Send message to Author

Paper short abstract

This paper re-evaluates Wenck (1959)’s hypothesis of intervocalic obstruent voicing in Old and Early Middle Japanese through comparison with Old Tuscan evidence, suggesting that voicing may be gradient and non-systematic, resulting from a probabilistically conditioned irregular sound change.

Paper long abstract

To make the well-known lenition pathway /-p-/ > /-w-/, traditionally labelled hagyō tenko, phonologically plausible, Wenck (1959: 81) proposed that /p/ was allophonically voiced word-internally in Old and Early Middle Japanese, and that hagyō tenko proceeded through successive weakening stages [b > *bβ > *β > w > ∅], with segmental loss blocked before /a/. Wenck further suggested that voicing extended to other obstruents in intervocalic position, adducing typological parallels from Korean and Northeastern Japanese dialects. From a sound-change perspective, this proposal offers a convincing basis for the lenition processes affecting the syllables involved in the so-called onbin changes. It has been supported, albeit with differing emphases, by Hayata (1977), Takayama (1992), Frellesvig (1991, 1995, 2010), and Hamano (2000), with partial agreement by Unger (2004). Nevertheless, many Japanese scholars continue to interpret hagyō tenko as a process of fricativisation, typically reconstructed as *p > *ɸ > w (Mabuchi 1971; Shibatani 1990, among others). Vovin (2020) likewise rejects systematic intervocalic voicing, appealing primarily to man’yōgana orthography, which does not consistently reflect such voicing. Beyond the orthography, a blanket voicing hypothesis raises two further issues: (i) it does not explain why lenition seems to be arrested in syllables that did not undergo onbin, and (ii) it entails a later fortition reversal, namely intervocalic devoicing in Late Middle Japanese, that is cross-linguistically uncommon and therefore difficult to justify. This paper reassesses these competing accounts in light of Romance historical phonology. In many Romance varieties, Latin intervocalic obstruents underwent voicing and subsequent lenition, yielding trajectories comparable to Wenck’s chain. Crucially, however, Old Tuscan did not develop an exceptionless intervocalic voicing rule; instead, weakening was selective, gradient, and sensitive to consonant type, prosodic position, and lexical diffusion (Canalis 2014, 2015). Treating intervocalic voicing as probabilistic rather than categorical thus offers a comparative model in which non-automatic, non-systematic outcomes may be expected. Applied to Japanese, this perspective supports a scenario in which voicing and lenition spread differentially across consonant types and environments, allowing the same intervocalic segment to lenite in some words but not others, without positing uniform system-wide voicing or an implausible later fortition change.

Send message to Author

Paper short abstract

This study investigates referential density in the history of Japanese using manually annotated data and corpus analysis. I demonstrate a diachronic increase in overt argument realisation and show that it is conditioned by morphosyntactic change, genre and register variation, and language contact.

Paper long abstract

This paper investigates referential density (RD) in the history of Japanese, asking how reference is distributed across overt argument NPs, pronominal forms, and morphology. Building on Bickel’s (2003) RD1 (overt argument NPs per available argument positions) and Noonan’s (2003) RD2 (overt arguments per verb), I introduce two indices capturing the referential contribution of exaltation morphology (keigo): RDex1 (markers per available arguments) and RDex2 (markers per verb).

Because RD is best assessed in extended discourse, the Old Japanese (OJ) component draws on the only substantial prose sources, Shoku-Nihongi Senmyō and Engi-shiki Norito, alongside their Contemporary Japanese translations. I triangulate the annotated results with corpus evidence (Oxford-NINJAL Corpus of Old Japanese and Kainoki Treebank), and extend the analysis to Early and Late Middle Japanese (EMJ – LMJ) materials spanning courtly narrative, diary prose, and contact-mediated didactic and translation prose.

Two findings stand out: (i) RD1 is broadly stable, while RD2 increases diachronically, indicating more overt arguments per predicate in later stages; and (ii) RDex decreases, reflecting reduced reliance on exaltation morphology as a referential resource.

The EMJ – LMJ comparison shows that, alongside morphosyntactic change, genre and register strongly condition RD. Courtly narrative and performative registers sustain low RD via morphology-mediated reference tracking and topic continuity, whereas expository and didactic prose favours overt NPs. Detailed annotation of Heike Monogatari suggests that exaltation-related effects on argument realisation are construction-sensitive, clustering in specific predicate morphologies and discourse environments.

Finally, I argue that language contact adds further pressure towards overt reference. Late Middle Japanese Christian translation prose, rich in light-verb constructions (LVCs) with Sino-Japanese verbal nouns, reshapes predicate packaging and argument expression. Likewise, post-Meiji contact with European languages correlates with increased pronoun use and more explicit participant tracking in translated and modern written styles. Methodologically, the paper highlights challenges in annotating zero forms and morphology-driven reference in historical corpora, and the resulting limits of corpus-only RD measures.

This study forms part of my DPhil project on reference tracking and argument realisation in Japanese.

Send message to Author

Paper short abstract

In Early Middle Japanese, predicates lacking tense-aspect suffixes are tenseless: time reference is inferred from boundedness (presence/absence of a temporal boundary). Bounded clauses are perfective and refer to specific past episodes; unbounded clauses report regularities concerning the present.

Paper long abstract

Narrative prose predominantly reports past events. Yet ninth–eleventh-century Classical Japanese prose contains remarkably few instances of the past tense suffixes -ki and -keri (e.g. ariki-ki, ariki-keri ‘walked’) relative to the number of past-time situations in the texts. Instead, past-time reference is typically associated with the perfective suffixes -tsu and -nu, or left without overt tense–aspect marking (i.e. zero-marked).

Many predicates referring to past events therefore lack tense–aspect morphology (e.g. ariku ‘walked’). Although zero-marked predicates may refer to past or present time, the grammatical principles determining their temporal construal—and their aspectual profile more generally—have remained largely unexplained.


In this paper, I argue that zero-marked predicates instantiate one of two aspectual profiles: one correlates with present-time reference, the other with past-time reference. In a wider study of 700 Classical Japanese examples, I show that, in the absence of overt tense–aspect marking, temporal construal is inferred pragmatically from boundedness—that is, the presence or absence of implicit or explicit reference to a temporal boundary within the clause. I identify boundedness on the basis of boundary cues (e.g., temporal adverbials of position, or overt specification of a preceding or following event).


Bounded clauses yield episodic perfective construal and are understood as referring to specific past events; unbounded clauses yield imperfective habitual/generic construal and are understood as reporting regularities concerning the present.

The two aspectual profiles are exemplified below. In the first example, the predicate 過ぐ sugu denotes a bounded situation and is understood as referring to a past episode:


またの日、山の端に日のかかるほど、住吉の浦を過ぐ。


“On the following day, just as the sun was setting behind the mountain ridge, we passed by Sumiyoshi Bay.”
 (Sarashina Nikki)


In the second, the predicate 食ふ kuu denotes an unbounded situation and is construed as a habit (imperfectively) related to the narrative present.


こと物は食はで、ただ仏の御おろしをのみ食ふか。


“Do you refrain from eating ordinary food and eat only altar offerings?”
(Makura no Sōshi, §83)

Collectively, these patterns indicate that zero-marked predicates in Early Middle Japanese are genuinely tenseless, with past or present time reference emerging from a systematic interaction between aspectual profile and pragmatic interpretation. These findings support an analysis of Early Middle Japanese as a tenseless language.

Send message to Author

Paper short abstract

In Japanese, first-person desiderative suffixes may exceptionally refer to a non–first-person entity when attached to existential verbs. This paper examines modern and classical desiderative markers and existential verbs to clarify the distribution and underlying causes of the phenomenon.

Paper long abstract

The Modern Japanese desiderative suffix -tai generally implies identity between speaker and agent and requires a first-person subject, a property shared with adjectives of emotion (kanjō keiyōshi) and known as person restriction (ninshō seigen). However, its combination with existential verbs gives rise to a peculiar semantic effect. In such cases, the resulting forms exceptionally permit the target of desire (i.e. the subject of the existential predicate) to be a non–first-person entity.

A concrete illustration of this phenomenon is provided by the combination of -tai with the existential verb aru, yielding the form aritai, which allows a reading in which the desire is directed toward the existence or state of affairs involving a third-person entity. For example, in “jibun no jinsei mo kō aritai” (BCCWJ PM22 00007) "I want my life to be like this too", the desiderative predicate does not express the speaker’s wish to exist, but rather a wish concerning how a third-person entity (“one’s life”) should exist. This phenomenon is not limited to the Modern Japanese suffix -tai, but is also observed with some classical desiderative markers, such as -mafosi and -baya.

The extent to which this relaxation of the person restriction arises is not uniform, and appears to vary depending both on the desiderative marker involved and on the particular existential verb with which it combines. For instance, while some markers tend to retain the original alignment between speaker and agent, -mafosi more consistently allow an agent shift.

The present study examines Modern (-tai) and Classical (-tasi, -mafosi, -baya, -(te/ni)sika) desiderative suffixes attested in combination with existential verbs. Other Old/Classical markers (-na, -koso, -namu), strictly non-desiderative but exhibiting related semantics in combination with existential verbs, will also be taken into account. First, newly collected data are used to clarify the distribution and usage of each suffix–verb combination, also taking diachronic change into consideration for classical lexemes. Finally, the study addresses the broader question of why desiderative morphology interacts differently with existential predicates, and why such interactions vary across individual lexemes. The analysis evaluates to which extent these effects reflect a shared cause, marker-specific semantics, and diachronic developments.