The Diachronic Shift of Japanese Transitive/Unaccusative Verb Pairs

This study explores how Japanese transitive/unaccusative verb pairs have transformed from being a substantive verb to the various forms they fulfil in Modern Japanese (i.e., an aspectual verb, a noun, an adjective, an adverb, a quantifier and a suffix) and how grammaticalisation and lexicalisation play an essential role during the processes. A working definition of ‘grammaticalisation’ and ‘lexicalisation’ that applies to Japanese is put forward, followed by a corpus-based investigation as well as a case study. The finding reveals that (a) the process by which a lexeme develops into a noun is a case of lexicalisation; the process by which a lexeme develops into an aspectual verb, an adverb, an adjective, a suffix or a quantifier is a case of grammaticalisation; (b) transitive verbs are more likely to convey ASPECT than unaccusatives are. The shift into a quantifier is limited to unaccusative verbs. Grammaticalisation (affixation) and lexicalisation in Japanese both require syntactic reduction and morphological alternation. The two differ in that lexicalisation does not require an alternation in writing, i.e., a lexicalised item can remain being written in Chinese characters (Note 1) whilst a grammaticalised item can only appear in kana script. Phonological alternation is obligatory in grammaticalisation but not required by lexicalisation. Lexicalisation appears to occur before grammaticalisation.

The purpose of this study is two-fold. First, it seeks to establish which form is the original base, and which is the derived form.
Second, transitive/unaccusative verbs have both derived new categories and meanings over the evolution of the language. For instance, in Old Japanese corpus, motsu was transitive in origin, meaning 'to carry something' (6a) and got lexicalised in 1707, fulfilling a noun function (6b). It further derived an unergative verb function in 1821 (6c). The shift from a VP (object + transitive verb) to a compound noun is an issue of lexicalisation due to the omission of the accusative case particle o and the morphological transition (motsu (conclusive form) → moti (continuous form)). body NOM hold.CONCL In (6b), moti is the continuous form of transitive verb motsu (carry). Syntactically, the compound noun okanemochi derives from the VP: okane o motsu [money ACC have]. When the accusative case particle 'o' is omitted, the VP turns into a compound noun. Semantically, the compound noun okane-mochi refers to a person who has lots of money, i.e., 'okane-o-motteiru-hito' [money-acc-possess. PROG -person].
(6c) is a case where the transitive verb motsu shifting into an unergative verb. The unergative motsu rather renders a state whilst the transitive motsu conveys an action, which inspires us to deduce that the agentivity of mostu has been weakened.
It occurs that the boundary of grammaticalisation and lexicalisation is fluid. This raises the second question as to how the two transition processes are distinct yet interrelated. For example, whether it is reasonable to postulate that the paths might be predictable for the two roots, in that a transitive root may be favoured for lexicalisation while an unaccusative root facilitates grammaticalisation, or the other way around.

Functions and Meanings of Transitive/Unaccusative Pairs in Modern Japanese
A search in the dictionary Koojien reveals that quite a few transitive verbs in Modern Japanese are multi-faced, i.e., function as a substantive transitive verb, a noun, an aspectual verb, an adverb, and a suffix. Consider kiru tran. 'cut', a typical action verb, for an instance. In the corpus of BCCWJ, kiru is found 5034 times, fulfilling seven categories. A total of 300 words were randomly selected and show that following distribution of kiru's multiple characters. The most frequent option is the transitive verb use (88 tokens). A total of 76 tokens had an aspectual function. 67 tokens were attributed to the transitive verb use with a lexicalised meaning. 58 went to the noun use. Six tokens showed an adverb use and three showed an adjective use. Finally, only two tokens were used as a suffix. Turning to the counterpart of kiru, i.e., the unaccusative kireru 'cut intran .'. The Koojien Dictionary indicates that kireru conveys five categories in Modern Japanese: an unaccusative verb, unaccusative verb with meaing lexicalised, a suffix, a noun, an aspectual verb, and a quantifier. Illustrations are provided in (8).  Based on the previous discussion, we can draw a comparison of the pair transitive kiru and unaccusative kireru as follows.  Table 3 suggests that: (i) There are more types and tokens of kiru than kireru with a substantive verb function.
(ii) Kireru is detected as a quantifier but there is no data confirming kiru as a quantifier.
(iii) The aspectual use in kireru has more tokens than the aspectual use in kiru.
(iv) There are more tokens of kiru (59 tokens) with a noun use than kireru (8 tokens).
(vi) Both kiru and kireru has a suffix function.
(vii) Kiru bears two additional functions, as an adverb and an adjective.
The process by which a lexeme develops into an aspectual verb, a quantifier, a suffix is a case of grammaticalisation; the process by which lexeme develops into a noun is a case of lexicalisation. Building on this, we see that both transitives and unaccusatives are involved with grammaticalisation as well as lexicalisation. This study looks at the evolution of transitive/unaccusative verb pairs, exploring their base and, how they have transformed from being a substantive transitive/unaccusative verb to the various forms them fulfil in Modern Japanese, namely an aspectual verb, a noun, an adjective, a suffix, a quantifier and an adverb.
Action verb kireru

Transitive Verb and Unaccusative Verbs in Old Japanese
Old Japanese (the Asuka and Nara periods: 700-800) is a dead language. The pure phonetic kana script had not been developed yet and thus Chinese characters were employed to represent the vernacular on paper. Three writing systems existed: Junsei-kanbun 'purely classical Chinese' (official documents), hentai-kanbun 'variant Chinese' (poems and tales) and man'yōgana (poems). The following illustration comes from Man'yōshū, vol.5, completed in the Nara Period (710 -794 A.D.), when the Japanese language is recorded for the first time.

Grammaticalisation and Lexicalisation
Before a corpus-based investigation, it is necessary to introduce two concepts that play an essential role in the evolution of a language, viz. grammaticalisation and lexicalisation.
Grammaticalisation refers to the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions, and, once grammaticalised, continue to develop new grammatical functions (Hopper & Traugott, 2003). The study of grammaticalisation has undergone for centuries. Bopp (1816), Schlegel (1818) and Humboldt's (1825) pilot research links it with evolutionary change for the first time. Meillet (1912) paid attention to the transformation instead of the grammatical forms and came up with the terminology 'grammaticalisation'. About a half century later, German linguist Lehmann (1982) measured grammaticality both synchronically and diachronically. Heine and Reh's (1984) work introduce the term into typological study on African languages, which kicked out the boom of the study on grammaticalisaiton in linguistic typological work. One crucial issue on the topis lies in that, it is rather difficult to capture all the grammaticalisation phenomena in diverse language families with one definition. Jerzy Kurylowicz (1965) paid attention to transition of morpheme, referring to grammaticalisation as consisting in the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status. Christian Lehmann (1982) in his Thoughts on Grammaticalization and New Reflections on Grammaticalisation and Lexicalisation, broaden the scope, i.e., grammaticalization is a process leading from lexemes to grammatical formatives. A number of semantic, syntactic and phonological processes interact in the grammaticalisation of morphemes and of whole constructions.
The transformation of content word into a function word takes place by a gradual series of shifts. This very shift follows a certain pattern, known as 'unidirectionality' (Heine & Kuteva, 2002), i.e., grammaticalisation always leads from less grammatical to more grammatical forms and constructions. This pattern represents human cognition and is cross-linguistically applied. A more specific evolution path is put forward by Hopper and Traugott (2003).
(17) content item > grammatical word > clitic > inflectional affix Another pathway may come from lexical semanticians who contend that grammaticalisation is perhaps a subtype of metaphor after searching on the shift of semantic meaning. A different view on grammaitcalisation comes from Himmelmann (2004, p. 31) and Traugott (2003, p. 6) who asks for the role of context on grammaticalisation.
Japanese, is an agglutinative language, featured with phono-morphologically, sequential voicing. Morpho-syntactically, verb compounding. The Indo-European language-oriented notions do not seem to apply to Japanese, given the typological distinctions between Indo-European languages and Altaic languages. First, the meaning bleaching is not enough for identifying the degree of grammaticalisation regarding Japanese. In the case of change-of-state verbs, when grammaticalised, it is the morphological change as well as the syntactic shift (decategorisation) that occurs; the semantic meanings retain. Second, as a moraic language, it is not phonological reduction as European languages display but sequential voicing (Note 13) that is an effect of lexicalisation. Sequential voicing accompanies morphological change ([-voiced] consonants shift into [+voiced]), but not morphological reduction. Third, Japanese aspectual verbs and potential verbs are an outcome of diachronic change.
Another concept involved in language evolution is lexicalisation, which refers to an item with a concrete semantic meaning that evolves into an item with an abstract or metaphorical reading.
Grammaticalisation and lexicalisation deal with different parts of transition: the former covers the syntactic function domain, and the latter covers the process of meaning extension. It appears that the development of Japanese compound nouns is a combination of grammaticalisation and lexicalisation.
For instance, shime-kiri 'deadline' derives from the compound verb shime-kiru 'to close'. Shime-kiru is a transitive compound verb, carrying two meanings: a concrete meaning 'to shut' and an abstract meaning 'to expire/to cut off'. The second constituent kiru of the compound verb shime-kiru initially appeared in Old Japanese (A.D. 759), behaving as a transitive verb and meaning 'to cut'. In shime-kiru, kiru conveys an ASPECT of the first constituent, i.e., shimeru 'close', which indicates the ACTION and is not grammaticalised. Kiru's transition from a substantive transitive verb into an aspectual verb in a compound verb can be argued to be the result of grammaticalisation.
Moreover, the conjunctive form of the verb compound is also noun form, i.e., shimekiri 'to expire/to cut off' (e.g., Gakkai wa boshū o shimekitta [The academic conference cut off the submission period]). CHJ brings us to the point that the first appearance of the nominal form, shimekiri was in 1888. This time, it is the whole compound verb shimekiru that turns into the noun form shimekiri and renders an extended meaning, i.e., 'deadline'. Since shimekiri involves morphological alternation (from conclusive form into conjunctive form), the noun shimekiri 'deadline' turns out to be lexicalisation. the process is described in (18).
Stage II: V2 being grammaticalised shime-kiru (ASPECT) Stage III: V-V being lexicalised into a noun, shimekiri 'deadline' rendering an extended meaning This invites the question whether the two processes (grammaticalisation and lexicalisation) is a constant change leading from grammaticalisation to lexicalisation?
A good deal of effort in the past has been devoted to addressing the relationship between grammaticalisation and lexicalisation. Previous work falls into two positions. One position alleges that lexicalisation is the reverse process of grammaticalisation (cf. Kuryłowicz, 1965). The other position conceives the two processes as a constant change leading from lexicalisation to grammaticalisation and vice versa. The two processes are in an orthogonal relationship (Lehmann, 2002;Himmelmann, 2004). The transition process of the above-mentioned shimekiri appears to support this view. Moreno Cabrera (1998) takes a cognitive approach, arguing that grammaticalisation is a metaphorical process while lexicalisation is a metonymical process; the two are assigned to a complementary relation. These demonstrations on the relationship between the two processes are summarised in Figure 1.

Figure 1. Previous proposals on the interrelationship between grammaticalisation and lexicalisation
The purpose of this study is three-fold: (a) To establish the original verb root as well as when a transitive or an unaccusative form is derived.
(b) How transitive/unaccusative verbs transformed from being a substantive verb to the various forms they fulfil in Modern Japanese.
(c) To investigate how the constantly changing journey of languages is accompanied by grammaticalisation and lexicalisation and, in essence, the fluid nature of the boundary between grammaticalisation and lexicalisation. For instance, the unaccusative verb kireru can either appear in a noun compound with the consonant ki being voiced, e.g., 幕切れ maku-gire [curtain-cut] 'end'; 燃料切れ nenryoo-gire [fuel-cut] 'out of fuel', etc., or can appear in a noun compound without the consonant ki being voiced, e.g., 板切れ ita-kire [wood-cut] 'a piece of wood'; 紙切れ kami-kire [paper-cut] 'a piece of paper'.
The question thus is whether the unaccusative verb kireru ought to be treated as a suffix (a process of grammaticalisation), semi-suffix or a lexicalised noun?
To answer these inquires, this study starts by providing a working definition of 'grammaticalisation' and 'lexicalisation' that apply to Japanese data. Then it turns to diachronic issues: the development of other categories. With this in place, it aims to pin down the distinctions and interrelationship between grammaticalisation and lexicalisation.
This paper is structured as follows. Section 3 presents quantitative data on the occurrence of the different uses of Japanese transitive/unaccusative word pairs and their derived forms in the various subperiods of the historical text corpus. Building on this, a hypothesis about their grammaticalisation-and/or lexicaliation processes is proposed. To further confirm the findings from Section 3, a case study on the categorial and semantic shift of the Japanese verb pair kiru/kireru (cut tran. /cut intran. ) is outlined in Section 4. Section 5 highlights the results and concludes the paper.
The data for Modern Japanese comes from the corpus of BCCWJ (Balanced Corpus of Modern Written Japanese) by the National Institute for Japanese language and linguistics. The data of Old Japanese is extracted from the CHJ (Japanese historical linguistic corpus) 2.5.0 by National Institute for Japanese language and linguistics. The data of for compound verbs are extracted from the compound verb corpus https://db4.ninjal.ac.jp/vvlexicon/db/ (last access 29 August 2020). The glossing follows the Leipzig Glossing Rules (https://www.eva.mpg.de/lingua/resources/glossing-rules.php last access 29 August 2020).
The categorisation of the development of the Japanese language is as follows:

Grammaticalisation and Lexicalisation Regarding Transitive/Unaccusative Verb Pairs
Drawing on the concepts highlighted above, this section focuses on the original root of vt/vi pairs. Building on this, we delve into how vt/vi transforms from a substantive verb to the various categories in the present days. To start, we set diagnoses for grammaticalisation regarding Japanese verb as follows: (20) Diagnoses of grammaticalisation in Japanese verb (a) The alternation of category is obligatory (either partially or fully) e.g., verb shifting into an affix (prefixation or suffixation, cf. (21)) or verb shifting into an auxiliary (22) or an aspectual verb (23), rendering RESULT, or, verb shifting into an adverb (24). Crucially, X-duki is highly productive. A search in the BCCWJ reveals about 201 types (2001 tokens) of X-duki. With this in place, the productivity and the consonant alternation (Note 14) can be deemed as a criterion of grammaticalisation.
(c) The fourth diagnosis of grammaticalisation is the alter of writing.
The lexeme that is fully grammaticalised would no longer appear in Chinese character in written form, but in kana script. See for instance, nuku. In Modern Japanese, it has two renderings: (i) when behaving as a substantive verb, it is written in the Chinese characters 抜く nuku, e.g., ha o nuku 'tooth-ACC-pull out'; (ii) when fulfilling the function of a suffix, it appears in kana script instead of 抜き nuki, e.g. asagohan nuki 'breakfast-without'.
Lexicalisation requires the following conditions.

(c) Phonological alternation
The initial consonant of the non-initial portion of a compound or prefixed word becomes voiced, e.g., gohan o tsukuru [meal ACC cook] 'cook meal' → gohan-zukuri 'meal cooking'.

Orignal Root and Derivation
In this section ten pairs are discussed. Given below is a   e. kom-was initially unaccusative (komu 'be crowded' 1887) and its transitive form komeru 'count something in' was derived in 1895.

(III) Derived at the same time
For one pair, both transitive and unaccusative form, appeared at the same time: deru 'come out'/dasu 'take out' (759 A.D.).
Having examined the base root, we are now in a position to engage with an analysis of vt/vi pairs' derivation of other functions. The following discussion starts with verb weakening. A search in the compound corpus brought us the following distribution of vt/vi pairs' derivation into an ASPECTUAL VERB.  Here, kiru is grammaticalised. The conjunctive form of shimekiru, i.e., shimekiri renders a metaphorical reading: deadline. In BCCWJ, the V-V shimekiru has merely eight types. It is its conjunctive form, shimekiri 'deadline', that appears to be mostly used (563 tokens).
This leads to a hypothesis that grammaticalisation and lexicalisation are perhaps assigned to an orthogonal relationship. To confirm this idea, the following sections draws a statistical analysis of vt/vi pairs' derivation into other categories: noun, suffix and quantifier as well as their lexicalisation possibilities.

vt/vi's Shift into Noun/Suffix/Quantifier (Grammaticalisation)
This study took 100 tokens of the targeted verb pairs at random and calculated the distribution of them behaving as noun, suffix and quantifier from the corpus CHJ. The findings are summarised in Table 6.
(b) Both transitives and unaccusatives may shift into a suffix, as illustrated in (27).
(27) Transitive: kiru → hitori kiri [one person-cut] 'all on one's own'; Unaccusative: nuku → gohan nuki [meal-pull] 'without eating' Among them, the suffixation of komu has the largest tokens, which suggests that the degree of grammaticalisation of komu is extremely high.
(c) Both unaccusative and transitive are subject to nominalisation.
(29) a. Tsuki no yoi secchakuzai 'the glue that is sticky' b. Taki ga nutteite, tsuki ga warui 'The firewood is wet, so it's hard to get burnt'  The above can be generalised as: • Three unaccusative verbs are grammaticalised.
All unaccusatives and transitives are capable of being lexicalised and grammaticalised. However, lexicalisation appears to occur before grammaticalisation. Furthermore, the process by which a lexeme develops into a noun is a case of lexicalisation; the process by which a lexeme develops into an aspectual verb, an adverb, and an adjective, a suffix and a quantifier is a case of grammaticalisation.

A Case Study on the Category and Meaning Shift
The previous section (31) has drawn an overall picture of the original root of vt/vi pairs and, the likelihood of vt/vi pairs' evolving into other grammatical categories. The findings support our hypothesis that grammaticalization and lexicalisation are assigned to an orthogonal relationship.
To confirm this hypothesis, a case study of the transition now becomes appropriate. This study selects the kiru/kireru pair, one of the most typical change-of-state verbs, as the suitable candidate. It was shown above that kiru/kireru originates in transitive root, i.e., kiru. This study therefore begins by looking at the diachronic shift of kiru. Table 8 shows the derivation period of the various categories. 'We may say that we can cut one end of the short thing.'

(MYS.5)
The substantive transitive function remained as the main function in every period.

Kiru in Early Middle Japanese
There are 54 tokens of kiru in Early Middle Japanese, 53 behaving as a transitive verb and one token as an aspectual verb.

Kiru as a Transitive Verb
Two forms of kiru, as a transitive verb, are observed: (a) as a single verb, e.g., (33); (b) as V1 in a compound verb, e.g., (34 In Kageroo Nikki (EMJ, 974), kiru forms a compound verb with another action verb, i.e., kudaku 'smash'. The two constituents render a substantive transitive function and syntactically weigh equally. There are 18 tokens in total of kiru's transitive verb use in compound verbs. Early Middle Japanese is a significant period in Japanese linguistics, as kana script was developed; the lexical integrity of the compounds becomes tighter. Each verb in multiple verb combinations is assigned to a coordinate or a successive relation and receive equal syntactic and semantic weight. Thus, the combined patterning may be a way of verb compounding.
This said, we cannot say that kiru merely bears substantive function. In the following compound verbs, it seems to render an aspect.

Kiru as an Aspectual Verb
In the work Genjimonogatari (Early Middle Japanese, 1008 A.D.), kiru fulfils an aspectual function. In the compound ori-kiru 'go down-completely', V1 oriru 'go down' denotes the MANNER of action and V2 conveys the ASPECT of the action, i.e., 'completely'. A similar formation is found in 這入り切る haihairikiru 'completely entre'.
Building on this, it seems sound to propose that the grammaticalisation of kiru started in the Kamakura Period, the beginning of Late Middle Japanese and becomes more solidified towards the end of Late Middle Japanese, i.e., the Aduchimomoyama Period.

Kiru in Early Modern Japanese
This section analyses Early Modern Japanese data, a period when kiru derived two new functions: adverb (four tokens) and noun (eight tokens). The transitive verb use is retained as the main function (48 tokens are confirmed).

Kiru Derives an Adverb (kiri)
Below is an illustration from 丹波与作待夜のこむろぶし, a work completed in 1707, where kiru displays an adverb function and is written as kiri ni. 'give it a go, seeing if it is an eight-copper coin or sixteen-coper coin.'

Kiru Derives a Noun (kiri)
Another intriguing finding is that kiru derives a noun function in Early Modern Japanese. The illustration comes from 冥途の飛脚 Meidonohikyaku (1711). 切り kiri is the continuative form of transitive verb kiru. Here, it modifies the degree of 模倣 mohoo 'imitate', meaning 'completely'.

kiru as an Aspectual Verb
In Taiyoo  Another striking finding in this period is that kiru begins to form a V-V with an unergative verb, i.e., 疲れ切る tsukare-kiru 'worn out-extremely'. The pattern of kiru's aspectualisation can be summarised as follows:

Stage I Early Middle Japanese
V-V kiri-kudaku, 'cut-crush', both V1 kiru and V2 kudaku are transitive. They render the MANNER of action and involve strong agentivity; no ASPECT is indicated. The two constituents are morphologically and syntactically equalled. The substantive function, i.e., 'to cut', is obtained in Early Middle Japanese.

Stage II Early Modern Japanese
The V-V omoi-kiru 'despair', and komari-kiru 'extremely distressed', the agentivity of kiru is reduced; kiru We cannot say that the above data is a matter of lexicalisation, as there is no syntactic reduction nor morphological alternation. Given this, it appears that the transitive root of kiru/kireru pair is not subject to lexicalisation. The following sections turn to the unaccusative verb kireru, seeing if it is subject to lexicalisation.

Kireru (cut unaccusative )
Having drawn a picture of kiru's evolution, this section analyses its unaccusative pair, kireru, by posing three question: (a) when it was derived from the transitive root (4.2.1); (b) how it derived other syntactic functions (4.2.2); (c) whether the process involved lexicalisation (4.2.3).
To begin with, Table 10 summarises the periods during which kireru derives other functions: The distribution of kireru in Late Middle Japanese as well as Early Modern Japanese remains similar to that in Early Middle Japanese: the majority has an unaccusative use and the rest has a noun use.

Kireru in Early Modern Japanese
In Early Modern Japanese, some data suggest a metaphorical reading of kireru. The aspectual function has been very extensively employed in this period. A total of 90 tokens are confirmed, cf.
(56). The reason of treating kireru as an aspectual verb rather than a suffix is due to the followings: kireru is not bound; the semantic meaning of kireru is not completely lost; it renderings RESULT of the action that is carried out by the first verb.
(56) Kireru as an ASPECTUAL verb やり切れる yarikireru 'carry out-throughout' (describe an action) 捨て切れる sutekireru 'give up-completely' (describe an action) 耐え切れる taekireru 'bear-throughout' (dscribe an action) 数え切れる kazoekireru 'caculable' (describe an action) 張り切れ harikireru 'burst' (describe an action) It ought to be noted that kireu can also be treated as the potential form of transitive verb kiru. In this regard, the original compound verb of (56) would be (57): (57) yarikiru; sutekiru; taekiru; kazoekiru, harikiru. The compound verbs in (57) differ from the compound verbs in (56) in that (56) describes an event that happens spontaneously. On the other hand, (57) involves an intention of carrying out an action thoroughly. The context of kireru in the corpus suggests a spontaneous event rather than an intended action. Therefore, kireru is treated as the unaccusative verb, functioning as an aspectual verb, rather than the potential form of transitive verb kiru. Building on this, we propose that unaccusative verbs can be ijel.ccsenet.org International Journal of English Linguistics Vol. 12, No. 3;2022 subject to grammaticalisation, rendering an aspect for the action verb (V1).

Kireru's Lexicalisation
Recall that lexicalisation regarding Japanese requires the following conditions.
(a) Syntactic reduction • VP [object + verb] has its accusative case particle omitted, • VP [subject + verb] has its nominal case particle ga lost The verb transits to a continuative form, conveying a noun function.
(c) Phonological alternation: the initial consonant of the non-initial portion of a compound or prefixed word becomes voiced  (Table 11), where, • Syntactically, the VP [subject + verb] lost the nominal case particle ga; • Morphologically, The verb transits to a continuative form; • Phonologically, the initial consonant of the non-initial portion of a compound or prefixed word becomes voiced. Among the 18 types (34 tokens) in Table 11, 14 types (26 tokens) are subject to sequential voicing (Note 23). The nominalised use of kireru in the above data undergo the following transitions: (a) The nominal case particle ga omitted, e.g.
検査が切れる→検査切れ; 燃料が切れる→燃料切れ; 電球が切れる→電球切れ, etc. ijel.ccsenet.org International Journal of English Linguistics Vol. 12, No. 3;2022 (b) The verb transits to noun form, e.g., kireru → kire (c) Though voicing occurs in the first consonant which is a criterion of grammaticalisation, 切れ cannot be treated as a suffix yet, because, most essentially, a suffix is written in kana rather than in Chinese characters.
With this in place, the evolution of kireru in Modern Japanese can be claimed to be a case of lexicalisation rather than affixation (i.e., grammaticalisation).

Summary
This section has tackled with kireru's decategorisation: how kireru is derived from its transitive pair kiru and, how it itself derived an aspect, a noun and a semi-suffix function. The finding suggests that kireru emerged in Early Middle Japanese, specifically in 951. In 1100, it derived a noun use. In Early Middle Japanese, the unaccusative use derived an aspectual function and the noun use began to render a metaphorical reading.
Kireru's lexicalisation appears to begin in Modern Japanese. The lengthy journey of grammaticalisation and lexicalisation is provided in Table 12. Comparing the evolutions of kireru and kiru, we arrive at the following: (I) kiru emerged in Old Japanese and kireru was derived in Early Middle Japanese; (II) The noun use of kireru was derived in Late middle Japanese and the noun use of kiru emerged in Early Modern Japanese; (III) The aspectual use of kireru emerged in Early Modern Japanese and the aspectual use of kiru is born in Late Middle Japanese; (IV) Kiru derived an adverb use in Early Modern Japanese and the adverb use is missing in kireru; (V) During Modern Japanese, kiru derived an apparent suffix function. Kireru got lexicalised.
The foregoing discussions are summarised in Table 13. Pulling these strands together, it seems that the degree of grammaticalisation in kiru is higher than that in kireru, as evidenced by the fact that kiru derived the aspectual function earlier than kireru; kiru derived the adverb use whilst kireru did not; kiru derived a suffix function in Modern Japanese whilst kireru underwent lexicalisation in the meantime. A proposal is thus put forward that: it is the transitive function that is likely to be subject to grammaticalisation and it is the intransitive function that tends to be subject to lexicalisation.

Conclusion
This study seeks an answer to the relationship between two evolution processes, grammaticalisation and lexicalisation, by examining data from Japanese transitive/unaccusative verb pairs' paths from a substantive verb to the various forms they fulfil in Modern Japanese, i.e., an aspectual verb, a noun, an adjective, an adverb, a quantifier and a suffix. A corpus-based investigation and a case study was carried out. This study further provided a working definition of 'grammaticalisation' and 'lexicalisation' that applies to Japanese. The findings brought us to the following main points: (a). Transitive verbs are more likely to convey ASPECT than unaccusatives do. The shift into a quantifier is limited to unaccusative verbs. Both transitives and unaccusatives appear to be capable of deriving a suffix function. However, the nominalised form by unaccusatives and transitives can both render a lexicalised meaning. A statistical analysis indicates that the nouns derived from an unaccusative verbs are likely to render a concrete reading; the nouns derived from a transitive verb tend to convey a metaphorical meaning. Furthermore, both unaccusatives and transitives are subject to nominalisation. Three ways are confirmed, namely, • N (dative case) + unaccusative verb, e.g., kawa-zoi 'rive side'; • N (nominative case) + unaccusative verb, e.g., gohan-taki 'rice cooking' • N (accusative case) + transitive verb, e.g., gohan-taki 'rice cooking' By tracing the evolution of how vit/vi develop into various categories in different periods, this study has pinned down the distinctions between grammaticalisation and lexicalisation, as shown in Table 14. Grammaticalisation (affixation) and lexicalisation in Japanese both require syntactic reduction and morphological alternation. The two differ in that lexicalisation does not require an alternation in writing, i.e., a lexicalised item can remain being written in Chinese characters whilst a grammaticalised item can only appear in kana script. Phonological alternation is obligatory in grammaticalisation but not required by lexicalisaiton. Lexicalisation appears to occur before grammaticalisation.