This paper addresses the semantics of the present form (known as the -ru form) of activity verbs in Japanese and examines the effect of these verbs in contrast to that of the inflected form (the -ta form). Garden-path sentences involving an ambiguity between a simple sentential reading and a relative clause reading generally show a preference for the former reading; when the preferred reading proves to be inconsistent with the correct reading of the sentence, the ensuing processing difficulty is known as the 'gardenpath effect.' Interestingly, it has been observed that the effect is reduced in sentences that contain activity verbs in the present -ru form in the adnominal clause. One major problem that arises in the interpretation of the -ru form in subordinate clauses is that it is temporally ambiguous, and may be interpreted as belonging to the matrix clause instead. To date, no uniform analysis has been developed to characterize the semantic nature of the -ru form in subordinate clauses. The main goal of this study was to develop a semantics of the -ru form. The data revealed some interesting findings suggesting some logical characteristics of the -ru form. The results help to clarify how the semantic nature of the -ru form exerts an effect on the processing of garden-path sentences.
This study examines Japanese garden-path sentences that
contain an active verb in an adnominal clause. Usually, an
adnominal clause that contains an activity verb marked for
past tense (the -ta form) has a ‘forward shifted’ reading in
which the event denoted by the adnominal clause
temporally precedes the matrix clause event. Sentences
with adnominal clauses pose some problems in processing
the semantic/syntactic relations linking their components.
In (1), when Kobayasi-ga (Kobayasi-NOM) is interpreted
as the subject of sikat-ta ‘scold-PAST,’ the sentence
yields a simple sentential reading meaning “Kobayasi
scolded the employee.”
(1) Kobayasi-ga syain-wo sikat-ta Yasuda-wo
Kobayasi-NOM employee-ACC scold-PAST Yasuda-ACC
yobituke-ta.
call-PAST
‘Kobayasi called Yasuda who had scolded the employee.’
But this interpretation crashes when another NP
Yasudawo (Yasuda-ACC) occurs after the V sikat-ta
(scoldPAST), signaling that the VP preceding Yasuda-wo
(Yasuda-ACC) must be construed as part of the
adnominal clause modifying Yasuda and that the first NP
Kobayasi-ga (Kobayasi-NOM) must be construed as the
subject of the sentence-final V yobituke-ta ‘call-PAST.’
This reinterpretation process, which requires some time, is
known as the garden-path (GP) effect. However, the GP
effect in adnominal clause constructions, such as in (1), is
slightly reduced when a bare NP such as syain ‘employee’
is replaced with a proper noun such as Imai, as in (2),
below
‘Kobayasi called Yasuda who had scolded Hirata.’
A similar influence of NP type on the magnitude of the GP effect is also observed in scrambled sentences. The sentences in (3c, d) contain an object–subject–verb word order that is assumed to be derived from the subject– object–verb word order contained in the sentences (3a, b) (Saito, 1985) 1. In a filler-driven parsing account of the processing of scrambled sentences, it is assumed that the object (e.g., Imai-wo nagut-ta Yasuda-wo ‘Yasuda who had hit Imai’) is reactivated at the trace position, so that scrambled sentences such as (3c, d) result in a diminished GP effect relative to sentences such as (3a, b), which exhibit canonical word order (Inoue, 2007; 2012)2.
However, there are some counterexamples to the general effect of NP type. The GP effect with adnominal clauses observed in the above examples is reduced only when the past -ta form is replaced with the present -ru form, as shown below: (1)’Kobayasi-ga syain-wo sikar-u
Kobayasi-NOM employee-ACC scold-PRESENT Yasuda-wo yobituke-ta.
Yasuda-ACC call-PAST
‘Kobayasi called Yasuda who scolds/was scolding Imai.’
1 This movement of the object is referred to as scrambling.
2 The judgment of the data is examined by using self-paced
moving-window reading paradigm
As mentioned above, garden-path sentences include a structural ambiguity between a simple sentential reading and a subordinate clause reading; it is assumed that this structural ambiguity is the cause of the GP effect. For example, in sentence (1), Kobayasi-ga (Kobayasi-NOM) potentially serves as the subject of both sikat-ta ‘scoldPAST’ and yobituke-ta ‘call-PAST,’ although it eventually becomes evident that Yasuda is the correct subject of sikat-ta ‘scold-PAST.’ If we faithfully follow the analysis that the source of the GP effect lies in the structural ambiguity of the sentence, we would expect all of the sentences in (1’) (2’) and (3’a, b) to result in a GP effect. However, (1’) is easily understood as “Kobayasi called Yasuda who scolds/was scolding the employee,” with little or no GP effect. The same result is observed in (2’) and (3’a, b). The question is why the GP effect is mitigated in these cases.
Before tackling this issue, it must be noted that data on sentences containing verbs that denote states also show reduced GP effects. Sentences (4a) and (4b), which have the same meaning, indicate that the contrast between the present -ru form and the past -ta form is abstracted away in the subordinate clause.
(4) a. Kare-ga i-ru koro-wa, yokat-ta.
He-NOM exist-PRES time-TOP good-PAST ‘The days where he was are the good days.’ b. Kare-ga i-ta koro-wa, yokat-ta.
He-NOM exist-PAST time-TOP good-PAST ‘The days where he was are the good days.’ Interestingly, the same is true in the GP sentences. When the activity verbs sikat-ta ‘scold-PAST’ in (1) and (2) and nagut-ta ‘hit-PAST’ in (3a, b) are replaced with a stative verb kirat-te-i-ru ‘hate’, the GP effect is not observed, regardless of NP type, as shown below.
(5)a. Kobayasi-ga syain-wo kirat-te-i-ru
Kobayasi-NOM employee-ACC hate-STATE Yasuda-wo yobituke-ta.
Yasuda-ACC call-PAST ‘Kobayasi called Yasuda who hates the employee.’ b.Kobayasi-ga Hirata-wo kirat-te-i-ru Yasuda-wo yobituke-ta.
‘Kobayasi called Yasuda who hates Hirata.’ (6)a. Kobayasi-ga Imai-wo kirat-te-i-ru Yasuda-wo home-ta.
‘Kobayasi praised Yasuda who hated Imai.’ b.C. Ronaldo-ga Imai-wo kirat-te-i-ru Yasuda-wo home-ta.
‘C. Ronaldo praised Yasuda who hated Imai.’
This result is significant because it suggests that the
present -ru form, when occurring with activity verbs, has
a stronger impact on the magnitude of the GP effect than
the syntactic structure of the sentence or the type of head
NP in the adnominal clause has. The fact that adnominal
clauses involving an activity verb in the present -ru form
exhibit a reduced GP effect cannot be accounted for by
previous analyses, in which the GP effect was attributed
to the structural ambiguity or NP type, as claimed, for
example, by
The present -ru form (in episodic sentences) is temporally neutral in contrast with the past -ta form. It could take on any reference, with the context determining its semantic value. (8) Hypothesis 2
The present -ru form (in episodic sentences) is a perspective shifter. It indicates the movement of point of view (for instance, from speaker to listener) in a given sentence. (9) Hypothesis 3
The present -ru form (in episodic sentences) denotes a ‘propositional concept’ whose truth-value is not determined. Thus, GP sentence (1) is interpretable either way, regardless of whether Kobayasi or Yasuda is construed as subject of the V nagu-ru. Hypothesis 1 appears to be the most plausible option, consistent with the observation these verbs in the -ru form (in episodic sentences) have various temporal readings, as illustrated in (10): (10) a. Ashita, Tokyo-ni ik-u. (Future)
Tomorrow, Tokyo-to go-FUTURE ‘Tomorrow, I will go to Tokyo.’ b. Ima, dekake-ru-tokoro-da. (Near future/present)
Now,go-out-PRESENT- place-COPULA-PRESENT ‘I am going to go out.’ c. Hati-wa mitu-wo atume-ru. (Generic sentence) Bee-TOP honey-ACC collect-PRESENT ‘A bee collects honey.’ d. Ken-wa kinoo takusan tabe-ru kara. (Past) Ken-TOP yesterday lot eat-PRESENT since, onaka-ga itai-nda stomach-NOM have a pain-PRESENT ‘Since Ken ate a lot yesterday, he has a stomachache.’ Hypothesis 2 is also reasonable. It is typical for the utterer to express her/his own point of view anchored in terms of “here,” indicating the place where she/he exists, and “now,” indicating the utterance time. The historic present is a rhetorical device for using the present tense to describe a past event. According to hypothesis 2, the processing of GP sentences such as (1’) could be described as follows: the V sikar-u ‘scold-PRESENT’ induces a shift in point of view from the sentential subject Kobayasi to the speaker or author of (1’). In other words, the adnominal clause in (1’) represents the event that is observed directly by the speaker or the author. As a result, the VP syain-wo sika-ru ‘scold-PRESENT the employee’ preceding Yasuda-wo (Yasuda-ACC) could be easily construed as part of the adnominal clause modifying Yasuda.
Hypothesis 3 appears to be the least preferred option because it runs contrary to the observation that in sentences (10a, b, d), the verbs in the present -ru form cooccur with the definite temporal adverbial ashita ‘tomorrow’ and kinoo ‘yesterday,’ and the propositions denoted by these sentences seem to be evaluated as true.
However, I will argue in this paper that Hypothesis 3 is ultimately the most plausible choice among the three, as the other two can be contradicted by empirical evidence.
This section examines Hypothesis 1, which is restated as (11) below: (11) a. The present -ru form (in episodic sentences) is temporally neutral in contrast to the past -ta form. b. It is could take on any reference, with the context determining its semantic value.
Data such as (10d) seem to support this line of analysis3. In this example, the literal meaning of the expression of the V tabe-ru ‘eat-PRESENT’ in the subordinate clause describes an event in the present, but the pragmatic meaning conveys a past event: (10d) can be paraphrased as (12), in which the past -ta form is used in the subordinate cause.
(12) Ken-wa kinoo takusan tabe-ta
Ken-TOP yesterday lot eat-PAST onaka-ga itai-nda stomach-NOM have a pain-PRESENT ‘Since Ken ate a lot yesterday, he has a stomachache.’ kara, (Past) since,
However, it must be noted that the sentence (10d) cannot be embedded in the negated complement of the factive verb sira-nai ‘do not know-PRESENT,’ as illustrated below: (13) a.* Ken-wa kinoo takusan tabe-ru kara, Ken-TOP yesterday lot eat-PRESENT since, onaka-ga itai koto-wo stomach-NOM have a pain-PRESENT COMP-ACC watasi-wa sira-nai.
I-TOP know-NEG-PRESENT
3 As for (10d), see
I-TOP know- PRESENT ‘I know that since Ken eats a lot yesterday, he has a stomachache.’ (14) a. Ken-wa kinoo takusan tabe-ta kara, Ken-TOP yesterday lot eat-PAST since, onaka-ga ita-i koto-wo stomach-NOM have a pain-PRESENT COMP-ACC watasi-wa sira-nai.
I-TOP know-NEG-PRESENT ‘I do not know that since Ken ate a lot yesterday, he has a stomachache.’ b. Ken-wa kinoo takusan tabe-ta kara, Ken-TOP yesterday lot eat-PAST since, onaka-ga ita-i koto-wo stomach-NOM have a pain-PRESENT COMP-ACC watasi-wa sit-te-i-ru.
I-TOP know-PRESENT
‘I know that since Ken ate a lot yesterday, he has a
stomachache.’
(15) a. Actually, I saw Ken eat yesterday, and I think that
Ken ate a lot, and this causes a stomachache. b. Actually, I did not see Ken eat yesterday. But it is objectively true that Ken ate a lot, and this causes a stomachache. (15a, b) are the predicted interpretations of (10d) and (12) respectively.
The fact that the sentence in (10d), where the -ru form is selected, is only available in the complement of sit-te-iru ‘know-PRESENT’ but is not available in the complement of sira-nai ‘know-NEG-PRESENT’ demonstrates that the subordinate clause in (10d) is truth-conditionally ambiguous/non-decisive, allowing each of the following possibilities regarding Ken’s eating activity (16): (16) a. Ken ate a lot.
b. It is not the case that Ken ate a lot.
c. Ken ate a little.
In contrast, it is quite obvious that the subordinate clause in (12), where the -ta form is selected, is truthconditionally decisive, allowing only the reading in (16a). This is confirmed by the fact that the sentence in (12) can be embedded in both the complement of sira-nai ‘knowNEG-PRESENT’ and sit-te-iru ‘know-PRESENT.’ Therefore, the available readings for (15a) (= (10d) and (15b) (= (12)) seem to be determined by the -ru form and the -ta form, respectively. These observations hint at the presence of a hidden cognitive subject other than the sentential subject Ken. However, Hypothesis 1 is not able to account for these nuances involving (16a-c).
The reader may argue that these nuances come from the context rather than the content of the sentence, thereby allowing Hypothesis 1 to be maintained. However, this line of reasoning cannot be supported because the reading of the -ta form in the subordinate clause must also be explained in terms of Hypothesis 1 (cf. (11a, b)), as shown in (17a, b) and (18a,b), where the -ta forms in the subordinate clause denote the time qua the past in (17a) and qua the future in (17b), and -ru forms denote the time qua the past in (18a) and qua the future in (18b). (17) a. Ken-wa furansu-ni it-ta toki,
Ken-TOP France-to go-PAST time, kaban-wo kat-ta. bag-ACC buy- PAST ‘Ken bought a bag when had been in France.’ b. Ken-wa furansu-ni it-ta toki,
Ken-TOP France-to go-PAST time, kaban-wo kaw-u. bag-ACC buy- PRESENT ‘Ken will buy a bag when he will be in France.’ (18) a. Ken-wa furansu-ni ik-u toki,
Ken-TOP France-to go-PRES time, kaban-wo kat-ta. bag-ACC buy-PAST ‘Ken bought a bag when he went to France.’ b. Ken-wa furansu-ni ik-u toki,
Ken-TOP France-to go-PRES time, kaban-wo kaw-u. bag-ACCbuy-FUTURE ‘Ken will buy a bag when he will go to France.’ These readings can be captured by means of the following principle4.
4 More precisely, the principle of (19a, b) reflects the SOT rule below.
The SOT (sequence of tense) rule:
If a tense feature B is the local tense feature of a tense
feature A at LF and A and B are occurrences of the same
feature (i.e., either [+past] or [+pres]), then A and the
tense associated with A (if any) are optionally deleted.
N.B.: (i) The tense features include [+past] and [+pres]
and nothing else. (ii) A tense feature A is “in the scope”
of a tense feature B iff B is associated with a common
noun and asymmetrically c-commands A or B is
associated with a tense or a perfect and asymmetrically
commands A. (iii) A tense feature B is the local tense
feature of a tense feature A iff A is “in the scope” of B
and there is no tense feature C “in the scope” of B such
that A is “in the scope” of C.
When the -ru form is replaced with the -ta form in the subordinate clause, a parallel temporal interpretation is obtained, unless both the matrix clauses are not same. Therefore, in conclusion, Hypothesis 1 is somewhat plausible, but given the parallel readings of the -ru and -ta forms in subordinate clauses, Hypothesis 1 is simply a different way of stating the principle in (19a, b), and it cannot adequately address the issue of why a reduction in the GP effect is obtained only for activity verbs in the present -ru form.
Next, we pursued the possibility that the present -ru form (in episodic sentences) is a perspective shifter. Under this account, the present -ru form indicates the shift in point of view in (1’) and (3’a, b). Given that there is no morphologically overt operator that indicates a perspective shift in these sentences, the -ru form is a good candidate for fulfilling this function. Hypothesis 2 is repeated in (20) below: (20)a. The present -ru form (in episodic sentences) is a perspective shifter. b. The present -ru form includes, as part of its meaning, the movement of the point of view within the sentence.
As mentioned in the previous section, the predicted interpretation of (10d), which includes the -ru form in the forward shifted reading, is repeated below in (15a): (15) a. Actually, I saw Ken eat yesterday, and I think that
Ken ate a lot, and this causes a stomachache.
Presumably, the sentences in (1’) and (10d) demonstrate that the subordinate clause is truth conditionally ambiguous/non-decisive. If this presumption is correct, the meaning of (10d), according to Hypothesis 2, is (15a), which includes a perspective shift from the sentential subject Ken to the speaker.
However, this analysis can be challenged in several ways. First, (1’) (repeated below) is the same as (10d) except that no acquaintance relationship is apparent, which would put a speaker in cognitive contact with the situation, “Kobayasi scolds the employee” or “Yasuda scolds the employee.” (1)’Kobayasi-ga syain-wo sikar-u Yasuda-wo yobituke-ta.
(unambiguous) This fact is not predicted under Hypothesis 2. When the present -ru form appears in the matrix clause, the acquaintance relationship disappears, but the present -ru form is still expected to induce the perspective shift from Kobayasi to Yasuda. However, the sentence in (1’) is highly distinguishable and no perspective shift is observed.
Second, suppose that the present -ru form triggers a shift in point of view as a last resort only when there is no other overt morphological element that can indicate a shift in point of view within the sentence. In that case, when the present -ru form is included in the sentences in (10a– c) (repeated below), we would expect them to exhibit a perspective shift.
(10)a. Ashita, Tokyo-ni ik-u. (Future) ‘Tomorrow, I will go to Tokyo.’ b.Ima, dekake-ru tokoro-da. (Near future/present) ‘I am going to go out.’ c. Hati-wa mitu-wo atume-ru. (Generic statement) ‘A bee collects honey.’ d.Ken-wa kinoo takusan tabe-ru kara onaka-ga itai-nda. (Past) ‘Since Ken ate a lot yesterday, he has a stomachache.’ However, this prediction is not borne out: the -ru forms in (10a–c) can be understood as “future,” “near future or present,” and “generic statement,” respectively. In these sentences, we do not observe the perspective shift. Rather, these sentences are bound together by a common characteristic: they are truth conditionally indecisive. Therefore, it seems reasonable to argue that the -ru form can be semantically characterized as a function from propositions to sets of possible worlds.
The general observation is as follows: A reduced GP effect in adnominal clauses depends upon the present -ru form. This close relationship between the GP effect and the -ru form suggests that the present -ru form itself serves as a function from propositions to sets of possible worlds.
The discussion above leads us to Hypothesis 3 below, at first glance, the least favored hypothesis: (21) a. The present -ru form (in episodic sentences) denotes a ‘propositional concept’. b. The present -ru form serves as a function from propositions to sets of possible worlds. Thus, a proposition that contains the -ru form is interpreted as being true in some presupposed possible world but not in all possible worlds that are accessible in a given context.
This may seem to be a peculiar hypothesis, but it accounts for the facts. First, the meaning of (1’) under Hypothesis 3 is as shown in (22). This is the same as seen in (1), which exhibits a GP effect, except that the GP effect is not observed in the reinterpretation process: it is easy to cancel the reading in (a) and shift to reading (b) and then (c). (22) a. “Kobayasi scolded the employee.” b. Reading (a) is accommodated as “Yasuda scolded the employee.” c. Reading (b) is accommodated as “Kobayasi called
Yasuda who had scolded the employee.”
In (10d), we saw that a subordinate clause involving the
-ru form expresses an ‘acquaintance relation’ in
Suppose that the speaker presupposes that Ken ate a lot,
while Ken himself thinks that it is not the case that he ate
a lot. Following
In this vein, the sentence in (1’) is also accounted for, since without the -ru form, there is no element that represents a different state of possible worlds against which to evaluate the given proposition. For convenience, let us suppose only two worlds, such that i is the world in which Kobayasi is taken to be the subject of the V sikar-u ‘scold-PRESENT’ and j is the world in which Yasuda is taken to be the subject; we can represent the adnominal proposition in (1’) using the two-dimensional matrix below: Matrix (24) represents the propositional concept corresponding to the adnominal clause in (1’). What the two-dimensional matrix conveys is roughly this: the proposition “Kobayasi scolds the employee” is true in i but not j. At the same time, the proposition “Yasuda scolds the employee” is true in j but not i. In a sense, the adnominal clause in the sentence (1’) extends worlds, that is, it extends perspectives. In this sense, the sentence (1’) represents a ‘diagonal proposition’. Therefore, the sentence in (1’) is interpretable either way, regardless of whether Kobayasi or Yasuda is construed as the subject of the V sikar-u. This explains why the -ru form can play a principal role in reducing the GP effect.
In contrast, it is obvious that the adnominal clause in (1) containing the ta-form is truth-conditionally decisive; in this case, the GP effect is observed. Let us suppose two worlds such that i is the world in which Kobayasi is taken to be the subject of the V sikat-ta ‘scold-PAST’ and j is the world in which Yasuda is taken to be the subject; we can represent the adnominal proposition in (1) using the one-dimensional matrix below: (25) Kobayasi-ga syain-wo sikat-ta Yasuda-wo yobituke-ta. (Kobayasi called Yasuda who scolded the employee.) i T j F i F j T Matrix (25) does not represent a diagonal proposition. The one-dimensional matrix conveys roughly the following: the proposition “Kobayasi scolded the employee” is true in i, and there is no world other than i. Thus, it is not possible for the proposition “Yasuda scolds the employee” to be true in j at the same time. The past -ta form blocks the extension of the worlds under which the proposition is evaluated and forces a one-dimensional perspective. This explains why the -ta form induces a GP effect.
The analysis provided in this paper can be summarized in (26) to (28): (26) Subordinate clauses containing active verbs in the present -ru form as the main predicate result in a reduction of the GP effect. (27) The present -ru form serves as a function from propositions to sets of possible worlds. Thus, a proposition containing the -ru form is interpreted as true in some presupposed possible world but not in all possible worlds accessible in a given context. (28) For the reason expressed in (27), the -ru form extends the worlds against which a given proposition is evaluated, yielding a ‘diagonal proposition’, which has the effect of reducing the GP effect.
Scientific Research (C) (25370447).