writing|escritura|idaztea

The Language Thief

12 June 2026

I recently heard someone I “know” only from social media talk about their years living abroad with their family. They said something to the effect that they had been able to survive the challenges of life in a different country, a different language even, in no small part thanks to the community they had found on social media. Online they were fed an endless stream of advice from around the world about the best places to shop, how to find the best schools for their kids, how to navigate local bureaucracy and the like.

I envied this person not one bit. To the contrary, I felt a twang of patronising pity. But this brief exchange was a gift of sorts. It made me understand how privileged I was to exist in this world before it was taken over by social media. Before the hell that is the family WhatsApp group; before parents could pick up a call from a needy teenager asking what cupboard the peanut butter is in while climbing Mount Everest; before you could answer an “emergency” text about the dog’s toothbrush while in bed with your lover in Paris. Before you could buy an amazing, and amazingly affordable, European wine without the aid of Google translate.

I am fond of saying that Spanish is one of the few things in life I got for free. Well, not entirely free. Learning the language cost me more glasses of cheap red wine than I can count, on top of the lesbian drama ignited by endless arguments with my first girlfriend. I was young and so was my brain, I was living in a country (Spain), not yet two decades out of a repressive, claustrophobic military dictatorship that had left the nation largely isolated; few people my age or older spoke English fluently (French was still the educated class’s world language of preference). I had little choice but to learn Spanish. And I had lots of time – not just the gift of youth, but the gift of life without the demands of a smart phone.

We didn’t have a phone in the flat I shared with a group of French students, so to ring my parents every few weeks I had to travel by bus or metro to the old Telefónica building on the Gran Vía in Madrid, where I paid a few pesos to make an overseas call in a little cabin. Sure I missed my family and friends. I wrote long and numerous letters, and received as many back. I have preserved all my late father’s blue airmail letters from that time, even though I can still only read every third or fourth word of his scrawl. My years of handwritten correspondence helped to develop my writing style in my own language, even as I was learning to speak, live and love in another.

Yes, I was a middle-class brat. But I was also a very lucky brat – not just because I had the financial and cultural capital to travel and live abroad for a couple years after university. But because I did this is an era before the internet cramped the world. More importantly, though, this was before the era of what Mark Fisher called “capitalist realism” “– “the widespread sense that not only is capitalism the only viable political and economic system, but also that it is now impossible even to imagine a coherent alternative to it”.

I think of Fisher’s thesis often today as I hear people (and not just cynical opportunists like Tony Blair) argue that there is no point in resisting Artificial Intelligence because it is to here stay; our only hope is to adapt as best we can. This fatalistic attitude, one based on the belief that there is no point in resistance because human beings cannot change the world, does not come out of nowhere. It grew out of the triumph of neoliberal capitalism in the late twentieth century, which coincided with the collapse of Soviet socialism, which, for all its horrors and errors, provided evidence that there was an alternative to capitalism.

In the era of capitalist realism it’s not just that technology increasingly steals our work by committing what Sarah Hall calls “creative larceny”. It’s that it has stolen our imaginations by making it impossible, as Fisher argues, even to dream outside capitalism. Today, it seems, the best one can do to resist digital technologies and AI is to invest time (and often money) in forms of individual acts of will and control: breaking up with your phone, censoring your kids’ social media intake, etc. These are not forms of resistance; they are forms of accommodation.

Of course people still learn languages and speak across them. There is still human translation (just). But the odds are stacked against anyone whose first language is English who wants to be less of a linguistic imperialist, to stretch themselves, experience the challenge and discomfort of isolation, loneliness even. The wider context actively discourages this, while actively encouraging over-reliance of technologies that drain our dreams.

I didn’t learn Spanish in Madrid in 1989, watching from across Europe as the Berlin Wall fell, because I had a special talent for languages, or because I was so disciplined that I refused to speak English when I had the opportunity. I learned it because my time and creativity had not been hijacked by social media and artificial intelligence. Most importantly, capitalist realism had not stolen my imagination.

And as we all know, there’s not much you can learn – including a new language – if you can’t imagine a world you have not yet encountered.

DID EMILY DICKINSON USE AI?

5 June 2026

On 1 April this year the Emily Dickinson Museum posted on its Instagram page: 

In an effort to silence critics who mistake the poet’s signature punctuation for AI, the Emily Dininson Museum is officially rebranding as the Em Dash Museum. The Museum aims to reclaim the dash from modern technological woes and return it to its rightful owner. After all, the em dash is named after Emily…

The April Fools’ Day joke provoked some fun comments about AI’s time-travelling talents and the esteemed origins of a common punctuation mark. The truth is more banal. The em dash (—) takes its name not from one of the most famous American poets, but from the typed letter (M) which is roughly the same length. 

According to the copyeditor’s bible, The Oxford Guide to Style (2002), em dashes have a number of functions. They can replace an introductory colon (“She has but one hobby—chocolate”) or be used to “to clarify sentence structure, to express a more pronounced break in a sentence structure than commas, and to draw more attention to the enclosed phrase than parentheses” (“Going—going—gone!”).  

Although the authors are at pains to stress that the em dash is not to be confused with mere brackets or the lowly comma, they also advise using it sparingly, as overuse “can appear jarring on the page”. The discerning editor who detects an over deployment of em dashes in a text is advised to replace some of these with alternatives. 

Much to the exasperation of editors and writers alike, em dashes now appear so frequently in published text that they have become (along with semi-colons and the “rule of three”) objects of suspicion, identified by AI-detection software as telltale signs that a text has been written with the aid of artificial intelligence. Why are chatbots enamoured of em dashes? One explanation is that people writing in English were already using them a lot before AI, so the large language models “trained” on human-written text are simply reproducing that tendency. Unless the models are later corrected by people, the patterns they pick up in early training become “hardwired” into LLMs. This could explain why, today, em dashes appear more often in AI-generated text than in text written by people – and why the machines designed to detect AI fraud increasingly reject texts with lots of em dashes. 

As a result, authors who have already seen our work plagiarised by LLMs, as well as editors who find ourselves internalising the paranoid messaging of the AI police, are now being told that we should avoid the em dash altogether lest we be accused of cheating. 

A more felicitous outcome of the (not strictly moral) panic around the em dash is the attention it draws to the finer points of punctuation in prose – something that hitherto probably flew over the heads of all but the geekiest of readers. Writers of fiction and nonfiction are rarely associated with a particular punctuational pattern. For poets, on the other hand, punctuation is a defining element of style. 

Take the poetry of Emily Dickinson, famously full of em dashes. Many of Dickinson’s lines end in them; in some poems (e.g., 386, “Answer July— ”) every line, including the last, ends with a dash. In poem 492 (“Civilization—spurns—the leopard!”) most lines have dashes between words and at the end (though the final line, like the first, ends in an exclamation mark). 

It’s difficult today to imagine Dickinson’s poetry without the ubiquitous em dash. Yet this telltale mark comes to us through the interpretive choices of the people who originally transcribed the poet’s handwriting. Scholars of Dickinson still debate whether the em dash – or the shorter en dash or hyphen, or any dash at all – is the best way of conveying Dickinson’s original punctuation in print. 

But the question of how the rise of AI affects readers’ relationship to Dickinson’s poetry does not end with the em dash. In a recent study, researchers showed students an original Dickinson poem (“Could I but ride indefinite / As doth the meadow-bee”) alongside one written by OpenAI’s GPT4o based on the poem’s first stanza. A significant majority of participants preferred the AI-generated poem. One criticism of Dickinson’s original work was indeed its (over)use of em dashes. But the students also found the AI-generated stanzas more emotionally engaging and evocative of their own experiences of nature. In contrast, they “dismissed” Dickinson’s, “unusual or disjunctive imagery as ‘stilted’ and ‘distinctly unevocative and jarring’”.

Other studies have come to similar conclusions: not only can human readers (including people with degrees in creative writing) often not distinguish between human-written and AI-generated prose or verse; they tend to prefer the latter, citing both consistency of style and emotional resonance. Yet, when the same readers are informed that their preferred text was generated by a machine, they report feeling disappointed, even upset. 

These reactions suggest that obsessions about the proliferation of the em dash in AI-generated text are a red herring. The more important question for the future of writing is not one that can be answered by the AI police, or even by a good editor (alas). The question is: what does it mean when readers who turn to literature seeking human connection discover that they are more likely to find it in a text strung together by a machine trained to predict their desires than in a text – with all its quirks and surprises – written by another person? 

In praise of self-editing

28 May 2026

Over the years I’ve heard different versions of the quotation, “I’m not a great writer, but I’m a great editor.” I finally googled it, and it seems the words are those of the late American writer James Michener. The exact phrase is, “I’m not a very good writer, but I’m an excellent rewriter.”

The point is straightforward enough. You don’t win a Pulitzer Prize by submitting a first draft. But these days, when everyone online is their own writer, editor and publisher rolled into one, there are a lot of first drafts out there. And too often they make the online reading experience a real drag. 

It’s not the typos and grammatical mistakes that betray these write-and-press-publish pieces. Even before generative AI began its demolition job on writing, autocorrect features helped writers to catch a lot of typos (though not the fun ones, like the “rook reviews” that appeared in the original version of my last post). It’s the casual tone and long-windedness that make so much online writing, including on platforms like Substack, read like first drafts. Their chattiness, their all-around-the-houses-to-make-a-single-point quality give them away.  

Part of the appeal of blogs when they first appeared in the nineties was the novelty of the online diary or journal. Blogs helped to make culture, politics and even arcane academic subjects more accessible by stripping away the formalities of much written prose. But thirty years later that style of writing no longer feels fresh – because it’s not. And a lot of what blogs used to do – whether as confessionals, political rants or mini physics lessons – is now done more effectively, with more bells and whistles, in video format.  

Blogs started long before smartphones and finger scrolling, and before these things started to kill attention spans. Today there’s an expectation that readers will scan or even skip most lines of a post. Why bother taking the time and effort to edit something that few, if any, people will read every word of?

The boring answer: we don’t just write to be read; we write to become better writers. Good writers care about our readers, so we take the time to make our writing worth reading, even if the only reader is ourselves.  

The better answer: there are very few good writers out there. But with a little self-editing, even the most mediocre can become good editors.

A MINI ODE TO THE BOOK REVIEW

A Gift to Writers and Readers Alike 

21 May 2026

Writing a book review is one of the best ways to engage closely with another’s writing. In turn, there are few things in a writer’s life more satisfying than to get a good review. By “good” I don’t mean a review that reads like a series of generic one-liners of the kind publishers plaster onto book covers (endorsements which, if not actually written by AI, certainly read like they could be. Just how many “best writers of their generation” can we have among a given cohort? How many decades can an individual writer be “at the height of their powers”? How many empty superlatives can a single publication generate?).

No, I mean the review written by someone who has taken the time to read the book carefully and is prepared to be honest about it, flaws and all. Reading for a book review – like reading for research and reading intensively for pleasure – requires what scientists call “deep reading”, a skill that is largely lost in the practices of skimming or scrolling. It requires not only intention and attention but memory. Taking notes, without which it would be impossible to review a book, helps the process because notes act as aides-mémoires.

Writing a good book review also enhances our editing skills because reviews are typically quite short (say 500 to 800 words, significantly shorter than this post, for example). That means lots of cutting and refining before submission or posting. Good writing may be a pleasure, but it’s one that requires some discipline – and readers as well as writers benefit from that discipline.

Finally, book reviews are reminders that writing and reading are communal activities, conversations across pages. A good reviewer is the ideal reader – the kind of reader every writer should aspire to be.

Read the full text here: https://peninfist.substack.com/p/do-we-write-too-much