COHESION AND CONNECTION – JAPANESE DATA

This small study came to be as a by-product of an attempt at annotation of various texts, carried out as part of a larger research on Japanese connectors. In fact, the material used this time, prose fiction, yielded little or no connectors at all, reflecting perhaps a certain ideal of laconicism. But does that mean that, for lack of overt markers, the text doesn’t have to be somehow held together? This function, as it turns out, is extensively filled in by cohesion devices.

The short story “Run, Melos” by a Japanese classic Osamu Dazai (first published in 1940) is slightly over 10,000 characters in size [5]. The texts originally has only 4 paragraphs, so it took subdividing it into 24 episodes based on changing participants, location and timing of events described. Further, using the standard annotation techniques [9] it was broken down to 692 units, roughly representing elementary units of discourse (EDUs) in rhetorical structure theory [8], [2]. In examples below (minimum for reasons of space) the former are given in circled numbers and the latter in superscript. More often than not, however, ascribing a rhetorical relation to a pair of units was made difficult through a deficit of overt markers which are by far not necessary to express relations between EDUs [3].

Cohesion devices admittedly include topic markers, elision, lexical repetition and substitution, as well as certain means of deixis – notably, textual [6]. All of them appear loosely linked to certain connection types [7].

1. Topics in written Japanese are generally marked, or omitted [1]. The most prominent topic marker is WA, occurring all in all 257 times. Its role in the text crucially depends on whether it marks a new entity or the same one that precedes it.

In the first case, it primarily marks shift of the writer’s attention to a new subject within his field of view. At least 17 cases account for this, a prerequisite condition being that Topic-2 belongs to the same cognitive frame that Topic-1, e.g. ⑦ 1 KEKKONSHIKI ‘wedding’– 2 SHINROOSHINPU ‘bridegroom and bride’ + KAMI E NO SENSEI ‘oath to the gods’; ⑫ 5 KAWA ‘river’ – 6 SUIGENCHI ‘source’ – 17 NAGARE ‘flow’; in ⑬ the picture goes round in circles describing participants along a scenario of ‘river crossing’ 1 DAKURYUU ‘turbid stream’ – 3 NAMI ‘waves’ – 8 DAKURYUU ‘turbid stream’ – 11 NAGARE ‘flow’ – 13 NAMI ‘waves’ – 16 NAGARE ‘flow’ - 23 TAIGAN ‘opposite bank’ – 31 TOOGE ‘mountain pass’. All this ensures smooth transition within the scope of Conjunction (of the narrative subtype, see Narration or Continuation in SDRT). See also an effect of a camera zooming up in ② 11 BOOKUN DIONISU ‘tyrant Dyonisus’ – 13 OO NO KAO ‘king’s face’ – 14 BIKEN NO SHIWA ‘crease between brows’. In ⑯ 23 SHAYOO ‘declining sun’ – 25 NICHIBOTSU ‘sunset’ a parallel slide along the temporal frame takes place.

Antonymy is a necessary part of a frame (4 cases), see ⑪ 16 AME ‘rain’ – 17 HI ‘sun’; ⑨ HANAMUKO ‘groom’ - HANAYOME ‘bride’. Twice, awakening is described after sleep, fulfilling an anticipation activized. Here we find more of Contrast. ⑧ 23 WATASHI WA TSUKARETE SHIMATTA KARA, 24 CHOTTO GOMEN KOOMUTTE 25 NEMURITAI. 26 ME GA SAMETARA, … ‘23 I’m so tired, so 24 if you’ll excuse me, 25 I’d like to have some sleep. [now,] 26 When I awake…’ No verbal connector introduces 26, running counter to 25 exclusively on lexical antonymy. The same device is effectively used to tie together episodes ⑮ and ⑯ divided by a spell of sleep.

In the following example metonymic relations between rather loosely verbalized topics pre-determine a connection of Elaboration (also approachable as illocutory, or pragmatic [10], connection of Reason - reversed): ⑯ 30 WATASHI NO INOCHI NANZO WA, MONDAI DE WA NAI. 31 SHINDE O-WABI, NADO TO 32 KI NO II KOTO WA ITTE IRARENU. ‘30 My life isn’t an issue here. 31 Dying won’t leave me any possibility for excuse, 32 or any other good deed’.

A topic marker may signify a paragraph topic, like KEKKONSHIKI ‘wedding’ in ⑦ 1 where, still, it doesn’t  come out of the blue but is anticipation-driven.

A shift is justified merely by Contrast between topics, rooted in encyclopedic knowledge: ⑱ 19 AI TO SEI NO CHIKARA O, IMA KOSO SHIRASETE YARU GA YOI. 20 FUUTAI NANKA WA, DOO DEMO II. ‘19Now is the time to show the strength of love and faith. 20Of appearances I care not’  Cf. English ‘as for…’, Russian ‘chto do…’ in similar occurrences. A connector seems redundant as long as the Contrast relation between notions is self-evident, but it would be in place had there been need for clarification.

On the contrary, the omnipresent hyperframe of ‘Time’ yields an element of contrasting events and temporality just at any given time, see ⑱ 19 MEROSU WA, IMA WA, HOTONDO ZENRATAI DE ATTA. ‘By now [contrary to what was before] Melos was almost completely naked’. 

An element of Contrast is always present in described interaction between multiple participants, whose actions are marked when a topicality shift (28 cases in 16 episodes). Usually, this would involve circular motion that can include ‘outside’ factors such as time (especially in this piece that is tight on its chronologically set): ⑤ 4 MURA ‘village’ – 5 YOO ‘sun’ – 6 MURATACHI ‘village folks’ – 7 MEROSU ‘Melos’; ㉑ GUNSHUU ‘crowd’ – MEROSU ‘Melos’ – GUNSHUU ‘crowd’.

More than anything else, the connection of the Contrast type comes into play in marking of shifts of turns within a dialogue (or other such interaction) – all in all in 36 cases in 12 scenes. Basically, this represents one of the means of marking speakers in a dialogue. But obviously it cannot be used indefinitely – rather the author uses it in the frontmost part and then lets the exchange run automatically, reserving a reoccurrence of a marked topic for some special occasion. This can be a relevant comment, or a change in the nature of the exchange. See in ②: 3-6 ROOYA ‘old man’, 4 MEROSU – and from 8 and further the turns go uncommented, until 30, where a change in the mode of interaction takes place – Melos becomes infuriated, which exceeds just going another turn in the exchange. Same in ③ - Tyrant, topicalized first in 11, then in 31 where he heaves a sigh, 41 looking up, and in 62 where his voice becomes coarse. Also ⑤ when, running outside the framework of pure dialogue, Melos’s sister reddens in the face (20) or he himself walks out (26). Twice both communicants, topicalized, come to the conclusion of the dialogue.

Twice, a repeated topic marker occurs at the beginning of internal speech. 

Actually in some cases an actual occurrence of a verbalized, and marked, topic is a sayonara sign to the object of reference, rather than a promise to dwell on it for some time more. (end of a period) before another line of narration – f.e. Melos ‘starting to say’ before he breaks off 

Par contraste, of course, no change of topic ever takes place with a series of elisions.

2.  WA re-marking the same element has the opposite function, that is, dwelling on the topic. Generally it would reflect a rhythm of topical prominence gradually fading away after an introduction of a new element, so after a number of elisions some ambiguity would build up, increasing the need for the topic to be reasserted – ‘just to keep the fire going’, see ⑩2-11-14.

A particularly clear case of it is the beginning of an episode (10 times). MEROSU, as the macrotopic of the whole text, doesn’t actually have to occur in the first sentence – say, it is 5 in ⑮ or even 11 in ⑯. This provokes a light cognitive tension increasing towards the story’s climax, while closer to the denouement the narrative tempo builds up again and the gap shortens to only 2 in ⑱ and then slacking down a bit to 4-5 in ⑳ + 2 in ㉔.

(end of period = excourse) 1

However, just this ‘thirst’ for clarity isn’t by far the only, nor even the most outstanding pattern of topic repetition. More often a repeated topic would come about with some progress in the state of affairs. It made mark a new thread of thoughts, thus ending a series of Elaborations and a shift to (illocutory) Conjunction (see this supported by a Conjunction marker SOOSHITE in ⑪1-6) or even Cause (7 occurrences). In part this, no doubt, is required by the very idea of dynamical narration.

⑧ 3 MEROSU WA, 4 ISSHOO KONO MAMA KOKO NI ITAI, TO OMOTTA. 5 KONO YOI HITOTACHI TO SHOUGAI KURASHITE IKITAI 6 TO NEGATTA GA, 7 IMA WA, JIBUN NO KARADA DE, 8 JIBUN NO MONO DE WA NAI. 9 MAMA NARANU KOTO DE ARU. 10 MEROSU WA, WAGAMI NI MUCHIUCHI, 11TSUINI SHUPPATSU O KETSUI SHITA.

Here a repeated topic marks arrival at a conclusion, also signifying end of a period (both semes supported by temporal TSUINI = at last), and a Reason type of connection.

A rather transitional case is the same entity re-marked with WA but repeated with a different case particle (19 occurrences). This reflects a change in the approach to the state of affairs, viewed at a different angle of a situation, typically representing Elaboration – the same idea is enlarged upon, bringing out new sides in it and creating new meanings, see ① 1-3(+case)-4-7-11(+case):

① 1 MEROSU WA GEKIDO-SHITA. 2 KANARAZU, KANO JACHI BOOKUN NO OO O NOZOKANAKEREBA NARANU TO KETSUI-SHITA. 3 MEROSU NI WA SEIJI GA WAKARANU. 4 MEROSU WA, MURA NO MAKITO DE ARU… ‘1Melos was infuriated. 2 [He] decided that he would by any means depose the tyrant king. 3Politics is uknown to Melos. 4 Melos is a shepherd in a village…’

Not in all cases topic marker “survives” necessary transformations. For instance, in cannot cooccur with genitive. But evidently, a succession of this kind is typically in the Mathesius’ style, and plays the same role (9 cases with MEROSU+GEN).

A wholly different story is the use of marked topic in delimiting boundaries of inner speech. Here it is the first person pronoun that is being marked at the beginning and end. See

Now, in ⑪ 9 a shift from WATASHI to MEROSU signals a jump from inner speech back to the objective description.

Furthermore, a subtler differentiation is being effectuated even within the flow of consciousness of the protagonist, where articulated sentences, topic-marked, are ‘diluted’, ‘filled in-between’ with not quite ‘objective’ but ‘sensed’, ‘automatic’ speech (9 cases in 4 scenes). This establishes implied connections of (illocutory) Cause:

⑰ 12 WATASHI WA, SEIGI NO SHI TOSHITE SHINU KOTO GA DEKIRU ZO. 13 AA, HI GA SHIZUMU. 14 ZUNZUN SHIZUMU. 15 MATTE KURE, ZEUSU YO. 16 WATASHI WA UMARETA TOKI KARA 17 SHOJIKINA OTOKO DE ATTA. ‘12 I am capable of dying for the cause of fairness. 13 Oh the sun is setting. 14 Setting faster and faster. 15 Please wait, oh Zeus! 16 Since birth 17 I have been an honest man.’

Rather than picking up a slackening reference, topics are here to mark expressions of decision, determination, action after more abstract reasoning in ⑲ 44 WATASHI WA NANDAKA 45 MOTTO OSOROSHIKU OOKII MONO NO TAME NI HASHITTE IRU NO DA. ‘44I am running 45for the sake of something bigger and more important’.

Even inside the spoken speech, where reference is most transparent, by overdoing the ‘ego’ (quite unlikely in everyday spoken Japanese) the protagonist Such is his last speech to Selintius: ㉒ 4 WATASHI O NAGURE… 6 WATASHI WA, TOCHUU DE ICHIDO, WARUI YUME O MITA… 8 WATASHI WA… ‘4Hit me… 6I say a bad dream on my way once… 8I…’

Events of ‘conscious existence’, marked in such a way thoughout the story, succeed in making the passages in the novel all the more striking as a manifest of human will.

3. Other topicalizing particles include DATTE (1 occurrence) and MO 13, which marks Contrast and Conjunction, or most often a syncretic combination of the two, e.g. ⑮ 53 KIMI WA, ITSUMO WATASHI O SHINJITA. 54 WATASHI MO KIMI O, AZAMUKANAKATTA ’53 You always trusted in me. 54 I also never betrayed you.’

Here is an example of a non-linear connection marked by MO. In this Conjunction, rather than just mechanically adding Melos to the number of believers (which would be absurd), it stands for the “Me not lying to you” added to the general count of the “right” events depicting a harmonious relationship. To account for this, Contrast would be needed.

MO can well link EDUs at a considerable distance, like AME ‘rain’ in ⑩ 12 after ⑧ 2 and then ⑪16, creating a wide-stretched web of associations particularly valuable for creating a holistic image of the text.

4. Chief anaphoric markers include SO- (14 occurrences including 3 SONNA with negation) and A-.

⑯ 33 WATASHI WA, SHINRAI NI MUKUINAKEREBA NARANU. 34 IMA WA TADA SONO HITOKOTO DA. ‘33I must gratify this trust. 34 Just this is needed now.’

Here the need for the following sentence originally arises out of a desire to offer more detail without derouting from the principal line of narration – that is, not shifting the topic. This typically coincides with the Level-of-Detail type of Expansion connection.

SO-words widely amalgamate to actual connectors, mostly temporal - SONO YORU in ⑤ ‘on the evening of that day’ (distant sequence) or SONO KORO ⑧ ‘at that period’ - same (approximate simultaneity). At first sight redundant, SO- strengthens and accentuates the connection otherwise more obtuse as with Cause in ⑲ 39 SOREDAKARA, HASHIRU NO DA. ‘This is why I run’

Normally SO- constitutes illocutory Conjunction (Narration in SDRT), e.g. ㉑ 13 MEROSU WA SORE O MOKUGEKI SHITE… ‘Melos saw that and…’

SO- is sharply opposed to KO- used by heroes in ‘their’ world in dialogued or inner world. On the other hand, A- (7 occurrences) as marker of mental space has a peculiar way of resting between the text and the event level, since the reader’s and the acting heroes’ mental spaces partially coincide. Occurring after a significant lapse it is aimed precisely to strengthen that impression, as in the beginning of ⑰ 3 ANO AKUMA NO SASAYAKI WA, ARE WA YUME DA ‘Those whisperings of the devil, that was a bad dream’ (the antecedent was 35 ECDs back, in the end of a pre-preceding episode. Same is intended in an overheard conversation in ⑱ concerning Melos himself – to show people having distant idea of the protagonist: ⑫ 24 TAIYOO MO SUDENI MAHIRUDOKI DESU. 25 ARE GA SHIZUNDE SHIMAWANU UCHI NI… ‘24 The sun is already in zenith. 25 He must have [come] before it goes down…’

5. Elision (zero realization) is posited at sentence level, so it has to be borne in mind that there are only 473 sentences in the whole text. Mostly it is used in monotonous topical surroundings to dwells on a compact zone of thought or description, without drastic developments [1] – which suits the type of rhetorical connection called Elaboration. More dynamical progress (causal or narrative), as well as the topic setting, is done my marked topics, as was shown above. Here is an example: ⑮ 89 WATASHI WA, SHINU YORI TSURAI. 90 WATASHI WA, EIEN NI URAGIRIMONO DA. 91 [0] CHIJOO DE MOTTOMO, FUMEIYO NO JINSHU DA. ‘89 I feel worse than dead. 90 I am for good a traitor. 91 [0] A most inglorious person on earth’ WA-Topic occurs when a new thesis comes in (connection 89-90 accountable as both Elaboration and illocutive Cause), ellipsis – when it is being explained and elaborated on (connection 89-90).

Predominant objects of elision are, consecutively, the main protagonists, holding prolonged attention of the author. ‘Melos’ is omitted all in all in 51 sentences (though marked as topic/subject it occurs 78 times), showing a staunch resilience in reoccurring even after strings of ellipsis for different nouns (e.g. without re-marking the topic after 2 sentences centered on his sister’s fiancé in ⑥). Tyrant/king (BOOKUN/OO), otherwise topic-marked on 8 occasions, goes unnamed 5 times as subject. Selentius and sister follow at 2, her fiancé occurs so once. Inanimate objects in descriptions are omitted only 3 times.

Elision is frequent in spoken speech of the heroes, mostly for first person singular pronoun (28 with Melos, 7 for the king, once for the fiancé and for the bandit. This is a natural trend in spoken Japanese in general, strings of marked ‘I’-subjects, on the contrary, conveying a peculiar mental tension (fast switch of important landmarks for attention) and a great deal of dramatic effect. Curiously, ‘You’ pronoun (OMAE), at 20 topic-marked occurrences throughout the text rather tending to be pronounced (except, of course, for imperatives), is omitted for the most part in contexts involving speaker’s power over his counterpart – say, 6 times for the king talking to Melos (none back), 7 times Melos addressing his sister (mute). Spoken dialogue is therefore a mesh of omissions, see the words of Tyrant: ③ 101 [OMAE WA] INOCHI GA DAIJI DATTARA, 102 OKURETE KOI. 103 OMAE NO KOKORO WA, [ORE WA] WAKATTE IRU ZO ‘101 If [your] life is dear [to you], 102 do come late. 103 [I] can very well read your heart.

‘He’ is not verbalized on 4 occasions, ‘we’ on 2.

Internal speech by Melos includes 19 more omissions of ‘I’ pronoun, thus in his case alone already exceeding the general number of topic-marked WATASHI in the text (43). However, in some cases defining subjects isn’t easy in Japanese for reason of rather blurred boundaries between indirect speech and ‘objective’ description, as with psychological predicates such as ④ 18 SUKOSHI DEMO NAGAKU KONO IE NI GUZUGUZU-TO TODOMATTE ITAKATTA. ‘It would be so nice to lag around for a while longer in this house’.

Omitted are also a macro-topic (see 31, 38 and 41 in ①, and even 1 and 2 in ②!) after MEROSU occurs in 28 and 34 ①), as well as the topic across turns in an exchange (9, 10, 11, 12, 28 after OO in 8 in ②, with just one reoccurrence 13). Generally, the factor of empathy seems to play a significant role, see the ending of ⑥: 13 MEROSU WA, 14 [ORE WA] MATSU KOTO HA DEKINU, 15 DOOKA ASHITA NI SHITE KURE TAMAE, 16 TO SARA-NI OSHITE TANONDA. 17 MUKO NO MAKITO MO GANKYOO DE ATTA. 18 [MUKO WA] NAKANAKA SHOODAKU SHITE KURENAI. 19 [FUTARI WA]  YOAKE MADE GIRON O TSUZUKETE, 20 [MEROSU WA] YATTO, DOONIKA MUKO O NADAME, SUKASHITE, TOKIFUSETA. ‘13 Melos said, 14 [I] can’t wait, 15 please find a way to make it tomorrow, - 16 pushing his request further. 17 The shepherd fiancé was stubborn. 18 [He] just wouldn’t yield. 19 [They] argued till dawn, 20 when at last [Melos] managed to cajole, reassure and concinve the fiancé.’ Here a decisive return to the perspective of Melos’s (to abandon a luring prospect of representing 18 as Melos’s point of view – given the speaker-oriented adverb NAKANAKA [4]) is reaffirmed in both his zero realization of his name and use of relational nomer for his counterpart in 20. Japanese writing shows here quite a way more empathic than pronoun-ridden Western. 

Empathy lurks behind yet another curious use of elision occurs in the beginning of an episode – with the aim to stress the continuity of ‘inner persective’ of the protagonist – whereupon it is ‘objectivized’ to a ‘third-person’ position in naming him outright:  ⑥1 ME GA SAMETA NO WA YORU DATTA. 2 MEROSU WA OKITE SUGU… ‘1 [He] awoke at night. 2 Melos stood up and… No doubt, this, strengthening the narrative Conjunction connection, assures a more intimate contact of the reader with the protagonist. 

Lastly, a purely literary trick can be played on, it seems, an interlinguistic scale with the order of naming and omitting reversed: ⑱ 24 MIERU. 25 HARUKA MUKOO NI CHIISAKU, 26 SHIRAKUSU NO SHI NO TOOROO GA MIERU. ‘24 [It] appears. 25 Far in the distance and tiny, the towers of the city of Siracuse appear’. Here, mimicking the process of recognition “before words”, a reversed elision nevertheless support Elaboration.

Overall number of elisions is impressive, making the piece, reportedly adopted from Schiller, an authentic masterpiece of original Japanese writing.

6. Pronouns are apparently and world-deictic rather than text-deictic, such as, in dialogues, ANOKATA ‘he’ (3) and ARE ‘it’ (3) including, unususally, substituting Selinuntius, a person, ② 75. Anaphoric are KARE ‘he’ (4) and FUTARI ‘the two of them/they’ (3). Non-textual deixis outside of exchanges includes KO- in internal speech: ④ 18 SUKOSHI DEMO NAGAKU KONO IE NI GUZUGUZU-TO TODOMATTE ITAKATTA. ‘It would be so nice to lag around for a while longer in this house’. Here the centre of empathy is Melos, and his statements follow a track of Elaboration before another jump in narration.

7. AGReement, as it arguably exists in Japanese, comes as a partial recompense for ellipsis. Here its use is twofold, firstly in 4 honorifics pertaining to the ‘king’ in third person (GOJISHIN ‘he himself’, GOMEIREI ‘his orders’) in ②. Here it functions exactly as elision, ensuring a monotonous development in Elaboration. A more dramatic connection takes place in ⑱ 20 O-URAMI MOOSHIAGEMASU ‘I hate you’, where, however, the cohesion device is hardly of any value to connections.

8. Lexical repetition is observed all in all in 71 cases, which makes it a rather dispreferred means of ensuring topical continuity unless motivated. Not unlike Russian (where it lexical repetition normally is obligatorily replaced by pronouns), one such motivation lies in questioning either the concept, or the code. For instance, HITO NO KOKORO ‘heart of a man’ in the king’s reasoning occurs in ③ 24 and again in 28, as long as he dwells on its nature. In ⑮ 16 Melos is called YUUJA ‘a brave one’, and in 29 it reoccurs in YUUJANI FUTSURIAI-NA ‘unworthy of a brave one’. Such rhetoric move, where the author or the protagonist picks up the term and scrutinizes it, naturally comprises Generalization and Contrast [7].

A further development of this semantics is achieved in Cause connection, with an element of ‘suspended factivity’ (= ‘if you admit that presumption A is true, than B, based on A): ⑰ 16 WATASHI WA UMARETA TOKI KARA 17 SHOJIKI-NA OTOKO DE ATTA. 18 SHOJIKI-NA OTOKO NO MAMA NI SHITE SHINASETE KUDASAI. ‘16 From the time of my birth 17 I was an honest man. 18 Let me die as an honest man’. This is an example of the mechanism behind illocutory Causality.

Similar connection belongs to the Q-A type (here, cognitive stimulus – cognitive response) in ⑯ 1 MIZU NO NAGARERU OTO GA SHITA… 4 MIZU GA NAGARERU RASHII ‘1 There was a sound of running water… 4 Apparently there is running water’ – here not an element of the code, but a piece of extralinguistic reality is tried out and reconfirmed.

Questioning code or concept backsides with a shade of self-reasoning in ⑰ 3 ARE WA YUME DA. 4 WARUI YUME DA. ‘3It is [only] a dream. 4A bad dream’, a modifier added working for Elaboration (then it reverts in a set in ㉒ 6). This approaches another forceful motivation for lexical repetition, when it becomes a device to make the speech sound heavier and more convincing. On 4 occasions chunks are repeated in completely identical setting – to create rhythm, setting stronger rhetorical impact: ⑮ 15 KOKO MADE TOPPA SHITE KITA MEROSU YO. 16 SHIN-NO JUUJA, MEROSU YO. ‘15 You’ve made it to this point, o Melos. 16 You are a real brave one, o Melos’.

Repetition isn’t limited to noun groups: ⑲ 39 SOREDAKARA HASHIRU NO DA. 40 SHINJIRARETE IRU KARA HASHIRU NO DA. …44 WATASHI WA, NANDAKA… OSOROSHIKU OOKII MONO NO TAME NI HASHITTE IRU NO DA. ‘39 That is why I run. 40 I am trusted – this is why I run. 44 I’m running for something tremendously great’. Same goes for NAGURE ‘strike’ in an example below, repetition producing an effect of incantation that adds rhetoric strength.

Above I already touched upon the repetition of the marked topic: ⑮ 31 WATASHI WA, KORE HODO DORYOKU SHITA NO DA… 34 WATASHI WA SEIIPAI-NI TSUTOMETE KITA NO DA… 36 WATASHI WA FUSHIN NO TO DE WA NAI. ‘31 I made such an effort… 34 I did my utmost… 36 I am not someone to distrust’. Here the repetition pattern grows complex. Besides cyclic reoccurrence of first person pronoun, quite uncharacteristic for ‘inner thought’ but very much so – for a vehement argument, counting on an audience, - we see words plucked from previous deliberations, notably DORYOKU ‘effort’, first a lexical substitution between these very 31 and 34, and FUSHIN ‘mistrust’ used in negation 16 EDUs before, thus by turn representing connection of Conrasting the new reality to the previous, with a more global pattern of Elaberation or Equivalence in the development of the argument. But all this strength is used – or, better said, wasted, on a immanently lost battle with oneself, being an attempt to strengthen an already lost, forceless argument! Here is an obvious example of why and how rhetoric devices are used in communication, en faute of more plausible means.

A similar example is the outbreak of ⑰ where a whole phrase is doubled, conveying an impression of “autohypnosis” – 1 WATASHI WA SHINRAI-SARETE IRU. 2 WATASHI WA SHINRAI-SARETE IRU. ‘1 I am trusted. 2 I am trusted’. Unprecedented in the otherwise very compact and laconic text, this produces an intense sense of concentration and resolve. Later on, WATASHI (WA) is brought up again 5 more times, in a more and more rapid succession (42-44-45-46-47), playing up to a climax in the novel’s plot.

Overly repetitive use of pronouns rare in spoken speech adds a lot to the horrendous tension of the final scene where almost nothing but the bare pronouns exchange is left, reminding the reader of an ascetic ancient drama: ㉒ 4 WATASHI O NAGURE. 6 WATASHI WA… 7 KIMI GA MOSHI WATASHI O NAGUTTE KURANAKATTARA, 8 WATASHI WA KIMI TO HOOYOO SURU SHIKAKU SAE NAI NODA. ‘4Strike me. 6I was… 7If you do not strike me, 8I will not be worth even embracing you’. Then later it is reverted in the response of his friend.

A change in case with a repetition is similar to the already described topicalized structure (14 occasions): ⑩ 11 MEROSU WA YUUYUUTO MIJITAKU O HAJIMETA. 12 AME MO, IKUBUN KOBURI NI NATTE IRU YOOSU DE ARU. 13 MIJITAKU WA DEKITA. ‘11 Melos slowly started his preparations. 12 The rain, too, was lightening a little. 13 The preparations were over.’ Later the word doesn’t come up again, so the Mathesius-style comment-topic treadmill is over. This marks a shift in Elaboration.

Another motivated use of complete repetition is reactivating the concept after a significant lapse. The further the distance with the antecedent the more important it is to offer an exact repetition, not infrequently including modifiers, e.g. ⑮ 55 YOI TOMO ‘good friend’, used previously in ⑫ 27, with a parallel slight modification of YOKI TOMO in ㉔ 3 after ④ 2. See also MIGAWARI NO OTOKO ‘the hostage man’ in ⑪ 3 after it occurred in ③ 86, 90, 93, 96. This strengthens the argument, establishing a long-ranging connection of Equivalence between the two situations. Bridging time spans are word-for-word repetitions of DAKURYUU ‘turbid flow’ and SANZOKU ‘bandits’ in a synopsis of past events in ㉑. Generally, repetitions across text serve as pivotal points for its message, the most prominent being the concept of ‘trust’ (that the story indeed evolves around), rendered in SHINJIRU/ FUSHIN/ SHINJITSU/ SHINRAI in ②23 ③30 71 88 ⑩9 ⑮18 20 36 60 62 65 94 100 ⑯29 33 ⑰1 2 ⑲ 32 38 40 and ㉓8. Another example is the ‘sun’ TAIYOO/HI (17), creating a continuous dramatic chronological counterpoint to the plot. Slight differentiation observable here is due to a comparatively loose notion of Sinographic stems in Japanese.

8. Lexical substitution is comparatively rare at 14 occasions. As Japanese has no articles and no plural, in theory an unmarked noun would refer either to the entire multitude of generic objects, or, with an antecedent, to a distinct single entity (or entities). Therefore, there is too little tying two alternative expressions together, if the lexical material is changed. What we see are examples of metonymy as in NINGEN ‘a human being’ in ③ 29 for HITO ‘man’ in 24, 28, which marks Generalization and possible a Cause connection. Another line of change is shift in empathy, as with ANI ‘older brother’ in ⑤ 9, 10, 15 instead of MEROSU in 7, when the talk is re-centred in his sister’s perspective. Actually, in ⑧ Melos uses it himself talking to his sister, instead of personal pronoun – this phenomenon is widely known in baby talk. Again, in ⑨ a shift from IMOOTO ‘little sister’ to HANAYOME ‘bride’ marks a move from Melos’s perspective to an objectivated narration.

A speaker tends to substitute an element of the code when he becomes unsure of the effect of his words. Thus, whilt repetition combined with an epithet added yields an Elaboration or Level-of-Detail (2 occasions), in substitution it is rather the connection of Equivalence that must be taking place, at least if the added material is of little or no value informationwise. Between 17 and 18, if not 16 and 17, is Paraphase at its pure: ㉑ 16 WATASHI DA, KEIRI! 17 KOROSARERU NO WA, WATASHI DA: 18 MEROSU DA. ‘16 It is me, hangman! 17 It is me who must be killed. 18 Me, Melos’. Similar case of Equivalence is the change of Included in the above is a quoted case in  ISSHOO ‘[my] life’ to SHOOGAI ‘entire life’ in ⑧ 3 - 5 when it makes repetition of a statement weighty.

Also, if repetition serves to focus on an element of the code or a concept, and add tension, lexical substitution aims at the opposite effect. E.g., when no doubt arises in the truthfulness of the code or reality in ⑯ 9– 10 with SHIMIZU ‘source’ to IZUMI ‘fountain’. More elegant still is the shift from MEROSU (㉔7) to YUUJA in ㉔8 – which not only lessens the tension and intensivity at the same time, but reestablishing a semantic loop comprising the entire story after ⑰9 and ⑮29 as well. Here YUUJA WA, HIDOKU SEKIMEN-SHITA ‘The brave one was greatly confused’ even an element Contrast can be detected, as the hero is discovered naked in public. Besides a touch of irony is daftly in place – the finale off the short novel veering a little off towards the universal.

Let us conclude by saying that cohesion in a multi-layered text [11] helps establish a clear structure, delimiting descriptions, interaction of participants and inner speech of the protagonist with its various nuances. In this sense, it helps create a multistorey, multifaceted texture, including connections between adjacent EDUs and longer spans, both within and outside of dialogue. Not that it substitutes the structure rhetoric connections, nor that all elements of it, however, are of equal value to the latter. Some of these patterns of interaction between cohesion and connection are more universal (like word-for-word repetition drawing attention to an element of the code or establishing remote equivalences, or substitution-(by-definition-)as-Equivalence), some less (as the use pronouns). That the topic marker WA has a great potential for marking connections of Contrast, may not be such big news in itself. Using WA helps establish a clearer recognition of this connection similar to that expressed in Russian connective A. Minimal marking of it in Japanese doesn’t make it inexistent, cf ‘full-fledged’ markers in English like ‘now’, ‘as for…’. What may come as a surprise is how deeply this type of connection permeates a prepared speech of any substantial length. For a significant part of it, the author is changing the angle of his vision, weighing different part and bits of reality against each other. In a certain way, this comprises close but non-identical theses that connect through Cause or Elaboration types. So Contrast, which by far not always invokes Contradiction, or Betrayed Expectations, appears as a rather primary and widespread type of logical connection – or mental operation, should we proceed from the Speaker’s standpoint.