Abstract
This paper examines the co-occurrence tendency of motion ideophones with different kinds of motion verbs (Ibarretxe-Antuñano, 2005; Toratani, 2012) as manifested in Japanese. A total of 2429 co-occurrences of motion verbs and ideophones extracted from the 100-million-word corpus of written Japanese (bccwj) were analyzed. A comparison of the results with the overall frequencies of different verbs in the same corpus reveals that motion ideophones tend to collocate better with manner verbs, occurring with a disproportionally large number of manner verbs in spite of their relative infrequency in this language. An item-by-item analysis further shows that individual ideophones tend to occur with manner or path verbs that contain meanings related to ideophones, analyzed in the revised framework of semantic grids of ideophones (cf. Ibarretxe-Antuñano, 2019). Strikingly, motion ideophones tend not to co-occur with simple deictic verbs, and this finding is discussed in relation to the pragmatics of informativeness.
1 Issue
So far, this sound-symbolic Manner-specifying constituent type seems to occur extensively only in verb-framed languages. In such languages, further, this constituent type seems not to be the sole (or perhaps even principal) locus for representing manner, but rather to be additional (and perhaps ancillary) to that of another constituent type: The more familiar type consisting of the gerundive or other non-finite form of a verb root. Accordingly, in a sentence containing both of these constituent types, Manner is represented distributedly.
To support this claim, Talmy refers to the patterns found in Spanish and Japanese. One Japanese sentence representing such a pattern is given in (1). In this sentence, the manner ideophone bura-bura(-to) occurs with the V-V compound verb aruki-mawar(-u), in which the manner verb aruki (the renyookei form of aruk(-u)) modifies the path verb mawar(-u), which serves as the head of the compound.
Similar observations have been made by other researchers. A strong collocation between motion ideophones/mimetics and manner verbs has been reported in different languages (Chinese: Kimura, 1976: 47–48; Japanese: Hirose, 1981; Emai: Schaefer, 2001, etc.). However, no comprehensive quantitative study has yet confirmed this phenomenon. Such a study should examine the frequencies of different motion verb types (e.g., manner verbs, path verbs) occurring with ideophones and compare these with the overall frequencies of these verb types (used with or without ideophones).2
The present paper conducts a corpus-based analysis to address this gap, examining the frequencies of motion ideophones and motion verbs in Japanese. The Balanced Corpus of Contemporary Written Japanese (bccwj) is used, which is a 100-million-word corpus of written materials published between 1976 and 2005. I will show that the above claim concerning the collocation of motion ideophones and motion verbs is statistically valid as a general tendency, though item-by-item analyses reveal that the co-occurrence patterns depend on the particular meanings of individual motion ideophones. These findings are discussed in relation to the pragmatics of informativeness: Ideophones are used when a high depictivity of manner or path is in demand, which affects the choice of other elements in the same sentence.
2 Japanese Motion Event Descriptions and Motion Ideophones
This section reviews previous findings on motion event descriptions and motion ideophones in Japanese.
2.1 Motion Event Descriptions in Japanese
Regarding the typology of motion event descriptions, Japanese has been classified as a language in which path is coded in the main verb (referred to as a “verb-framed language” in Talmy [1991, 2000] and “head path-coding language” in Matsumoto [2017, 2025]). However, when deixis is present, it occupies the main verb slot, with a path verb (if any) modifying the deictic verb (Matsumoto, 2020; Koga, 2025). Manner is typically indicated by an element other than the main verb, and basic manners like ‘walking’ and ‘running’ are often expressed using “converbal” verbs marked by -te, which modify the main verb, as in (2).
2.2 Japanese Motion Verbs
A classification of Japanese motion verbs is proposed in Matsumoto (1997, 2020), dividing mono-morphemic verbs into three categories: manner verbs (M), path verbs (P), and deictic verbs (D), as shown in (3).3
Deictic verbs do contain a path component (e.g., ku(-ru) ‘come’ means ‘move (in)to the speaker’s domain’) (Matsumoto et al., 2017), crucially involving to (or to.in). However, distributional evidence suggests a separation between path and deictic verbs in Japanese (as we will see shortly), which is relevant to the focus of this paper.
Japanese has a relatively rich set of path verbs, while its manner verb inventory is limited. Manner verbs in this language are all general in meaning. Unlike English (Snell-Hornby, 1983; Levin, 1993; and Slobin, 1996), Japanese lacks highly specific mono-morphemic manner verbs (e.g., ‘stride’, ‘swagger’), especially those which describe: 1) attitudes of the subject entity (e.g., ‘swagger’, ‘strut’), 2) special motor patterns (e.g., ‘shuffle’, ‘scurry’), and 3) sounds accompanying motion (e.g., ‘rattle’) (cf. the semantics of ideophones in Section 3.1).4 In this sense, semantic detailedness is restricted in manner verbs in Japanese.
Some of these manner and path verbs encode multiple components of meaning. Some path verbs encode an additional semantic component. For example, the verb nobor(-u) ‘climb up’ includes the agentivity component in addition to path (the upward direction of motion); it would sound unnatural for this verb to have a non-agentive entity such as an elevator or fireworks as its subject.5 Manner verbs can also have an additional component. For example, the manner verbs tob(-u) ‘fly, jump’ and maw(-u) ‘flutter, dance in the air’ in a sense contain a path specification, as the motion must be in the air. Such cases potentially challenge the categorization of motion verbs into different types. However, because of distributional restrictions that we will see shortly, it is possible to categorize such verbs as either primarily a manner verb or a path verb.
One characteristic of the Japanese motion verb inventory is the existence of verbal complexes (Matsumoto, 1996, 1997). There are two types of verbal complexes, i.e., V-V compounds and V-te V complex predicates, and each has distinct restrictions on verb combinations (Matsumoto, 1997, 2020, 2021).
V-V-compounds are exemplified in (4), and their combination types are given in (5) (“-” is used to mark concatenation in compounding). Most are composed of a manner verb as V1 and a path verb as V2 (M-P). In other cases, compounds combine two path verbs (P-P), or an action verb (A) as a V1 and a manner or path verb as V2 (A-M, A-P). Verbs participating in the last three types are limited (see Matsumoto, 1997).
These patterns in compounding indicate a distinction of verb types. Notably, deictic verbs do not participate in V-V compounds, suggesting a clear distinction between deictic verbs and other verbs (Matsumoto, 2021).7 Manner verbs and path verbs can also be distinguished. Although the verb maw(-u) ‘flutter, dance in the air’ includes a path component in addition to a manner component, it behaves as a manner verb, as it consistently occurs in the M position in compounds (e.g., mai-agar(-u) (flutter-ascend) ‘soar up’ (M-P)), and never in the P position (e.g., *toori-maw(-u) (pass-flutter) (P-P)).
V-te V complex predicates have a deictic verb as the final verb (Matsumoto, 1996, 2020), as exemplified in (6). The semantic combinations of verbs in the V-te V complexes are given in (7), with examples (“+” is used to mark concatenation with -te marking). The first verb of a V-te V complex can be a compound verb, forming an M-P+D sequence, for example.
In this paper, the term simplex verb is used to refer to standalone motion verbs, not in the form of compounds or complex predicates. In addition, the term motion verb is used as a cover term for simplex verbs, compound verbs, and complex predicates representing motion.
The frequency of each motion verb is needed to compare the general frequencies of various kinds of motion verbs with those of the various kinds of motion verbs occurring with motion ideophones. Such frequency data are available in some studies. Matsumoto (2017) analyzed a small sample from the pre-publication version of the bccwj, and identified 882 tokens of motion verb uses (including verbal complexes). These tokens exclude the non-motion uses of motion verbs (e.g., metaphorically extended uses of motion verbs). He classifies them into different types, as shown in Table 1. This kind of table is what is needed for the present purpose. However, a larger data set is desirable.
Other studies provide frequency data from the entire bccwj. However, existing databases fail to provide the data needed for the present purpose. Tono, Yamazaki, and Maekawa (2013) provide the frequencies of the 5000 most common Japanese lexical items in the entire bccwj, but this does not contain information on all motion verbs. A lexical profiler for the bccwj called “ninjal-lwp for bccwj” gives the frequencies of all individual motion verbs and some compound verbs listed in the dictionary used by the ninjal-lwp tool. However, 1) the frequencies that can be obtained are the totals of the motion and non-motion uses of motion verbs, and 2) the frequencies of some compound verbs and almost all of the V-te V complex predicates are not available (i.e., these instances are included in the counts of constituent verbs). There is now a version of the bccwj called “bccwj-wlsp” in which each word is tagged with the semantic domain of the particular sense of the word, using the semantic category numbers of the Bunrui-Goihyo (Word List by Semantic Principles). However, a close look at a small sample of bccwj-wlsp has shown that examples identified as motion uses in fact include metaphorical motion (e.g., degree change). Thus, existing databases will not suffice for our purpose if we want to have a larger sample size of frequency data than Matsumoto (2017).
2.3 Japanese Motion Ideophones
Japanese is rich in ideophones (Kakehi et al., 1996; Akita, 2009), including a large number of motion ideophones (Toratani, 2012; Akita, 2020). Most of them are adverbs. Japanese only has basic manner-of-motion verbs, and finer manner meanings are expressed through motion ideophones (Matsumoto, 1997; Akita & Matsumoto, 2020). Most of the motion ideophones, such as bura-bura(-to) ‘walking leisurely’ in (1) and suta-suta(-to) ‘walking briskly’, encode manner information, while others, such as guru-guru(-to) ‘moving in a big circle’, encode path information. Notably, many such ideophones do not encode only a single notion. Inspired by Ibarretxe-Antuñano’s (2006) semantic grid analysis of Basque ideophones, Toratani (2012) and Akita (2020) analyzed Japanese motion ideophones in terms of multiple types of information, and showed that many motion ideophones encode different types of information simultaneously. The ideophone koro-koro(-to) ‘rolling repeatedly’ is a case in point: this ideophone encodes information on manner (rolling), figure (small spherical object), and aspect (repeatedness).8 Ibarretxe-Antuñano (2019) also analyzed 13 Japanese motion ideophones in her semantic grid approach, showing that these ideophones encode several notions within a single form. One may note, however, that she claims that path is encoded in all the Japanese ideophones she examined. This claim about path is mistaken, as her own data in the paper’s appendix do not support this claim.
In these previous studies (Toratani, 2012; Akita, 2020; and Ibarretxe-Antuñano, 2019), slightly different lists of semantic components have been used. Akita’s list of semantic components is shown in (8).
Such lists of semantic information require refinement. Toratani (2012), for example, groups direction and aspect together with motor pattern, while Akita (2020) omits direction and aspect altogether. None of these three previous studies address the factor of evaluation, which is acknowledged as an important part of the meanings of ideophones (Dingemanse & Majid, 2012; Van Hoey, 2024, etc.) or of manner semantics in general (Stosic, 2019).9
2.4 Co-occurrence of Motion Ideophones and Motion Verbs
Two important previous studies have addressed the issue of the co-occurrence of Japanese ideophones and verbs. Akita (2012) found a strong tendency for certain pairs of ideophones and verbs/nominals to collocate with each other. He argued that such pairs tend to evoke common “frames” (Fillmore, 1982; Fillmore & Baker, 2015). He showed that these collocations are stronger than those of non-ideophone adverbs and verbs. This finding aligns with the hypothesis stated in Section 1 above, in that in both cases ideophones tend to occur with verbs that are semantically related to the ideophones. However, Akita’s analysis employs very broad frame categories, treating all motion verbs and ideophones under “the motion frame,” without distinguishing between collocations involving different kinds of motion verbs (i.e., manner verbs, path verbs, etc.).
Perhaps the most relevant previous study is Toratani’s (2012). She examined the co-occurrence of Japanese motion ideophones and motion verbs in 12 literary works. She made two key observations. First, she observed that manner ideophones very often occur with path verbs, providing the sole “co-event” (manner) information, an observation that appears to challenge the hypothesis mentioned earlier. She has also observed that specific motion ideophones tend to occur with the verb of their “superordinate” category (e.g., ideophones describing different walking manners occur with aruk(-u) ‘walk’). Toratani’s first observation requires careful evaluation, since such co-occurrence patterns must be evaluated against the overall frequencies of manner, path, and deictic verbs in Japanese.
3 The Present Study
This paper investigates the use of motion ideophones and motion verbs in the bccwj. The lexical profiler tool called ninjal-lwp (ver. 1.30) was used to search for motion ideophones in all bccwj subcorpora except the newspaper subcorpus due to tool limitation. ninjal-lwp gives a profile of lexical items (headwords), providing their collocation patterns and frequencies. ninjal-lwp uses the ipa dictionary for morphemic analysis. This dictionary is somewhat incomplete in its listing of ideophones, and a specialized list of ideophones has been compiled by Kimi Akita, which can be used in the searches within ninjal-lwp. The present paper used this special list of ideophones to obtain the data for motion ideophones and co-occurring motion verbs. The same corpus was used to determine the overall frequencies of motion verbs (used with or without ideophones).
3.1 Selection and Analysis of Motion Ideophones
The motion ideophones to be examined in this paper were selected in the following way. The 800 most frequent ideophones were examined to see how often they occur with motion verbs, using the co-occurrence tables provided by ninjal-lwp for each ideophone. Ideophones with general meanings not specifically used for motion (e.g., nonbiri ‘slowly and leisurely’) were excluded, and only those that are predominantly used for motion were identified. In this way the top 33 motion ideophones were identified. They are listed in Table 2, which includes 19 items also listed in Akita (2020). Their frequencies of occurrence with motion verbs are also given in the table. These frequencies do not include the instances of ideophones occurring with non-motion uses of motion verbs (instances representing self-contained motion (e.g., rotating) are regarded as non-motion uses).
Semantic analysis of these 33 motion ideophones was conducted, using introspection. They are examined in light of a list of semantic components modified from Akita (2020) and Ibarretxe-Antuñano (2019). The results are presented in Table 3. In this table, the ideophones are grouped into eight categories, but these should not be taken to be rigid categories because of the presence of overlapping features.
Many ideophones encode the attitude of a moving person. The criterion used to identify attitudinal meanings is compatibility with the nominal taido ‘attitude (of doing …)’. For instance, a verb phrase with a motion verb and the ideophone noko-noko(-to) ‘going somewhere shamelessly’ sounds natural modifying this nominal (i.e., noko-noko-to yatte kuru toyuu taido ‘the attitude of coming over shamelessly’), but the acceptability is reduced in the case of suta-suta(-to) ‘walking briskly’ (i.e.,??suta-suta-to yatte kuru toyuu taido ‘the attitude of coming over briskly’).10 The necessary involvement of alternating leg movement can be identified by compatibility with such phrases as jitensha-de ‘by bicycle’. For instance, the ideophone suta-suta(-to) is incompatible with this phrase, but noko-noko(-to) is compatible with it, suggesting that alternating leg movement is necessary in the former but not so in the latter. Path-related meanings can be identified by the occurrence with specific path expressions. The ideophone noko-noko(-to) sounds somewhat strange with aruki-mawar(-u) ‘walk around’, but sounds natural with haitte kuru ‘come entering’, indicating a goal-oriented nuance. In contrast, suta-suta(-to) is compatible with all of these.
Table 3 suggests that many ideophones (e.g., suta-suta(-to) ‘walking briskly’) are primarily manner-based, while some others are path-based (e.g., guru-guru(-to) ‘moving around in a big circle’). Notably, many have specifications of more than one category (figure, manner, and path), forming a gestalt of features. The most common feature is Figure_category, followed by General_aspectuality, Manner_rate, General_evaluation, Figure_attitude, and Manner_motor pattern. Path information is encoded in many but not all ideophones (unlike what Ibarretxe-Antuñano (2019) claims).11
3.2 Overall Frequencies of Motion Verbs
This paper compares the frequencies of motion verbs occurring with motion ideophones with the overall frequencies of motion verbs in the same corpus. A new count of each motion verb in bccwj (excluding the newspaper subcorpus) was conducted, with the assistance of Yuqi Yin and Yuko Morokuma. Forty-six simplex motion verbs including those listed in Matsumoto (1997) as manner, path, or deictic verbs were searched to identify their motion uses (see Appendix 1 for the list of simplex verbs examined). Since almost all motion verbs are polysemous, the counts must exclude those examples of these verbs used in non-motion senses in order to determine the frequencies of motion verb uses. This was done by examining the co-occurring case-marked nominal arguments/adjuncts of these verbs. For example, the degree-change meaning (‘increase’) of agar(-u) ‘go up’ (e.g., ondo-ga agat-ta ‘The temperature rose’) can be identified by the nominal of the nominative-marked subject; the non-motion examples identified this way were removed from the counts.12 Sentences mistakenly listed as examples of motion verbs were also excluded.
The frequencies of compound verbs were calculated in the following way. Some compound verbs have their own entries as headwords in ninjal-lwp, and the motion uses of those compound verbs were identified in the same way as those of simplex motion verbs. Other compound verbs lacking their own entries were identified by examining the instances of their constituent verbs. The instances of those compound verbs were excluded from the counts of the constituent verbs, in order to avoid the double counting of these instances as instances of compound verbs and those of their member verbs. Frequencies of 93 compound verbs were examined in this way (see Appendix 2 for the list of compound verbs examined). No V-te V complexes have their own entries in ninjal-lwp, so the frequencies of these predicates were determined by examining the combinations of constituent verbs. These complex predicates were excluded from the counts of constituent verbs.
4 The Co-occurrence of Motion Ideophones and Motion Verbs
4.1 General Tendency
The motion verbs occuring with motion ideophones were classified into the different types listed in (3), (5), and (7). The results are presented in Table 4. The table shows that motion ideophones occur most frequently with simplex path verbs than with any other kinds of verbs. This result seemingly contradicts the claim that motion ideophones tend to occur with manner verbs. However, as stated above, these counts must be evaluated against the general frequencies of different types of motion verbs.
Using the procedures described in Section 3.2, a total of 358,739 instances of motion verbs were identified in the bccwj. The frequencies of different motion verb types were calculated, and the results are shown in Table 5.
A comparison between Tables 4 and 5 reveals the following. Table 5 shows that simplex path verbs (P) are the most frequent category, and they naturally account for a large part of those occurring with ideophones in Table 4. Simplex manner verbs (M), on the other hand, account for only 11.1% of all motion verbs, but they account for 25.2% of those paired with ideophones, indicating a disproportionally strong collocation. Deictic verbs in their simplex use (D) account for 24.9% of all motion verbs, but only 3.3% of those occurring with ideophones, indicating a significant underrepresentation. A chi-square analysis of the frequencies of motion verbs and those of motion verbs occurring with ideophones was significant (χ2=1689.2, df=8, p <.001). Residual analysis (Table 6) further reveals that motion ideophones tend to occur with simplex manner verbs (M) and in their complex predicate use (M+D) significantly more often than expected (p <.001), while simplex deictic verbs (D) are used significantly less often with ideophones than expected (p <.001). In comparison to simplex D, complex predicates involving D are more frequent with ideophones than expected (M+D, P+D, M-P+D).
A further examination of each motion verb suggests that certain verbs exhibit a particularly strong tendency to collocate with ideophones. The two most frequent verbs (in their simplex use) occurring with motion ideophones were mawar(-u) ‘move around’ (368 instances; 24.2% of all co-occurring verbs in their simplex use) and aruk(-u) ‘walk’ (363 instances; 23.9%). These are striking since these verbs account for only 1.5% and 3.8% of all simplex motion verbs in the corpus, respectively.14
4.2 Variations among Ideophones
Motion ideophones differ in whether they tend to occur with manner verbs or path verbs. Those that tend to occur with verbs/verbal complexes containing manner or path are given in Tables 7 and 8, respectively. In these tables, all motion verbs/verbal complexes with a manner meaning (M, M-P, M+D, M-P+D) and those with a path meaning (P, M-P, P+D, M-P+D) were counted for “Manner verbs/verbal complexes” and “Path verbs/verbal complexes,” respectively (M-P and M-P+D were counted in both). The most frequent manner or path verb for each ideophone is also given.
The results show the following tendencies, given the semantic grouping of ideophones in Table 3. Ideophones that tend to occur with manner verbs/verbal complexes are listed in (9), together with the particular verbs they tend to occur with.
Ideophones that tend to occur with path verbs/verbal complexes are listed in (10):
These suggest that the congruence between verb meanings and ideophone meanings can be seen at a very specific semantic level. Motion ideophones tend to occur with verbs containing a similar/related meaning.
5 General Discussion
The findings above reveal that motion ideophones occur disproportionally more often with manner verbs. Japanese has a number of specific manner ideophones occurring with manner verbs, especially the verb of walking, confirming the pattern observed by Talmy and discussed in Ibarretxe-Antuñano (2005). On the other hand, certain motion ideophones tend to occur with path verbs, and they have meanings related to the path that the verbs express.
A notable tendency is the avoidance of motion ideophones occurring with simplex deictic verbs. This avoidance is not seen for complex predicates with deixis, suggesting that it is not the deictic meaning that is incompatible with ideophones, but rather the lack of manner or path information in the simplex use of deictic verbs.
So why do motion ideophones tend not to occur with verbs that lack manner or path information? An answer can be sought in the informativeness for expressing events. In selecting expressions, speakers are sensitive to the level of informativeness required (Grice, 1975; Matsumoto, 1995; Levinson, 2000; Davies & Arnold, 2019). Such informativeness is sensitive to the context. In the literature of conversational implicature, it is known that the use of a relatively less informative expression instead of a more informative alternative suggests that the speaker is not in a position to use the more informative option, especially when detailed information is expected in the context (Hirschberg, 1985; Matsumoto, 1995). For example, Matsumoto (1995) points out that using the general term dog instead of a specific breed name, such as English cocker spaniel, in a lost dog notice like (11) would be unusual. Such a use would suggest as one possibility that the dog is a mongrel. This becomes especially clear when similar notices posted on the same board include breed names.
Motion ideophones convey detailed and gestalt information on the nature of motion, especially concerning manner and/or path, giving a concrete image of the motion described. Some argue that speakers are employing a distinctive mode called the depictive mode when using ideophones (Akita, 2019). If speakers choose to be very depictive and informative concerning manner/path in using ideophones, they are also expected to be highly informative in choosing verbs occurring in the same clause. Using a simplex deictic verb in spite of the availability of a choice richer in manner/path information in this context would undermine this expectation of informativeness. Thus, the use of ik(-u) ‘go’ with an ideophone such as tobo-tobo(-to) ‘plodding’ would suggest that the speaker has some reason for not using a verb carrying information about manner such as aruk(-u) ‘walk’: most likely the manner of motion was not in fact walking. But this would contradict the meaning of tobo-tobo(-to), hence the avoidance of a simplex deictic verb.15
It is an interesting question to examine whether this avoidance of simplex deictic verbs is found in the use of non-ideophonic adverbs. The answer appears to be yes, as far as non-ideophonic adverbs carrying highly specific manner information are concerned. My examination of two very specific manner adverbs, oomata-de (big.stride-inst) ‘in big strides’ and shinobi-ashi-de (do.by.stealth-foot-inst) ‘stealthily’, in the bccwj shows that they occur with manner verbs/verbal complexes relatively frequently (50.5% and 41.7%, respectively) and rarely with a simplex deictic verb (4.0% and 4.2%, respectively). When these adverbs are used, specificity is also expected in the verbs, similar to ideophones.16
The present findings also raise the question of whether similar tendencies occur in other languages. In this regard, some languages provide data worth careful examination. In the medal project on motion event descriptions (Matsumoto, 2025), Kathmandu Newar (Tibeto-Burman) is the language in which ideophonic adverbs were most often used in the results of a video-based elicitation experiment. In this language, ideophones such as phatā-phata ‘walking steadily’ were used in 10.6% of responses for walking scenes (Matsuse, 2025). The language does not have a manner verb for walking, but it has complex predicates with a nominalized form of an archaic verb and a deictic verb (nyāsi ‘walking’ + wane ‘go’/waye ‘come’), which were used in only 14.4% of all responses for walking scenes. Crucially, ideophones occur with nyāsi +wane/waye in only 17.4% of their uses in walking scenes, not showing a strong tendency to occur with the predicates. Instead, they occur frequently with a simplex deictic verb. Possible reasons for this are the restricted use and formal complexity of the nyāsi +wane/waye complex, which give the speaker a reason not to choose it. A close examination of motion verbs and motion ideophones in Kathmandu Newar would reveal more about the relationship between them in light of the pragmatics of lexical choices.
6 Conclusion
This paper has shown that Japanese motion ideophones tend to occur disproportionally more often with manner verbs, though they do occur frequently with path verbs as well. Particular ideophones tend to occur with motion verbs with related meanings. They tend not to occur with simplex deictic verbs, and this may be due to the pragmatics of informativeness. This study further shows that a detailed semantic study of ideophones and a close examination of the corpus for frequencies are essential for understanding the semantic nature of expressions.
Acknowledgement
In writing this paper I am grateful to Yuqi Yin and Yuko Morokuma for their assistance in preparing frequency data for motion verbs. I am also indebted to Kimi Akita for discussion. This work has been supported by kakenhi 19H01264 and 23H00629 and ninjal project “Data-oriented Typological Study of the Semantics and Grammar of Predicates.”
References
Akita, Kimi. 2012. Toward a frame-semantic definition of sound-symbolic words: A collocational analysis of Japanese mimetics. Cognitive Linguistics 23(1): 67–90.
Akita, Kimi. 2019. Mimetics, gaze, and facial expression in a multimodal corpus of Japanese. In K. Akita and P. Pardeshi (eds.), Ideophones, Mimetics and Expressives, 229–247. Amsterdam: John Benjamins.
Akita, Kimi and Mark Dingemanse. 2019. Ideophones (mimetics, expressives). In M. Aronoff (ed.), Oxford Research Encyclopedia of Linguistics. Oxford: Oxford University Press.
Akita, Kimi and Yo Matsumoto. 2020. A fine-grained analysis of manner salience: Experimental evidence from Japanese and English. In Y. Matsumoto and K. Kawachi (eds.), Broader Perspectives on Motion Event Descriptions, 143–179. Amsterdam & Philadelphia: John Benjamins.
Chen, Yi-Ting and Yo Matsumoto. 2018. Nihongo Goiteki Fukugoodooshi no Imi to Taikei [The Semantics and System of Japanese Lexical Compound Verbs]. Tokyo: Hituzi Syobo.
Davies, Catherine and Jennifer Arnold. 2019. Reference and informativeness: How context shapes referential choice. In C. Cummins and N. Katsos (eds.), The Oxford Handbook of Experimental Semantics and Pragmatics, 474–493. Oxford: Oxford University Press.
Dingemanse, Mark and Asifa Majid. 2012. The semantic structure of sensory vocabulary in an African language. In N. Miyake, D. Peebles, and R. P. Cooper (eds.), Proceedings of the 34th Annual Conference of the Cognitive Science Society, 300–305. Austin, TX: Cognitive Science Society.
Fillmore, Charles J. 1982. Frame semantics. In Linguistic Society of Korea (ed.), Linguistics in the Morning Calm, 111–138. Seoul: Hanshin Publishing.
Fillmore, Charles J. and Colin Baker. 2015. A frames approach to semantic analysis. In B. Heine and H. Narrog (eds.), The Oxford Handbook of Linguistic Analysis (2nd ed.), 791–816. Oxford: Oxford University Press.
Grice, Paul. 1975. Logic and conversation. In P. Cole and J. Morgan (eds.), Syntax and Semantics, 3: Speech Acts, 41–58. New York: Academic Press. Reprinted in Paul Grice, Studies in the Way of Words, 22–40. Cambridge, MA: Harvard University Press.
Hamano, Shoko. 1998. The Sound-Symbolic System of Japanese. Tokyo: Kurosio.
Hirose, Masayoshi. 1981. Japanese and English Contrastive Lexicology: The Role of Japanese “Mimetic Adverbs.” Ph.D. dissertation, University of California, Berkeley.
Hirschberg, Julia. 1985. A Theory of Scalar Implicature. PhD dissertation, University of Pennsylvania.
Ibarretxe-Antuñano, Iraide. 2005. Leonard Talmy. A windowing to conceptual structure and language: Part 1: Lexicalisation and typology. Annual Review of Cognitive Linguistics 3(1): 325–347.
Ibarretxe-Antuñano, Iraide. 2006. Sound Symbolism and Motion in Basque. Münich: Lincom Europa.
Ibarretxe-Antuñano, Iraide. 2019. Towards a semantic typological classification of motion ideophones: The motion semantic grid. In K. Akita and P. Pardeshi (eds.), Ideophones, Mimetics and Expressives, 137–166. Amsterdam: John Benjamins.
Kakehi, Hisao, Ikuhiro Tamori and Lawrence Schourup. 1996. Dictionary of Iconic Expressions in Japanese. Berlin & New York: Mouton de Gruyter.
Kimura, Hideki. 1976. Chī le dàbǐng to kāi le dàmén－Gendai chuugokugo no kanryoo-soo doohinkoozoo ni kakawaru hitotsu no mondai [Chī le dàbǐng and kāi le dàmén: One issue concerning the perfective verb-complement structure in Contemporary Chinese]. Computational Analyses of Asian and African Languages 6: 43–50.
Koga, Hiroaki. 2025. Motion event descriptions in Japanese: The use of verbal complexes and its impact on typological issues. In Y. Matsumoto (ed.), Motion Event Descriptions across Languages, vol. 1, Case Studies of Linguistic Representations of Motion, 507–547. Berlin: De Gruyter Mouton.
Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation. Chicago: University of Chicago Press.
Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: mit Press.
Matsumoto, Yo. 1995. The conversational condition on Horn scales. Linguistics and Philosophy 18: 21–60.
Matsumoto, Yo. 1996. Complex Predicates in Japanese: A Syntactic and Semantic Study of the Notion ‘Word’. Stanford, CA: csli Publications.
Matsumoto, Yo. 1997. Kuukan-idoo no gengo-hyoogen to sono kakuchoo [Linguistic expressions of spatial motion and their extensions]. In Sh. Tanaka and Y. Matsumoto, Kuukan to Idoo no Hyoogen [Expressions of Space and Motion], 125–230. Tokyo: Kenkyusha.
Matsumoto, Yo. 2017. Nihongo ni okeru jishoo-hyoogen-taipu to keiro no hyoogen [Event-based expression types and expressions of path in Japanese]. In Y. Matsumoto (ed.), Idoo-hyoogen no Ruikeiron [The Typology of Motion Expressions], 247–274. Tokyo: Kurosio.
Matsumoto, Yo. 2020. Motion verbs in Japanese. In M. Aronoff (ed.), Oxford Research Encyclopedia of Linguistics. Oxford: Oxford University Press.
Matsumoto, Yo. 2021. The semantic differentiation of verb-te verb complexes and verb-verb compounds in Japanese. In T. Kageyama and P. Pardeshi (eds.), Verb-Verb Complexes in Asian Languages, 139–164. Oxford: Oxford University Press.
Matsumoto, Yo. 2025. Introduction: ninjal Project on motion event descriptions across languages (medal). In Y. Matsumoto (ed.), Motion Event Descriptions across Languages, vol. 1, Case Studies of Linguistic Representations of Motion, 1–52. Berlin: De Gruyter Mouton.
Matsumoto, Yo, Kimi Akita and Kiyoko Takahashi. 2017. The functional nature of deictic verbs and the coding patterns of deixis: An experimental study in English, Japanese, and Thai. In I. Ibarretxe-Antuano (ed.), Motion and Space across Languages: Theory and Applications, 95–122. Amsterdam: John Benjamins.
Matsuse, Ikuko. 2025. Motion event descriptions and deictic verbs in Kathmandu Newar: An experimental study. In Y. Matsumoto (ed.), Motion Event Descriptions across Languages, vol. 1, Case Studies of Linguistic Representations of Motion, 589–626. Berlin: De Gruyter Mouton.
Schaefer, Ronald P. 2001. Ideophonic adverbs and manner gaps in Emai. In E. F. K. Voeltz and C. Kilian-Hatz (eds.), Ideophones, 339–354. Amsterdam & Philadelphia: John Benjamins.
Slobin, Dan. 1996. Two ways to travel: Verbs of motion in English and Spanish. In M. Shibatani and S. A. Thompson (eds.), Grammatical Constructions: Their Form and Meaning, 195–217. Oxford: Oxford University Press.
Snell-Hornby, Mary. 1983. Verb Descriptivity in German and English: A Contrastive Study in Semantic Fields. Heidelberg: Carl Winter Universitätsverlag.
Stefanowitsch, Anatol and Stefan Th. Gries. 2003. Collostructions: Investigating the interaction between words and constructions. International Journal of Corpus Linguistics 8(2): 209–243.
Stosic, Dejan. 2019. Manner as a cluster concept: What does lexical coding of manner of motion tell us about manner? In M. Aurnague and D. Stosic (eds.), The Semantics of Dynamic Space in French: Descriptive, Experimental and Formal Studies on Motion Expression, 142–177. Amsterdam: John Benjamins.
Talmy, Leonard. 2000. Toward a Cognitive Semantics, vol.ii: Typology and Process in Concept Structuring. Cambridge, MA: The mit Press.
Toratani, Kiyoko. 2012. The role of sound-symbolic forms in motion event descriptions. Review of Cognitive Linguistics 10(1): 90–132.
Van Hoey, Thomas. 2024. A semantic map for ideophones. In Th. F. Li (ed.), Handbook of Cognitive Semantics Online. Leiden: De Gruyter Brill.
The quotative marker -to is obligatory for some ideophones, but otherwise it is optional. This marker is indicated in parentheses when it is optional.
See Stefanowitsch and Gries (2003) for examples of such statistical analyses.
Citation forms in Japanese verbs include the non-past tense marker -(r)u, which is not glossed in the text.
There are some complex manner verbs consisting of an ideophonic root and a verbalizing suffix that encode attitudinal meanings (e.g., uro-tsuk(-u) ‘loiter’, based on the root uro used in uro-uro(-to) ‘loitering’).
The verb nobor(-u) can be used with a subject nominal representing a person who uses an elevator. In this case the event is treated as a willful motion of a person.
The meanings of compound verbs can be non-compositional, containing meanings that do not directly reflect those of constituent verbs (Chen & Matsumoto, 2018). The verb uri-aruk(-u) ‘walk around selling’ can be used even if the sales trip partially involves automobile use. However, some walking must be included, since this verb cannot be used if, e.g., only a boat is used for motion. For this reason, the second verb in this compound still retains a manner meaning.
There are compound verbs with ik(-u) ‘go’ used as a V1, as in iki-tsuk(-u) (go-arrive) ‘reach’, etc. (Matsumoto, 2021). Interestingly, in such cases ik(-u) does not have a deictic meaning, as can be seen in the acceptability of koko-ni iki-tsuk(-u) ‘reach here’, and the lack of a contrasting form with ku(-ru) ‘come’ (i.e., *ki-tsuk(-u) (come-arrive)).
Some information conveyed by ideophones is based on morphological patterning and voicing contrast (see Hamano, 1998; Akita, 2009). Reduplication (e.g., guru-guru(-to) ‘moving in a big circle’ as opposed to grut-to ‘going all the way around’) signifies repeated or prolonged processes, and the voicing of the initial consonant of roots (e.g., guru-guru(-to) as opposed to kuru-kuru(-to) ‘moving in a small circle’) signifies an increased intensity, size, or negative connotation.
A semantic grid analysis is also useful in the analysis of manner verb meanings. Many English manner verbs (e.g., scurry, totter) similarly encode a gestalt of multiple information. See also Stosic (2019) for components of manner encoded in manner verbs.
A reviewer questioned my analysis of the difference between tsuka-tsuka(-to) ‘walking without hesitation’ (involving an attitude) and suta-suta(-to) ‘walking briskly’ (not involving an attitude). My small study suggests that some speakers, including myself, have the sense of determination/no hesitation in the meaning of the former. (Note that tsuka-tsuka(-to) yatte kuru toyuu taido ‘the attitude of coming without hesitation’ is acceptable [see also Kakehi, Tamori and Schourup, 1996: 1202]. This attitudinal meaning is clearer in its voiced counterpart zuka-zuka(-to) ‘going somewhere rudely’.) Other speakers, however, do not seem to have a clear sense of this component. For such speakers, this ideophone would be placed in the “brisk/energetic human movements” category.
One may note that, unlike manner and path verbs, which show clear distributional differences distinguishing them, no such ordering constraints can be found in ideophones. For instance, suta-suta guru-guru aruki-mawar(-u) ‘walk around briskly’ and guru-guru suta-suta aruki-mawar(-u) ‘walk around briskly’ are both acceptable. It does not seem to be possible to find a distributional criterion to distinguish manner-based and path-based ideophones. The distinction between manner ideophones and path ideophones appears to be more blurred than the one between manner verbs and path verbs.
In this way, examples with nominals that are incompatible with motion uses were listed and excluded. In some cases, such as with sugi(-ru) ‘pass’, koe(-ru) ‘move over’, and araware(-ru) ‘appear’, there were far more non-motion uses than motion uses. In such cases, the motion uses were identified by the co-occurring case-marked nominals that are compatible with motion uses. These procedures were also taken to identify the motion uses of compound verbs and V-te V complex predicates.
Details of the data are available at https://osf.io/nfza3/.
The relatively low frequency of the verb of walking in Japanese is due to the omission of walking information when it is inferable from the context (Koga, 2025).
Akita (2010) points out that ideophones can represent deixis through suprasegmental features, with increasing loudness signifying approach to the speaker. Interestingly, such an indication of deixis in ideophones is most natural when occurring with a complex predicate with a deictic verb rather than a simplex manner verb, as in (i):
This means that the specificity of ideophones requires predicates that are also specific in terms of the information expressed in the ideophones.
Interestingly, this is not the case at all with the manner adverbs kake-ashi-de ‘at a run’ and toho-de ‘on foot’, which infrequently occur with manner verbs/verbal complexes (15.9% and 14.6%, respectively), and one of them occur more often with a simplex deictic verb (4.0% for kake-ashi-de and 25.8% for toho-de). In these cases the information carried by these adverbs is not so much different from information conveyed by manner verbs hashir(-u) ‘run’ and aruk(-u) ‘walk’ and so the use of manner verbs with them is redundant. Thus, the avoidance of redundancy can override the tendency to be informative in verbs.
appendix 1 List of simplex motion verbs whose frequencies are examined
M: aruk(-u) ‘walk’, hane(-ru) ‘bounce’, hashir(-u) ‘run’, haw(-u) ‘crawl’, isog(-u) ‘hurry’, kake(-ru) ‘run’, korogar(-u) ‘roll’, maw(-u) ‘dance in the air’, nagare(-ru) ‘flow’, oyog(-u) ‘swim’, shitatar(-u) ‘drip’, sube(-ru) ‘slip’, tob(-u) ‘fly, jump’
P: agar(-u) ‘ascend’, araware(-ru) ‘appear’, chikayor(-u) ‘draw near’, chikazuk(-u) ‘approach’, dekake(-ru) ‘depart’, de(-ru) ‘exit’, hair(-u) ‘enter’, kae(-ru) ‘return (home)’, kayow(-u) ‘commute’, koe(-ru) ‘move over’, kudar(-u) ‘move down’, kugur(-u) ‘move under’, mawar(-u) ‘move around’, megur(-u) ‘move around’, modor(-u) ‘return’, mogur(-u) ‘move into water’, mukaw(-u) ‘head for’, nige(-ru) ‘flee’, nobor(-u) ‘climb up’, nuke(-ru) ‘pass through, fall out’, ochi(-ru) ‘fall’, ori(-ru) ‘move down, get off’, sar(-u) ‘leave’, shizum(-u) ‘sink’, sugi(-ru) ‘move past’, susum(-u) ‘proceed’, toor(-u) ‘pass by’, tsuk(-u) ‘arrive’, watar(-u) ‘cross, move to other side’, yokogir(-u) ‘cross’, yor(-u) ‘draw near, stop by’
D: ik(-u) ‘go’, ku(-ru) ‘come’
appendix 2 List of compound motion verbs whose frequencies are examined
M-P:aruki-mawar(-u) ‘walk around’, aruki-sar(-u) ‘walk away’, ayumi-sar(-u) ‘walk away’, ayumi-yor(-u) ‘walk toward’, fuki-agar(-u) ‘blow up’, fuki-kom(-u) ‘blow in’, hai-agar(-u) ‘crawl up’, hai-das(-u) ‘crawl out’, hai-mawar(-u) ‘crawl around’, hairi-kom(-u) ‘enter in’, hane-agar(-u) ‘bounce up’, hane-mawar(-u) ‘bounce around’, hashiri-kom(-u) ‘run in’, hashiri-mawar(-u) ‘run around’, hashiri-sar(-u) ‘run away’, kake-agar(-u) ‘run up’, kake-das(-u) ‘run out’, kake-kom(-u) ‘run in’, kake-mawar(-u) ‘run around’, kake-megur(-u) ‘run around’, kake-modor(-u) ‘run back’, kake-ori(-ru) ‘run down’, kake-tsuke(-ru) ‘run over to’, kakezuri-mawar(-u) ‘scurry about’, korogari-de(-ru) ‘roll out’, korogari-kom(-u) ‘roll in’, korogari-mawar(-u) ‘roll around’, korogari-ochi(-ru) ‘roll down’, koroge-ochi(-ru) ‘tumble down’, mai-agar(-u) ‘soar up’, mai-chir(-u) ‘flutter down’, mai-kom(-u) ‘flutter in’, mai-modor(-u) ‘dance back’, mai-ochi(-ru) ‘flutter down’, mai-ori(-ru) ‘fly/float down’, nadare-kom(-u) ‘avalanche into’, nagare-kom(-u) ‘flow in’, nagare-ochi(-ru) ‘flow down’, nagare-sar(-u) ‘flow away’, nagare-tsuk(-u) ‘flow/drift to’, notauchi-mawar(-u) ‘thrash around’, odori-agar(-u) ‘jump up dancing’, oyogi-mawar(-u) ‘swim around’, shinobi-kom(-u) ‘sneak in’, suberi-kom(-u) ‘slide in’, suberi-ochi(-ru) ‘slide/slip down’, suberi-ori(-ru) ‘slide down’, tobi-agar(-u) ‘jump up’, tobi-das(-u) ‘jump out’, tobi-koe(-ru) ‘jump over’, tobi-kom(-u) ‘jump in’, tobi-mawar(-u) ‘jump/fly around’, tobi-nor(-u) ‘jump onto’, tobi-ori(-ru) ‘jump/fly down’, tobi-sar(-u) ‘fly away’, ukabi-agar(-u) ‘float up, emerge’, uki-agar(-u) ‘float up’
M-M: mai-tob(-u) ‘fly fluttering’, tobi-hane(-ru) ‘bounce’,
P-P:mawari-kom(-u) ‘go around in’, moguri-kom(-u) ‘dive in’, nige-das(-u) ‘escape from’, nige-kaer(-u) ‘escape back’, nige-kom(-u) ‘escape into’, nige-mawar(-u) ‘flee around’, nuke-das(-u) ‘get out of’, ochi-kom(-u) ‘fall in’, shizumi-kom(-u) ‘sink in’, sugi-sar(-u) ‘pass away’, toori-nuke(-ru) ‘pass through’, toori-sugi(-ru) ‘pass by’, tsuk-kom(-u) ‘plunge in’
A-M:nomi-aruk(-u) ‘walk around drinking’, tabe-aruk(-u) ‘walk around eating’, tazune-aruk(-u) ‘walk visiting’, uri-aruk(-u) ‘walk around selling’
A-P:fumi-kom(-u) ‘step in’, hik-kom(-u) ‘draw back’, kagi-mawar(-u) ‘stiff around’, magire-kom(-u) ‘blend into’, mi-mawar(-u) ‘go looking around, patrol’, mochi-agar(-u) ‘be lifted up’, nori-kom(-u) ‘get on board’, sagashi-mawar(-u) ‘go around searching’, seme-kom(-u) ‘attack/press into’, shimi-kom(-u) ‘soak into’, tachi-mawar(-u) ‘move about busily’, tachi-modor(-u) ‘return to the starting point’, tachi-sar(-u) ‘leave’, tachi-yor(-u) ‘stop by’, uchi-agar(-u) ‘be launched up’, ugoki-mawar(-u) ‘move about’, wari-kom(-u) ‘force one’s way in’
