for the article in the Electronic Book Review
by John Cayley [ Programmatology ]


Images is one of the sections of my book, Image Generation, further documented on my personal website, Programmatology. It is also part of a long standing engagement with a piece by Samuel Beckett that was incorporated into the English version of How It Is.

As set out in the article, for some years now, I have been fascinated by what is considered Samuel Beckett’s last novel, Comment c’est (1961), translated by Beckett himself as How It Is (1964), and, in particular, I have used twenty two paragraphs from the first part of this novel (as well as the entirety of the novel) as supply text for the generation of pieces that I consider to be works of language art with computation (or, if you prefer, eliterature, or digital language art). These twenty-two paragraphs were also separately published by Beckett in French as a sort of narrative prose poem, ‘l’Image’ (1956), ‘The Image’ in English. Thus, this sequence of paragraphs have a certain integrity of their own within the novel. A number of the resultant pieces – the outcomes of algorithmic processes with quasi-random, non-deterministic aspects – are published in Image Generation (2023).

As a supplement to the EBR article, I invite readers to explore the underlying processes that I have applied to ‘The Image’ – as supply text or ‘prompt’ – along with the outcomes from these processes, and compare them to very recent and admittedly cursory chat interactions with the OpenAI and Microsoft New Bing GPTs and their models. In particular I recommended reading through my most recent engagement, ‘“l’Image” in How It Is’ and thinking of it as way of generating a new text, in the style of, in hommage to, and digging into Beckett’s writing and thinking for How It Is. Clearly, generating outcomes from a direct chat request for text ‘in the style of Beckett’, or by way of my own idiosyncratic descriptions of ‘The Image’s style and content is a quite distinct procedure as compared to the elaborate programming behind the outcomes published on the web and in Image Generation. But that is also the point. Which way of working gives us more to think about as readers and artists of language? How do the outcomes compare as such, as things to read?

Two other procedural and computational engagements with ‘The Image’ are also found in Image Generation and listed here below. However, their outcomes are less obviously ‘in the style of’ or ‘expansions on’, or ‘extensions’ of Beckett’s writing whereas, in making this iteration of ‘“l’Image” in How It Is’, I was conscious of trying to stay faithful to Beckett’s impetus in How It Is, the novel.

‘“l’Image” in How It Is can be read as a reconfiguration of the two manifestations of Beckett’s last novel and its published antecedents, in both French and English. If we think of the words and/or constituent ‘grams’ of Beckett’s work, in either French or English, as the ‘figures’ of certain levels (or pseudolevels) of linguistic structure, then this piece both configures and then reconfigures them.

‘“l’Image” in How It Is’ is part of what will be a larger project I am calling Comment’tis in so far as it is addressed to Beckett’s novel and its French original Comment c’est. In the printed version of the augmented Image Generation only one outcome from the project is presented. The underlying paragrammic algorithms will also be applied to other supply texts in the future. Both a static and a ‘live' version of the outcome from this work can be accessed by following the appropriate sidebar links.

Procedure. I surmised at some point during 2022 – when the possibility of a new edition of Image Generation was in the air – that I should be able to write an algorithm that would take all the words of Comment c’est – call it the ‘guest’ – and find them or their constituent letter sequences in the English of How It Is – the ‘host’ – and vice versa. What you read in the printed version of ‘“l’Image” in How It Is’ is a somewhat more legible proof of this potential, augmented with my own non-algorithmic (but digitally-assisted) ‘writing into’ what is produced by the algorithm.

Rather than applying the algorithm to all of Comment c’est, I take a portion, twenty-two paragraphs, of the novel’s first part – separately published by Beckett as ‘l’Image’ – and use this as the ‘guest.’ I then find all of its constituent letter sequences within the entirety of How It Is.

Thus, when you turn to the beginning of the book’s printed outcome, you are reading words from How It Is that contain all of what I call ‘grams’ (or pseudograms) of the French in ‘l’Image’. The letters of the French are set in bold. You can pick out and then read the French, if you have necessary inclination and patience. They are paragrams with respect to their (phonogrammic) hosts and the hospitable text that contains them.

There are a lot of peripheral complications and no little complexity here, along with a set of parameters some of which become important for later, related work. For the moment, suffice it to say that in this rendition, what I call the ‘definition’ of a particular French word is stable throughout the generated, hospitable text. ‘la’ will always be hosted by the word ‘last’ for example. This is the first possible host the algorithmic reader comes across that contains the ‘la’ sequence, starting from How It Is paragraph 150.

The algorithm seeks hosts for all of the guests, trying, in succession, to find host (English) words that: (a) match the French word exactly, (b) contain all of it, (c) contain syllables of the French word, or finally, if all else fails, (d) contain longest-possible sequences of letters from the French.

With regards to what I have ’written into’ the outcome as published, all those English words that do not contain French sequences of letters have been added by myself. The initial algorithm produces a less coherent text that consists exclusively in the English words that ‘define’ the French. The intervening words that I have added are not just ‘mine’, however. They are all phrases that I have found in How It Is and they are also phrases that I have found collocated with one or other of the words that do contain French sequences of letters. I made an arbitrary rule not to add words between those that ‘define’ and contain a single French word.

‘one image tongue’ is a zero-count stitching of the entire text of ‘The Image.’ It is a hybrid performance-targeted version of the text assembled from two iterations of the generative code. In 2011 the piece was performed by Ian Hatcher and John Cayley and filmed by Peter Bussigel.

‘“The Image” in Common Tongues’ is a loose-linked micro-collage that also contains the same text – that is, all of its successive common phrases (not always the longest) – found, by manual internet search, within fragments of language that were not composed by Beckett. They were not, for that matter, composed by the author, although the author did do the hand stitching. There is more information on this process in articles for The Electronic Book Review and Amodern.