constitutes other parameters for comparison, which return specific values for each language.
|
|
Manner/path co-expressions
|
|
|
V‑framed construction
|
S‑framed construction
|
Path and Manner on single lexeme
|
V‑framed + sub. manner verbs
|
Deixis + other paths
|
Japanese
|
√
|
×
|
×
|
√
|
√
|
French
|
√
|
−
|
−
|
−
|
×
|
√: possible, ×: impossible, −: not impossible
|
Table 7: Parameters for Japanese and French
80To summarize, if we ignore other parameters and only consider the verb-framed construction, then Japanese and French could be grouped into the same category. With respect to the other parameters, they show either slight or even striking differences. If we do not account for all of these parameters, the distinctions between Japanese and French are not correctly understood and cross-linguistic comparison would become more difficult.
81The morphosyntactic realization of the path information is the most important parameter given that it is pertinent to other constructions such as a resultative or change of state. In this respect, one of our criteria — “verb-framed + subordinate manner verbs” — would not be so decisive because the main verb always encodes the path component. However, this criterion should be considered as a type of “manner/path co-expressions” that pertains to the ease of the event integration itself. Table 7 correctly shows that French does not develop a means for manner/path co-expressions.
82Table 7 may account for some related constructions. For instance, in a transitive resultative like ‘hit something broken’, a complex predicate tataki-kowasu (break hitting) is available in Japanese, while in French, it is preferable to choose casser (break) or frapper (hit) and the expression of the result is generally preferred (casser en frappant ‘break hitting’ is not impossible, but hardly used). By contrast, in a caused motion event like ‘pull something out of somewhere’, a construction analogous to the satellite-framing ‘tirer (pull) something de (from) somewhere’ does not seem strange in French, and another complex predicate hippari-dasu (put out pulling) is available in Japanese. The distinction between languages like Japanese in which the manner/path or cause/result relation can readily be expressed in one clause, and other languages like French in which only one information tends to appear or more clauses are needed to express manner/path information, may be considered as a parameter for cross-linguistic comparison (cf. Bohnemeyer et al. 2008).
83The expression of manner and deixis highlights the differences between Japanese and French. In verb-framed languages, the expression of manner pertains to the ease of event integration, and Japanese deictic verbs reveal another level of motion event; these verbs function at the discursive or cognitive level of a speaker that cannot be reduced to the current model of motion events.
84The parametric approach summarized in Table 7 accommodates these differences and the traditional similarity of verb-framed languages. However, many problems still remain. The parameters that we have proposed are merely based on the data of autonomous motion events in Japanese and French literary works. When other languages or other constructions such as caused motion events or resultatives are taken into consideration, the number of parameters is expected to increase. Moreover, our data set is suitable for the analysis of complex structures that seldom appear in conversation, but it is not appropriate for investigating the cognitive aspects of language use. For example, to examine the use of the deictic verbs under controlled experiments will allow us to take further steps. Similarly, the relation between syntactic dependency and narrative production, or between cognitive cost and fore/back grounding of linguistic expressions (cf. Talmy 2000), is also a question to be challenged.
85Although we could not treat these cognitive aspects in this paper, we observed a correlation between syntactic dependency and the frequency of manner expressing subordinate elements has been confirmed. For example, among Japanese i/e-converbs, te-converbs, and French gerundives, i/e-converbs are more closely associated to the main verb morphosyntactically, and they are the most frequent in literary data. On the other hand, the French gerundive is syntactically the most independent element and appears least in our data. This paper has treated only syntactic and semantic aspects of motion events in Japanese and French collected from literary works (therefore different from the Talmy’s typology aimed at colloquially frequent expressions), and it is necessary, needless to say, to perform further experiments and analyses in order to approach the cognitive problems of languages.