Introduction

What patience he did not take Roger Little in 1965 to dial “manually” the first index of the work of Saint-John Perse, Word Index of the complete Poetry and Prose of Saint-John Perse[1] ! How many successive checks, and different accounts randomly displaced accents or homographs escaped the vigilance of the reader, before it could establish the exact amount every time !

But, since the publication of the book, the tedious work which he had compelled revealed its usefulness to researchers. They were able to, flipping pages, compare the frequency of words, or check what their intuition reader had their sense : the importance of the sea vocabulary, words “dream”, sands”, or “cell” or the abundant use of adjectives “grand”, “top”, “while”, etc. . .

However, whatever may have been the value of the company, the program has quickly shown its limits. Completeness of the count has been compromised by the publication of new texts : the speech “Dante” in 1965 and other writings published in the edition of the Complete Works procured by the poet in the Pleiades in 1972, new poems – “Nocturne” in 1972, “Drought” in 1974 – and finally the ultimate collection, Song for an Equinox. En effet, a few months before his death in 1975, Saint-John Perse had regrouped under this title the last two poems and two former texts, one of which is already entitled “Song for an Equinox”.

Otherwise, the dual statement that the English critic was in the title of poetry and prose was confusing. Saint-John Perse has often commented on what he called verse or verse to the renunciation even poetic prose appellation for his published collections. [2]

But what about ceremonial speech, tributes or certain witnesses ? Are they prose or poetry ?

Certainly the pace, images or sequential organization reminiscent of certain poems, and often we feel it's purpose “make artwork”. This is also the case for a number of letters chosen by Saint-John Perse for the edition of the Pleiades, especially for Asian Letters[3].

However the repository nature of these circumstances did not allow their works to be included in the poetic work with poems published as such by Saint-John Perse and are “irreducible to the whole temporal order, free from any time as any place”[4].

In any case, it is the poet himself, in the edition of the Pleiades in 1972, has clearly defined what constituted his poetry. His death, in 1975, has finalized the corpus as Saint-John Perse had rejected in advance any published posthumously under the title. And reissue Complete Works in 1982 has only added new poems in the collection song for an equinox, which he himself had prepared the publication.

It now became possible to work on a corpus rigorously defined and published in a collection, the reference applies to all researchers. The development of computer technology did the rest by ensuring the completeness and reliability of the indexation. It was enough just to set the parameters of the protocol according to the particularities of the work.

As Nancy linguistic database did not have all the poems of Saint-John Perse, The full text of the poetic work was “scanned” and transferred to a word processor by the team of technicians[5] de l'UPR 6861 C. N. R. S. at the University of Nice-Sophia Antipolis. Proofreading and successive audits have allowed us to almost completely eliminate mistakes or typos, entrust to reliable text software Hyperbase[6] used for indexing.

The corpus was allocated over 9 cards combined into a single file in a stack of this software. Each card consists of the text of a collection or set of poems as they were defined and paginated in the edition of the Pleiades published 1982. For the ninth card, it seemed more appropriate, however, to restore the organization and the general title Song for an equinox as Saint-John Perse had set for the publication of his collection by Gallimard in 1975 ; rather than booking an extra two last poems, “Drought” and “Nocturne”, as was done in the edition of the Pleiades 1982, likely to meet the pagination of the previous edition of the complete works. The reference “Chant” relates therefore successively “Drought”, “Song for an Equinox”, “Nocturne” and “Sung by one who was there”. But it is obviously the pagination of the Pleiades designated by the references.

Finally it seemed essential to correct two shells of this edition and return one where instead of one or (in the last line of the page 410 for Birds ), and, as did Albert Henry in the critical edition’Friendship of the Prince[7], pure figs instead of ripe figs in p. 67.

In the treatment of lexical corpus, however, recognize the limitations of a Hyperbase software that has dissociated O so that the O’ it was the same form, with or without upper because of its position at the top or within the sentence. We must therefore add the two sets of occurrences to assess the significance of the invocation. But we could also note that this limit back to advantage in a case homography. En effet, not more than in the other index, it was only possible to obtain an automatic distinction here between homographs ; except for the French word notice and notice that Latin, having taken timely acclaim in the ornithological nomenclature for use by poets, found himself, as in the index in the text, capped by a capital.

If reliability is the first virtue of the computer, completeness depends on the program that is offered. But this index has wanted closer to a poet's words directory as the lexicographical tradition. This is why one does not find, grouped, different forms of a dictionary entry. And if lemmatisation was not performed, this is not so because the software does not allow the automatic realization that because the different forms were more interesting in themselves. En effet, for a poetry text where phonemes are so important, is it not artificial to give an infinitive unaware that the poems, the singular of a word that Saint-John Perse never uses the plural? Surely available in alphabetical order is a first artifice, but maintaining all the separated forms, recorded and referenced, to better appreciate the choice of time and the phonetic qualities of genres, numbers, or conjugated forms. This choice makes the sensible particularly frequent and significant use of the plural in the writing of Saint-John Perse which brings in big crowds “all sorts of men in their tracks and ways”[8].

It's still the same principle that spread the English and Latin quotations highlighted some poems since, by definition, the words they contained had not been manipulated by Saint-John Perse. On the other hand, were kept English words (Thanksgiving Day, starling), Latin (ave, notice), and the words of the islands and elsewhere (Annaô, anhinga, Ologhi, Yaghé) it was appropriate for their rhythmic connotations or their virtues. It was the same for words in titles, that have been recognized under their respective forms.

However, to create an index that is useful and clear enough for the greatest number of readers, it is necessary to operate some filtering.

While, the relative thinness of the poetic work of Saint John Perse had allowed indexing of all forms, but the identification of punctuation (5777, / 1892. / 1067 … / 1003 ! ), it is valuable to study a language or stylistic, have occupied excessive space in a work intended for all researchers.

It was for the same reasons given for indexing words tools, articles[9], very common prepositions (at / to / aux / from / d’ / from June / of / in / sur / in), Very high frequencies generally for recovering these forms in most texts, their identification did not seem to make any special revelation. However Annexed the list of forms and eliminated, with an indication of their frequency.

The rule was to keep only forms whose frequency was below 500 occurrences. However 6 forms have escaped this rule : we (865) / that (689) / is (667) / as (596) / plus (511) and finally the sea word, Featured with the lexicon 530 occurrences for the singular and 35 for plural.

The absence of any other filter allows to find all the relevant forms for stylistic study and a large number of elements of linguistic analysis.

But he had to take into account the ambiguity qu'entretiennent homographs. For up, in all cases where the problem may arise, it was therefore necessary to turn to the text and examine the context. Note simply that if this disambiguation is necessary when one wants to deal with such words as is (verbal form) and is (point cardinal), or not (substantive) and not (adverb of negation), polysemy on which nature plays poetry, and especially that of Persia in which distant etymologies and phono-semantic all games generated by the context secretly working words, polysemia this does not lead, satisfactory, as part of a single index, disambiguation an operation that would have required the multiplication of too many forms. So, for example, it has not been possible to thwart the antanaclasis under the square shape that equates element of the verbal noun instead take place[10]. But in all cases where it was necessary, the grammatical category of the form – preposition, adjective, participle past, adverb, verb or noun – was identified subentry.

Finally, we must recognize that compound words and capital letters are the limits of the reliability of the process of indexing as Hyperbase does not distinguish the internal hyphen other separators, and it accounted separately the same forms as they had, or, the shift to the initial. This required return “manually” the compound forms. Several cross-checks have limited the risk of errors or oversights. They probably have not been completely eliminated.

Compound words are indexed withdrawal 5 spacings (to indicate that all their elements are already recorded with their respective entries) just after the first word of the compound form. And they replace, without removal, if the first word has no other occurrence.

If the first word is not the main word, They also reappear in their compound form, and withdrawal, but this time preceded by an asterisk, just after the main word group.

Finally, each other within them is indexed in its place in alphabetical order. If it has any other occurrence in the text, it appears preceded and / or followed by a – according to its place in the compound word. It is deleted if it artificially introduced ambiguity of a namesake. For example shallows is indexed back down and just after just after funds, but bats comes from the beginning of the line instead of bats, unknown form in the poems of Persia, and does not appear as the mouse poetry ignores these vulgar rodents. As for the name of Pi-lu plant, the second element is not indexed separately because obviously can not be considered as read. There are already enough homographs in simple forms that we do not venture to add more with foreign body members!

If all the compound forms have been listed, some of them, created by the poet for a single appearance, does not appear in their complete form. This is so for very fine / very white (Elog 7a) / very skinny (ROIS 65a) / very real (AMER 372d) A little-wet (Rois 60a) / Man very attractive or custom-Sans-among-us (ROIS 65f), or called for the verb phrase-by-their-name (Elog 25e). But it is mainly for one reason typographical, because it was too long, the poetic hapax yellow stained-de-black- purple-à-la-base (Elog 36b) was not retained in the index, while the grass to Madame-Lalie managed to drag them whole!

Anyway, include all elements of these compound forms, recorded and referenced instead that their assigns alphabetical order. And just to cross two to regain expression reference. It is the same for compound pronouns those / that one, or adverbs up there / over there / above or numbers as thirty-two.

The capital was restored only for so-called proper names, or more specifically for names of persons, Peoples, of country, of countries or even feasts. It has not been possible to identify other words(substantivized nouns or adjectives) that, equipped with an uppercase, suddenly took on the appearance of a title ; sometimes at all jobs like Narrator or Storyteller, sometimes in some instances only as for seas, islands or chronic, and often in the plural .

Thus limited, Capitalize the names were too few to constitute a separate list. And keeping them in the General Index allowed to have phonetically similar sets (at least for the first syllable) and could suggest a research track.

But if the software had allowed to index all the words that Persia had allocated a capital, it would have been interesting to present from proper names, in a particular list, All common names and ennobled[11], and which are very often a hapax, like all those Whistleblowers, Disputants and other Inquisitors. And we could see God arise suddenly force, eleven times, in the last two poems, whereas, in bitters, Poet always the graphic sign of his nobility.

Anyway all these reserves, references and frequencies remain above suspicion. They are the “bitter” index provided that, d’ Island hopping, we take the time to visit the archipelago.

Like all other, this index allows to first easily find a quote : from a word you have in mind or, even faster, from two words. Just cross all their references. In the vast majority of cases, one of them is their common : it is that of the searched citation.

Mention may also be, from the index, study the use of color : which, above all, under what works ? And, for example, play all green[12] the whole work with their namesakes.

Finally, we can compare the frequencies and noted for example that if the dream has noun 8 occurrences, dreams never appears. While dreams can be found 39 time, far behind the singular dream which has 134 occurrences. How then do not give in to the fascination of this “black hole” Charles Muller called zero frequency?

In short the index is a useful map that starts with the text adventure. But the crossing control to know where, when and how “tack”. Is-not unique to “Master of the stars and navigation” that, this time at least, could be the researcher?

Éveline Caduc

[1] The book published by the author in Durham (University Press) in 1966, included 290 pages. But the second edition, to Southampton 1967, already had an extra A 44 pages.
[2] See example, the letter of Mrs Francis Biddle 12 December 1955 : “… the issue of internal metric, rigorously treated in the general distribution and articulation of large masses prosodic (which are blocked, strophic, or leashes, in the same broad and contraction, with the same fate, all treated as individual elements to regular – what they actually are). It is easy, Obviously the foreign reader to misunderstand the general economy of a specific verse, yet qu'inapparente, and that has absolutely nothing in common with the current designs “free verse”, from June “prose poem” or with the large “poetic prose”. This is even the opposite that this ” (Pléiade, Complete Works, p. 922).
[3] And the demonstration has been made that some of them (in particular letters to his mother, said Asia, and the famous letter to Conrad) were composed or, at least, largely recomposed for publishing Complete Works. See the thesis (still unpublished) Catherine Mayaux Chinese referent in the work of Saint-John Perse (University of Pau and Pays de l'Adour, June 1991) ; first part : “False real letters and real fake letters of Asia”.
[4] Letter to the Monnier 26 March 1948, Pléiade, Complete Works, p. 552.
[5] Pierrette Uboldi for text input and Jacques Hammerschmitt for survey.
[6] This software, designed by Etienne Brunet, is a very specific adaptation of HyperCard software for statistical analysis of literary data.
[7] “Friendship of the Prince” de Saint-John Perse. Edition critique. Transcription of manuscripts states. Etudes. Publications of the Saint-John Perse Foundation , N.R.F. Gallimard 1979.
[8] Anabasis X p.112.
[9] Unfortunately eliminating definite articles inevitably led to that of their homographs, the pronouns, the, l’, the.
[10] Cf. Anabasis X p.113d.
[11] Stylème This is one component of the inflation process that I described in Saint-John Perse and Knowledge Creation, J. Courts 1977, p.122-135.
[12] The study of green was the subject of an article I wrote after completing the constitution of the Index “The green of Saint-John Perse “, in Alain Vuillemin Literary databases (book in print).