Regulated verse and semiotic structure: a poem by Li Shangyin (李商隠)

2.1.1 Two matrices

In my expanded theory, the above rules become more complex in that every modern text has not one but two paradigmatic sets of images generated by two matricial propositions, linked together by a syntagmatic relation, which may be of several types: negation, difference of scale, etc.

In the course of the chapter devoted to Riffaterre in The pursuit of signs (Culler 1981), Jonathan Culler makes a suggestion that is very relevant to my argument in favor of this type of binarism. Observing that the Riffaterrian principle of unity requires that all the images of a poem be “variants of the same structural matrix” (R.1978: 82), Culler doubts that the semiotic structure of a poetic text is necessarily as simple as that (Culler 1981: 92–93). He prefers to say that “in many cases readings unify a poem with the aid of some elementary binary opposition,” citing as an example Rimbaud’s Fêtes de la faim (Culler 1981: 93). This “binary opposition” is evidently suggested to Culler by the analysis which Riffaterre himself initially proposes of Rimbaud’s text into two sections or expansions, each generated, I would say, by a different matrix: “eating the inedible, eating the edible” (R.1978: 78).

It proves to be the case that bimatriciality, far from being the exception, is apparently a very powerful rule. Our 4-line example of Tang Chinese “regulated verse” [lü-shi] also happens to have bimatricial structure. The two matricial phrases will be called “MP1” and “MP2” (the latter being the more innovative, and more difficult to identify) according to my convention in Hopkins 1994. I approach Eliot’s long poem from the same “bimatricial” perspective (see Hopkins 2016).

2.1.2 Intertext and sociolectic context

Even acknowledging the notion that the intertext detailed in Section 1.2 may constitute a sort of “literary context” – in its role as structural model – it is an important part of my theory that a text also possesses a context in the narrow sense mentioned above: i.e. a proposition homologous with the matricial structure of the text on the level of vocabulary. In making this proposal, I am assuming that poetry communicates, not only in relation to other texts, but also in relation to that sociolectic background – or sociolectic context (SC) – which the reader brings to it. For example, if the primary text compares the power of human intelligence unfavorably to the power of Nature, the sociolectic context of this message includes these same lexical elements, which, however, are not related in the same fashion as in the primary text. Thus, we have similarity on the lexical level and contrast on the structural level.

My prime reason for proposing the existence of a sociolectic context is the notion that, for the reader to be able to adequately interpret the semiotic structure of the modernist literary work, he must first set it against the familiar structures of the semiotic framework of the sociolect within which he normally communicates.

2.1.3 Interpretant and perceptual change

In the chapter of La structure du texte artistique titled “Le texte et les structures artistiques extra-textuelles,” the Russian semiotician Juri Lotman begins by stating that the interpretation of the work as an artistic text is possible only after the difference between artistic and non-artistic texts has been recognized (Lotman 1973, [1970]: 390–406). This recalls Riffaterre’s own emphasis on the impossibility of interpreting an artistic text according to the rules of sociolectic discourse. Unlike Riffaterre, however, Lotman takes the interpretive process beyond the text: he points out that, once the text has been decoded according to the rules of the literary idiolect, the reader’s previous world-view (“le vieux modèle du monde”) is swept away by the “nouveau modèle plus complet du réel” (the “new, more complete, model of the real”) offered by the text. Lotman calls this process “modélisation,” which I translate as the “remodeling” power of the literary work. I should add that this power is characteristic of modernist texts.

The interpretant of the textual matricial structure (TMS) is abduced by the reader from a comparison of TMS and intertext. This interpretant will contain enough of the TMS to present to the reader a new view of the world. Subsequently, the reader projects this new world-view “not only onto the structure of his artistic experience, but also onto the structure of his experience of life,” i.e., onto his sociolectic preconceptions (Lotman 1973 [1970]: 392, 396). The result will be a change in those preconceptions.

Umberto Eco, also, is very explicit about the power of “aesthetic texts [to] modify our concrete approach to states of the world” (Eco 1976: 275). Among literary semioticians apart from Lotman, Eco is the most explicit concerning how the relations between text and sociolect might be accounted for, without doing violence (as Barthes does, for example) to the notion of the text as an impermeable – or “monumental” – semiotic structure. Eco does not deny the importance of establishing rules governing the operation of the internal signifying system of the literary text. In his terminology, one would speak of an intensional semantics, which, he insists, should be considered a prerequisite for any extensional semantics, i.e., one that would deal with the relations between text and sociolect, which “[…] controls the correspondence between a sign-function and a given state of the world” (Eco 1979: 179).

3.2.1 Formal parallels and semantic structure

1. Let us look first at the rhyme-scheme: the end-rhyme of in/in in the reconstructed Tang pronunciation of lines 1, 2, and 4. Even though pronunciation has changed over time, the modern reader knows to expect rhymes in these lines (incidentally, Tang pronunciation tends to be preserved in modern Japanese). If rhyme indicates some sort of similarity on the semantic level, then the “deep” and “sunken” at the ends of lines 1 and 2 may share semes with the state of the heart at the end of line 4, i.e., the hearts of both protagonists would be “sunken” or downcast. In isolation this would be perfectly valid, and it does not contradict the larger metaphor of the whole text, in which Chang-O’s situation is juxtaposed with the situation in lines 1 and 2.

2. In addition to rhyme, the last three characters of lines 1 and 2 also have more or less parallel syntax (adj.-noun-past part.). The shared idea here is that “the night is almost over”; i.e. the “depth” of the candle’s reflection indicates the lateness of the hour.

3. (a) In addition to rhyming end-words, lines 1 and 4 have the same tone pattern, as well as basically similar syntax. There are, however, certain syntactic differences, which are detailed in 3.2.2 just below. Jakobson might go so far as to ignore these and, relying on the tone pattern plus rhyme, set up a semantic parallel of A’ is to B’ as A is to B (“semiotic square”) configuration, which implies a semantic relation between the mica screen and the sea and sky, on one hand, and the depth of the candle-reflection and the heart, on the other (cf. Greimas 1970: 138). This would be an attractive conceit. (b) There would also be some justification for a syntactic parallel in the last three characters of each line, where “heart” could function semi-adjectivally. Tone pattern and rhyme are parallel. The common result would be to impute a connection between the moon spirit’s heartache (minus the reason for her regrets) and the human lady’s lonely vigil. The regrets of line 3 are key to the relation between matrices.

4. Lines 2 and 3 both begin with two words of very similar pronunciation (including identical tone pattern). May we conclude that the “Long River” (the Milky Way) thereby takes on some of Chang-O’s attributes? In any case, this is a nice example of phono-lexical parallelism: both expressions involve the night sky.

5. The above parallel may be thought to extend to the first four characters of lines 2 and 3 on the grounds of identical tone pattern. The syntax of these two groups is almost parallel: noun phrase-adv.-verb/proper noun-aux.-verb. The semantic parallel involved, if it were allowable, would have the effect of gaining inevitability for Chang-O’s regrets from the inexorable sinking of the Milky Way. But closer examination reveals that the verb is intransitive in line 2 and transitive in line 3, where we are really dealing with a verb phrase which extends to the end of line 3 (“must regret stealing spirit drug”). Thus, syntax gainsays the parallel suggested by tone.

6. The grouping of three characters involving “sinking,” etc., in lines 1 and 2 is a possibility: these have the same word-class, and also the same “water” radical, and this points the way – if pointer be needed – to their semantic common ground.

7. Within line 2, we have a strict parallel between two adj.-noun-v.i. groups (the adverb “gradually” being a minor anomaly): “Long River [gradually] descends/morning star sinks.” On its own, the semantic correspondence thus identified is fairly simple, but the syntactic parallel is so clear that it definitely attracts the reader’s attention to the location of an image: “our lady’s heart sinks with the morning stars.”

8. Parallel no. 7 (l. 2) may be extended on grounds of syntax (similar adj.-noun groups plus a verb: “deep” as a past participle would be a quasi-verbal) plus the “graphic cohesion” of parallel no. 6, to include the last three characters of line 1. This would be semantically more interesting, as it could be taken to impute a connection between the movement of heavenly bodies and the human situation: their downward – i.e. negatively marked – movement, and their inevitability.

3.2.2 Problems and anomalies

It is important to note that matricial structure is larger in scale than the formal parallels of regulated verse.

The Jakobsonian approach of collecting the maximum number of formal parallels turns up some interesting isolated semantic correspondences. But some of these seem to conflict with each other, and none extends through the whole text. The way prosodic categories are distributed shows that they are not designed to delineate a common text-wide semiotic structure, nor a pair of structures like the two matrices detailed in 3.3. Rhyme (parallel no. 1) links the final character of lines 1, 2, and 4, whereas tone pattern links the first four characters of lines 2 and 3 in one instance and lines 1 and 4 in the other. These three groupings between them conflict with the intuition of two matrices involving both lines 1–2 and 3–4 respectively.

The tone plus syntactic parallel of lines 1 and 4 (no. 3) comes the closest in semantic effect to this latter intuition. However, there remains the fact that there are clashes between syntactic and other categories. We have noted above the conflict between tone pattern, line position, and rhyme on one hand, and syntax on the other, in respect of “deep,” “heart,” and “candle-shadow” (3.2.1, 3b). As for the apparent parallel between the “mica screen” of line 1 and the “sea” and “sky” of line 4, at first sight we have two successive noun phrases – but in fact “mica” modifies “screen,” whereas “sea” does not modify “sky.” It is doubtful that we could avoid this problem of syntax by simply giving the tonal parallel precedence.

These formal-level contradictions are not so important if we allow formal parallels to function heuristically, and in respect of semantic parallels in the first and second parts of each line, rather than the two lines as units. In other words, without adhering to any of the formal parallels too strictly, once these have attracted the reader’s attention, he notices the possibility that line 4 is a macro-scale representation of the human situation in line 1: Chang-O’s nightly heartache might be seen as “reflected” – like the candle in the screen of line 1 – against the backdrop of sea and sky, a screen on a huge scale.

But although this somewhat fanciful parallel of lines 1 and 4 would not contradict our intuition of a matricial structure, neither can it account for the vital element of the regrets imputed to Chang-O, which need to be factored into the moon spirit’s “heartbreak” if it is to be juxtaposed with the “deeply” embedded candle-shadow. As we shall see, the only structure which can act as a model for this larger parallel – which extends throughout the text – is intertextual. To repeat, only a parallel extending across lines 1 and 2 on the one hand and 3 and 4 on the other could explain the text-wide metaphor involving the important point of Chang-O’s regrets. None of the above formal parallels – whether based on rhyme, syntax, or tone – can be said to do this. The only kind of parallel that could encompass the whole text would be syntactic: either all four lines would have to share the same syntactic pattern, or there could be two couplets of parallel syntax. The second of these two cases is not inconceivable, but it does not obtain here.

If such a global formal parallel did exist, that would tend to mean that everything in the poem is generated by a single matrix, i.e., an A’ is to B’ as A is to B (“semiotic square”) structure would tend to point to the existence of a unimatricial underlying structure, since this type of parallel would imply a semantic parallel between its two components.

3.3.1 Liu and Takahashi

The solution to the dilemma of Liu’s and Takahashi’s differing interpretations is supplied by an intertextual model. Takahashi gives no intertextual justification for his interpretation, whereas Liu relates this poem to a number of others in his selection.

But first things first: let us look more closely at the poem’s images. Examples of anomalous use of lexemes which qualifies as Riffaterrian ungrammaticality are found, signaling indirection of expression in three lines of the text, the other line (l. 2) containing the formal parallel no. 7 (the heuristic effect of this kind of parallel is discussed in 3.2.1). In fact, as is usually the case in poems of this genre, we have an image per line. Briefly, the ungrammaticalities are the use of “deep” to modify the candle’s shadow or reflection in line 1, the poet’s assertion that Chang-O “must regret” her lonely isolation in the moon in line 3, and finally the unexpected use of the single character for ‘heart’ in line 4.

From the first line we know that someone is sitting up late at night in a rather elegant interior; the word “mother” (Liu’s translation for “mica” reverses the order of the two characters “cloud-mother”), as well as semes like “delicate,” “creating subtle reflections,” of the material mica suggest that this is a feminine environment, contra Takahashi. Line 2 tells us that this someone has been waiting all night. This is plainly additional scene-setting, but here the internal syntactic parallel certainly has the effect of foregrounding the unusual use of verbs representing downward movement. The connotative affinity of these verbs with line 1’s “deep” (parallel no. 6) (a) tells us that both lines belong to the same matricial environment and (b) helps explain this unusual use of “deep” by relating it to the lateness of the hour. If it were not for no. 6, line 2’s parallel might indeed be semantically quite banal; but “descends” and “sinks” are already overdetermined by the semic load of the “deep” candle-reflection, and thus may act as figures for the state of the lady’s heart. With line 3, there is an abrupt switch of scene and protagonist to the world of the Chang-O legend, and the reason is not clear from a mimetic reading. With its apparent novelty of regret, this line is the nearest to direct (mimetic) expression in the whole poem. That the explanation is an affair of the heart is hinted at by the obtrusive use of that word at the end of the last line; the expression “green sea blue sky” tells us that this is indeed the macro-environment of the moon goddess’s lonely exile, as opposed to the human micro-environment of lines 1–2.

We should note that the lady waiting alone is an understood subject in line 2. In line 4, the word “heart” is foregrounded by coming as the final word of the poem. This is reflected in Liu’s translation of line 4: “her heart every night” (cf. 2.3). Line 2 and line 4 are thus secondary images generated by Matricial Phrase 1 and MP2, which are based on lines 1 and 3, respectively. So, even in this short poem, we have two images per matrix – a pair for each textual situation.

3.3.2 One matrix or two? Intertextual evidence

The change of protagonist in line 3 is the most obvious clue to the existence of two matrices. Also, it seems plain from the above that we have two image-generated environments: that of the woman waiting alone until late at night, and that involving Chang-O and her regrets. Each line contains a single image, the first two being generated by the first matrix, and the last two by the second matrix. At one point Liu puzzlingly talks of the poet’s representation of the mythological scene as if the imagery of the first two lines is part of it. Line 2 obviously could be, but the candle in line 1 (plainly an item of a human interior) does not fit this “unimatricial” scenario. Liu was perhaps influenced by the formal parallels described in 3.2.1 (no. 3a), for example, the possibility that the mica screen has something in common semantically with the expanse of sea and sky. Indeed, if the whole poem were nothing more than a representation of Chang-O’s loneliness, then it could have a unimatricial structure. But this would of course conflict not only with Liu’s own summation of its structure, but also with all the intertextual evidence (“Chang-O” is poem no. 28 in Liu 1969: 99.)

If the “domestic environment” semes in “mica screen” and “candle” are not enough to dissuade the reader from a unimatricial interpretation, the existence of a number of other poems by Li comparing a human woman’s situation with that of Chang-O should convince. Particularly apposite is poem no. 35 in Liu’s selection of works, the title of which is Again to the Sung Sisters of Hua-Yang Temple, on a Moonlit Night. This poem also likens Chang-O’s isolated situation to that of the nuns in their temple; the first line tells all: “Stealing the peaches and pilfering the elixir cannot both be done.” The former expression is a common Li euphemism for clandestine love, and the latter alludes to the inaccessibility of someone committed to the monastic life (via the legend of Chang-O’s exile in the moon as a result of her theft). The message is that shutting yourself away in a temple (like Chang-O on the moon) is going to make you miss a lot of fun. This is corroborated in Liu (1969: 106). The nun is obviously being chided for her inaccessibility. The two situations are very similar to those in our poem, and the relation between them – the “intermatricial relation” – is the same: exile is going to make you sorry. This element of scolding is the essential of the intertextual structure, and it is surprising that Liu omits to mention it in his notes on “Chang-O” despite its presence in no. 35 and other texts.

According to Liu, poems 28–34 in his book all take their cue from no. 35 and refer to a romantic relationship with a Taoist nun. Interestingly, he omits poem no. 36, where a male lover prepares wine but awaits in vain, i.e., the latter is much closer to Takahashi’s interpretation, but Liu does not consider it relevant to the case of “Chang-O.” For Takahashi’s interpretation to be valid, we would require an intertext where a male lover was waiting in vain. Of course intertextual models need not have the same lexicon as the primary text; but it is hard to see how a male lover’s wait could be reconciled with Li’s use of Chang-O’s self-imposed exile on the moon in so many other of his poems as an extreme version of a lady’s situation, in order to demonstrate what might happen if she carries on as she has been doing: sealing herself off in a world of her own.

The object of the love is not so explicit in the two poems which follow “Chang-O” in Liu’s collection, but both are clearly divided into two matrices corresponding in extent to the couplets – the first of which describes a nocturnal scene of this world, and the second, the solitary watch of Chang-O on the moon. Further, no. 29 contains an element of reproach in addition to the word “heart.” Once again, the message is that a human person should not imitate Chang-O, and this idea constitutes the relation between the two matrices of our intertextual matricial structure. Briefly, in each poem sharing this structure, we always meet the following two situations: (MP1) the feminine protagonist isolated from the world, and (MP2) the regretful goddess in exile. The fact that the “regrets” are integral to this intertextual structure would tend to prevent them from being a novelty in our primary text. Moreover, the regrets of Li Shangyin’s poem must be seen against the background of Bai Juyi’s famous preceding longer poem Chang Hen Ge (長恨歌, ‘Song of Everlasting Regret’). If “Chang-O” could be taken in isolation, the regrets imputed to the awaiting lady might be considered innovative – but that would be to ignore literary tradition. Thus, our poem can hardly be considered “proto-modernist,” despite its clearly paradigmatic semiotic structure.