Nina Leontyeva

Nina Nikolaevna Leontyeva (born 21 November 1936) is a Soviet and Russian linguist who spent her whole working life inside one of the twentieth century's strangest scientific projects — teaching machines to understand language. She was born in the tundra at Naryan-Mar, above the Arctic Circle, registered at birth under a name her father accidentally invented while humming; she became the first hired employee of the USSR's Laboratory of Machine Translation; she built and led FRAP, the French-to-Russian automatic translation system, at the All-Union Centre of Translation — the only woman and, she notes, the only non-Party member ever to head a laboratory there; and when the official campaign against her system came, she outlasted it, defended a doctorate of technical sciences, and carried the underlying idea — automatic understanding of text — into the semantic dictionary ROSS, the analyzer RUSLAN, and a textbook that still teaches it.^[1]

She was also among the first researchers anywhere to study semantic valencies — now the subject of work by hundreds of scholars worldwide — and, in the judgment of her colleagues, produced what may be the first clear definition of a coherent text in the scientific literature.^[2] But the thing that makes her singular is not any one system. It is that she kept a complete record. In retirement she wrote and deposited with the Prozhito corpus six notebooks — childhood, work, people, family, a diary written at eighty-four, and a book of verse — that together form one of the fullest first-person archives we have of Soviet computational linguistics: what it felt like, from inside a communal barracks and a succession of institute corridors, to spend sixty years trying to give a machine a mind for meaning.^[3]

This article draws on that corpus throughout. Leontyeva is its sole author and licensed it herself; unless a source is given otherwise, episodes and judgments recounted here are hers — her memories, her attributions, her candor — rendered in the ALB Team's own English, not translated verbatim.

A name assembled from a refrain

Her name was an accident of music. While her mother was in labor in Naryan-Mar, her father paced the next room humming a refrain of his own invention — Nari, Nari, Narina — a pseudo-"Georgian" song he had made up. Sent to register the birth, he was still humming it; asked the child's name, he broke off mid-syllable, and the clerk wrote down Narina. Naryan-Mar itself, she notes, was widely mispronounced Narinmar — the invented name sounded like it belonged there.

Her mother was not pleased. She revered Saint Nina, the enlightener of Georgia, and had the name changed. (The grown daughter later discovered that Saint Nina is Georgia's patron — a father's fake Georgian song had landed, by pure chance, on a genuinely Georgian name — and came to love it.) But the discarded birth-name refused to die. As a child she cycled through self-chosen names — Alya, the childhood one; Nika, adopted in fourth or fifth grade to sound nearer to her father Nikolai; later Rina, Irina; in Poland in 1980 her hostess Maria Lvovna Mayenowa called her Yanina; and Nino and Narin floated nearest the Georgian original. Then, in old age, when she sat down to write her memories, she reached back for the first name of all: the notebooks are narrated by, or on behalf of, Grandmother Narina — her literary self, a twice-great-grandmother who tells the child's story in the third person.^[4]

Origins: the North, the dispossessed, the writer and the accountant

Both of Nina's lines ran through the same two catastrophes of the early Soviet village: war and dekulakization.

Her father's family were the Leontyevs of Matigory, in the Kholmogory district of Arkhangelsk province — the Russian North, Lomonosov's own coastline. Grandfather Pavel Andreyevich Leontyev had come back from the First World War a St. George's Cross cavalier, was later dispossessed, and ended his working life as a bakery accountant with a passion for Saltykov-Shchedrin — the great satirist of officialdom, a telling taste in a man whom officialdom had ruined. There were four Leontyev brothers, and the family memory kept them sorted like a folk-tale set: Alexander, who died in 1941; Andrey, who became a colonel in the transport ministry; Alexey, the poet and war correspondent who studied at IFLI; and Nikolai — Nina's father.^[5]

Nikolai Pavlovich Leontyev — "NikPal," as his daughter styles him — deserves a paragraph of his own, because without him there is no article. He was a writer: one of the last pre-war collectors of northern Russian folklore, an unpaid outside reviewer for Novy Mir who, in his daughter's phrase, beat graphomaniacs with their own weapon, and a member of the Writers' Union who for years held a singular distinction — its only officially homeless member. A housing bureaucracy letter informed him, literally, "You are officially NOT there." He had volunteered for the front in 1941 without even stopping at home to say goodbye; the war cost him both legs, and he finished his life in a wheelchair, still a dandy — he died, she records, with a napkin tucked into his collar, prompting the family epithet "Aristocrat!" Among the papers that survive are a 1941 hospital letter written left-handed while his right was unusable, and a poem addressed to little Nina in 1944.^[6]

Her mother, Raisa Danilovna Zarubina, came from Saratov by way of her own dispossession. She had been expelled from Saratov University for the class crime of being a kulak's daughter — her father Danila Zarubin, a one-legged peasant (he walked on a wooden peg the granddaughter remembered in detail), had been dekulakized; neighbors smuggled him out the night before his arrest; and years later, in hospital, he took his own life — in the family's understanding, so as not to stand in the way of his children's lives. Raisa drifted where the unpersoned drifted — to Tajikistan — and became an expedition accountant, the trade she would practice from the Pamir to Taymyr. Once, on a mountain road, her vehicle went off a Pamir cliff; she survived. She lived to ninety-three, and her last recorded request to her daughter was: "Kiss me."^[7]

The parents met where the dispossessed met — on a geological expedition in Tajikistan, in basmachi country: he a hydrotechnician and aspiring writer, she the accountant. Their children were born wherever the work was, a geography the father later turned into a newspaper sketch he called "Living Geography": son Tolya at Yangi-Bazar near Kunya-Urgench; daughter Nina in the tundra at Naryan-Mar; and daughter Tanya at Katyryk on the Taymyr peninsula — each child a different end of the Soviet map.^[8]

Naryan-Mar: a childhood at the end of the map

Naryan-Mar, in the Nenets tundra beyond the Arctic Circle, was where the parents' expedition work had brought them: he as a newspaper correspondent and folklore collector for the Northern Sea Route administration, she as senior accountant in the same Glavsevmorput system. Their daughter's earliest memories, filtered through Grandmother Narina, have the texture of the North — enforced quiet, long indoors, adults gone at work.

The most telling is the tea episode. Guests came at the lunch break; the little girl sat silent in her crib — until a neighbor noticed her rocking and murmuring. Listening closer, he heard a small private refrain: "Want tea... want tea... want tea..." — not a request, a self-lullaby, working up to sleep on sheer repetition. The parents took pity and brought the tea. "A calm girl," the mother confirmed — "as long as you didn't leave her alone with her brother." What happened once when they did, Narina declines to recall; she loved her brother. "The main thing: everybody survived."^[9]

There was a gramophone — somehow — and she walked around the table singing with it, and she kept the habit her whole life: still, in her eighties, when sad she sings the crackly shellac songs and conducts the waves with her hands. And there was the wax duckling. Someone had given three-year-old Alya a yellow-and-red duckling of wax; she fed it, walked it, pressed it to her yellow dress. Persuaded at naptime that the duckling was tired too and would sleep on the porch, she woke and ran out into the fierce sun — and found a bright, slightly convex spot. Wax does not survive a Tsaritsyn summer. She fell seriously ill soon after — the abrupt swing from tundra to southern heat, the doctor said — and the family fled by plane for Moscow. The duckling stays in the notebooks as her first lesson that love and matter are not the same substance.^[10]

Her mother had fought to leave the North at all — expedition systems did not release good accountants easily — and they reached Kratovo, outside Moscow, where the father had rented a room, by way of Stalingrad. In Kratovo the memories solidify: the neighbor's chicks she loved with suffocating thoroughness (some literally — a scene she later found mirrored in Steinbeck); the white fur coat in which she fell backward into snow making "little people" all down the cleared path, and talked to them. It was, she says plainly, a happy childhood — and it ended in June 1941.

The war: the last train south

The father volunteered at once — he did not even detour through Kratovo to say goodbye or leave instructions. The mother, left with the child and a Moscow that was beginning to burn, made it onto what she remembered as the last southbound train — in 1942, on Literary Fund evacuation arrangements. The journey took a month through a rail system given over to troops and the wounded.

At Tashkent they were thrown off. The train needed room for soldiers already climbing through the windows without tickets; a mother with tickets to the end of the line, a sick five-year-old, one basket, and three rubles did not outrank them. She stood by the car begging as the departure signal sounded — and the conductor pulled them both bodily into the vestibule. Soldiers made room on a lower bunk for the sick child (the talk was of inflamed glands, or typhus — the memoir preserves the murmur, not the diagnosis). One act of wartime mercy, and the family survived the war on it.^[11]

Stalinabad — today's Dushanbe — took them in because that was where the paternal grandparents had been swept by the same machinery. Nobody met them; the mother hired a Tajik cart-driver to carry her and the bedridden child to an address on Shota Rustaveli Street. The grandmother, Matryona Stepanovna, nursed Alya back onto her feet; grandfather Pavel Andreyevich — the St. George's cavalier, the ex-kulak, the bakery accountant — stood in for the absent father.

The Stalinabad years, 1942 to 1945, are the memoir's warmest wartime pages, precisely because a child experiences a courtyard, not a front line:

• The spool carts. Grandmother sewed for the market on a sewing machine, and the empty spools became Alya's fleet — wheeled carts with a stick for an axle, even a four-wheeled cart. The problem was cargo: she had nothing to carry. Toys and dolls belonged to her friend Nadya, whose father was a minister or something like it, and who was, Alya notes without rancor, beautiful and clean in a way the courtyard Tajik girls were not.
• The doll that wasn't. Grandmother taught her cross-stitch; Alya designed whole wardrobes for the doll she did not have, even built it a carriage. She pinned her hopes on the kindergarten graduation — surely they would give the tall girl at the head of the exercise line, the one who knew all the routines, a doll. They gave her a large second-hand hare with thick wadded arms and legs already clothed in trousers. She decided to sew for the hare anyway.
• Malyuta the cow, the household's milk supply, which the children drove with the others up into the Pamir foothills where almond trees grew — where they gathered sweet pods and mulberries before the Tajik families picked them for market.
• The German suitcase. One morning the courtyard erupted: the upstairs neighbor had come home straight from Berlin, whole of limb, in new uniform and boots, and his open suitcase on the ground held row upon row of sharpened colored pencils whose smell, Alya says, she dreamed about for years afterward. No dolls in the suitcase. She ran back to the games — skipping ropes, hopscotch, Cossacks-and-robbers.

She was in the courtyard gang that raced to the main bazaar square on Victory Day to collect spent salute casings, not entirely sure what had been won. The aunt from Moscow, Tamara — wife of uncle Andrey — arrived with a slab of nut halva ("Eat, children, your fill") and then took Alya and Tolya home to Moscow: seven days by train, at the end of which waited the best gift of all — a little sister, Tanya, born on Taymyr after the war began and met only now, a living doll.^[12]

Her father came back from the war on no legs. The hospital letter of 1941 — written left-handed, his right arm then useless — and the 1944 poem to his daughter stayed in the family; the daughter kept them, and quoted them, sixty years on.

Postwar hunger: Burkovo, the barracks, the bread

What the children came back to was not recovery. It was the long postwar poverty of the Moscow periphery: ration cards that did not cover dependents, rooms rented or begged, a family of five now — mother, father in a wheelchair, Tolya, Alya, baby Tanya — moving from corner to corner.

They lived first at Losinoostrovskaya in a tiny room of aunt Tamara's, then in Valentinovka, then — on foot, across the railway — in the village of Burkovo on the Klyazma, where the landlady Anna Ilyinichna welcomed them with the memorable announcement: "I love writers — Valery Yakovlevich used to live here," meaning Bryusov; girls had hidden in the lilac bushes by the gate to call him out. His room was bright and spacious, with an alcove just right for mother and Tanya. Then the money ran out, and the landlady offered a corner of the shed — thin wall, chickens, one rooster, dawn reveille. They cooked nettle shchi with pickings from the landlady's garden. Bryusov's ghost overhead and a rooster through the wall: the memoir does not underline the irony.^[13]

Hunger organized everything. Bread came on cards — dependents had none — and the children's job was to fetch the family's two kilos of black bread: a loaf plus the dovesok, the scale-balancing extra chunk. They begged the seller for the smallest possible loaf and the largest possible dovesok, because the loaf went straight to market — traded for firewood or kasha — and the dovesok was the day's food. One spring day, walking home from school along the railway embankment, dizzy-hungry, nine-year-old Alya ate the dovesok on the way. Her mother said one word: "Egoist!" "I have known that word very well ever since," the adult writes — and it is worth noticing that the linguist who would spend her career on how much meaning a bare text leaves unstated learned it first as a single word carrying a whole household.

There is a companion episode, the apples vprok. Their mothers' generation had a word — vprok, "put by," "for later" — that meant beautiful things are not for you. Under her sister's bed the mother kept a hoard of Antonov apples for winter, for the sickly Tanya, whose anemia sent the mother to the hospital each morning with grated carrot and apple. Alya, who knew nothing of this, found the hoard — and began stealing one apple at a time, because school fashion demanded hedgehogs: you took a steel pen nib, drilled conical plugs out of a firm green apple, reinserted them blunt-end-out, and produced a hedgehog; after lessons, the hedgehogs were eaten, hunger not being an aunt. Her hedgehogs were the class's best — her apples were the firmest. When the pile visibly shrank and the truth came out, the mother nearly wept: those were winter apples for Tanya. The memoir leaves the guilt, doubled on both sides, exactly where it fell.^[14]

Yet the same years produced the sentences a future linguist would build on. In third grade at the Kostino elementary school her first English lesson arrived — a young, charming teacher fresh out of the Institute of Foreign Languages — and Alya fell in love on the spot; English stayed a lifelong line in her work. For the first class party she staged a tiny play in English, Chicken-Licken — deliberately, she admits, so that the elderly homeroom teacher she disliked would not understand it. Already the double move: language as beloved object, and language as a wall you can build on purpose.

There was also a poet of the household's own. In Valentinovka the family had lodged with Maremyana Romanovna Golubkova, a skazitelnitsa — a traditional epic narrator — whom the father had brought down out of the North and who became his co-author on books including Olen'yi kraya ("Reindeer Lands"). Through her the girl's Moscow childhood kept one door open onto the tundra and its tale-tellers. And on one hungry birthday, brother Tolya produced two glazed curd bars; watching how fast she ate hers, he pushed his own across — said he didn't want it — and left for school having left behind a small Lermontov volume inscribed, in effect: to Nina — modest, but from the heart. Her first gift. She kept it all her life.^[15]

Kostino: the barracks, the POWs, and Faust

The family's long postwar station was Kostino — now inside the city of Korolyov — where the mother, as the motor depot's accountant, was at last allotted rooms in a workers' barracks that officially did not exist: it had been struck from the register as demolished thirty years earlier. "You are officially NOT there," the letter said — to the writer who was already the Union's only homeless member. One hundred meters away stood a camp of German POWs; the children knew the wire, and little Tanya once slipped through it. The father, away in the North on assignment for years, finally came home after what the family remembered as a council of war in letters.

The barracks sang. In the communal evenings the neighbors' voices made a choir, and the father — legless, wheeled, published — was its seventeenth voice. She sat her school years out there, third grade through graduation (1947–54): positions of responsibility from the third year on, then a public stripping of Komsomol posts in ninth grade in the sour post-Stalin atmosphere — an early education in how institutions distribute honor and withdraw it. For literature she did what the family did: read through the barracks' evenings. Pasternak's Faust and Onegin under the tomato bushes of the mother's miraculous Kostino garden — the depot accountant grew a garden her daughter remembered as a marvel — while, three meters from their window, a new officers' dormitory went up for people the register did recognize.

In 1962 the Kostino gorispolkom finally allotted the war-invalid writer an apartment. The daughter heard his unmistakable northern okanye — the full, unreduced O of Arkhangelsk speech — down a phone line, at the Machine Translation Laboratory where she then worked, and the sound of it is in the notebooks like a homecoming.^[7]

A profession found by elimination

She knew only that she wanted one clear edge — her word, gran', the boundary or facet — and she recognized decades later, with some wonder, that the description of edges and boundaries in text is precisely what her specialty became. Getting there was a process of elimination conducted with a stubbornness she admits cost people who wished her well.

School suggested two paths. Her beloved English teacher pushed the Institute of Foreign Languages — translation, a plausible career. Her inborn literacy suggested proofreading. Both she set aside as hobbies, not professions. Physics she respected but could not feel. Mathematics she loved as problems — she would sit over one for days and refuse hints out of pure obstinacy — but a technical institute's pace was not hers (the irony is on the record: she ended her career a doctor of technical sciences — "theoretical informatics," she smiles). Pedagogy she refused outright — she did not feel psychology and was afraid of children. What remained was texts.

Then the system intervened — by timetable, not by verdict. Moscow State was her dream ("I wanted, above all, to study — I am such a non-student!"), but suburban schools issued their certificates and medals a month after the Moscow ones, and by the time the Kostino medalists arrived, the places reserved for medalists were already full. In despair she walked toward the Economics institute, which vacuumed up the stranded — and her brother literally seized her arm: "Where?! You're a philologist!" They rode into Moscow together and found the last philology faculty still open — at the Moscow City Pedagogical Institute named for Potyomkin, where she, the only non-Muscovite, was admitted without the exams that had already exhausted them in school. At the interview, dean Yuliya Ivanovna had her talk about Gorky's Klim Samgin and asked whether she had read Ehrenburg's The Thaw, just out. She had — it was still "singing inside" her. The Thaw, in both senses, was on.^[16]

MGPI: an institute on borrowed culture

MGPI Potyomkin — long since absorbed into the modern Moscow pedagogical university — turned out to be the accident that worked. The faculty was studded with scholars on their way to real science — the literary scholar Leonid Grossman among them — and the students who were hungry for more simply commuted to Moscow State for lectures. With friends Ira and Masha she went to hear Sergei Bondi ("Bondichka"), and once, drawn by two irresistible names on a poster — Rimbaud, lectured by Golenishchev-Kutuzov — four of them sneaked into an MSU hall and took the least visible back bench. The lecturer — "a tall nobleman" — made them stand and file one by one to the front row, directly before him, then asked the hall which of three languages he should speak; Russian was not among the three. The hall chose French. Four interlopers understood nothing — and stayed anyway. She calls the episode by its moral: "the ear gets used to it." It is not a bad summary of her entire method — sit in the front row of a language you cannot yet parse, and let exposure do its work.^[17]

Two school-and-institute threads that later mattered:

• The piano and the harpsichord. She went through music school, adored her instrument, and the notebooks keep the music pages — including one pairing the harpsichord with the name Vyacheslav Vsevolodovich Ivanov. Ivanov — poet, scholar, and member of the fathers' generation of Russian intellectual life — enters her story at the family dacha years later, in the fire; the seed was here.
• The stubbornness. Out of the family verdict that she was "grass" — her brother's word: grows but doesn't get smarter — she built a counter-principle the notebooks state outright: resist collective fashion, take nobody's ready thought for your own, hoard your own models. The girl who refused to read Cyrano precisely because everyone was mad for the production became the scientist who, when the whole MT establishment declared her system impossible, quietly built it anyway.

And then geology. After her second year she sat her exams early, walked out of the institute, and hired on as a worker in a geological prospecting expedition — as her mother had done, as the family had always done. She moved from expedition to expedition and rose to tekhnik-vychislitel' — technician-calculator, a human computer in the era before the other kind — before returning to Moscow and re-enrolling, this time on the evening faculty, finishing in 1959 while already doing catch-all contract work at the Institute of Linguistics.^[18]

It was a strange trajectory for a future doctor of technical sciences: tundra childhood, evacuee, chicken-shed lodger, barracks schoolgirl, geology-field technician, evening student. But it was the ordinary trajectory of her generation and class — the children of the dispossessed, for whom the Soviet 1950s offered exactly one escalator: education, seized wherever it could be found. The escalator's next stop was a staircase landing where, in 1956, a graduate student named Igor Melchuk was explaining to anyone who would listen that machines could be taught grammar.

The staircase: how she entered machine translation

The usual version of this story — she published it herself in 2009 under the title "How I ended up in machine translation" — begins not with linguistics but with dirt.^[19] In her second year at MGPI she quit, sat her exams early, and hired on as a worker in a geological prospecting expedition — first the Tian Shan mountains hunting minerals, then the Ustyurt plateau in the Karakum hunting water, where she rose to tekhnik-vychislitel', "technician-calculator": a human computer, working a slide rule to compute soil resistivity. It is worth pausing on that job title. Before she ever touched a computer, she was one.

Back in Moscow she re-enrolled on the evening faculty — determined to work rather than "waste" days on lectures — and luck arrived the first evening: a secretary came in announcing that the Institute of Linguistics needed "two girls." Thirty evening students surged forward shouting Me!; the tiny secretary parted them and beckoned to the quiet girl in the far corner: she worked all summer, so she needs work more. That is how Nina Leontyeva entered Soviet linguistics — chosen for having calluses.

At the Institute of Russian Language she did contract catch-all work — the Pushkin-dictionary sector, Reformatsky's sector, the Ethnography institute — and there she met Igor Melchuk, then the young star of the new machine-translation circle. His opening question is preserved: "Do you even know what an algorithm is?" She did not; a passing Misha Beletsky explained algorithm notation in five minutes, and she was hooked. Melchuk set her a first real task — proving that the inflections of the Russian adjective were largely redundant for analysis — and then did the thing that made him her idol for years: he crossed out her prose mercilessly and made her write it again. She argued back constantly, not understanding, she says, that she was spending a great man's precious time. At some point he said: now you can give a paper at the machine-translation conference. (The redundancy result itself got into print — Rozentsveig's 1958 English-language survey of Soviet MT cites "N. N. Leont'eva and G. Kh. Vavilova" on the redundancy of grammatical gender for analysis.)^[20]

The First All-Union Conference on Machine Translation, May 1958, was held at the First Moscow State Pedagogical Institute of Foreign Languages — later "imeni Maurice Thorez," today Moscow State Linguistic University. Two details survived in her memory: the famous Steblin-Kamensky opened the first session with a line of Bryusov — "But you, who will destroy me, I greet with a welcoming hymn" — addressed to the human translators who already felt the machine breathing down their necks; and the first great fight over the intermediary language — Melchuk insisting it must be a system of correspondences, Leningrad's Andreyev defending mysterious "semoglyphs," Belskaya insisting that an ordinary natural language would do. The argument was never fully settled; Leontyeva spent her career on her own side of it.

And then the staircase. Running out of the Institute one day after the conference, she tried to slip unnoticed past the great Vyacheslav Vsevolodovich Ivanov — already a celebrity, linguist, and member of the MT circle — who stopped her at the cloakroom landing and offered her, on the spot, a purely technical post: a Laboratory of Machine Translation was opening at In-Yaz, and there was a lab-assistant position. The exchange, as she recorded it:

"But the salary is very small." — "I can work for free, Vyacheslav Vsevolodovich!" "There's a lot of technical work." — "I'm ready to wash the floors." "Can you type?" — "No, but I'll learn very fast." (answers, she noted, straight out of Jack London's The Mexican*)*

She ran to In-Yaz immediately, sat an hour in a neighboring classroom listening through the wall to Viktor Yulievich Rozentsveig's soft baritone doing spoken French with its beautiful gargled r's — and was hired. Her labor book reads "Enrolled from 1 May 1959"; the order was signed the 15th; she had already been working for weeks. That date — the enrollment of its first lab assistant — is the laboratory's birthday. There was no formal founding order at all: the Laboratory of Machine Translation came into existence when Nina Leontyeva was hired into it.^[21]

Employee number one: the LMP years

The LMP of 1959–1966 was, in her own reckoning, a "virtual" laboratory — it never received official status as a scientific laboratory; its linguists were enrolled as "engineers" and lab assistants, and Rozentsveig ran it essentially pro bono, on top of his teaching. Around it orbited the Association for Machine Translation — a card-file of some two hundred members that reads like a roll of honor of Soviet structural science: the mathematicians P. S. Kuznetsov and A. A. Lyapunov, the philologists A. A. Reformatsky and D. G. Lakhuti, V. K. Finn, Yu. A. Shreyder, the dissident logician A. S. Yesenin-Volpin, and of course Melchuk, Apresyan, Paducheva, Kulagina. Reports drew analogies between MT and statistics, logic, coding theory — even decipherment, with Yuri Knorozov's Maya work held up as a model.

The lab's first composition, after employee number one: Yuri Martemyanov over from the teaching staff; Alexander Zholkovsky and Yuri Shcheglov fresh out of MSU; Konstantin Erastov a year later; and a penumbra of MT-Department students — Mushanov, Korolyov, Vasilevich. (The Department itself, opened September 1959, admitted twenty-five students — all boys.) A second generation — Gindin, Shalyapina, Pertsov, Polivanova — followed.

Vyacheslav Ivanov set the laboratory's great task: find the elementary meanings — semantic primitives, "semantic multipliers" — out of which all translation could be assembled. The idea was beautiful, obvious, and constructive, and it captured them all; Martemyanov unrolled a roll of graph paper "the width of the corridor" in a little-used corner of In-Yaz and began filling in the great table of semantic multipliers. Zholkovsky, amused by how much she argued, dedicated the multiplier negation to her — in a line that is itself a pun on negation. The assault failed, as assaults do: they produced not a working translator but, in Melchuk's terminology, a "functioning model." Yet the same walls incubated the discipline that outlived the goal — the Meaning⇄Text model of Melchuk and Zholkovsky, the lexical functions, and the journal "Machine Translation and Applied Linguistics" (MP i PL), which Leontyeva edited, proofread, carried to printers, and literally sold and mailed to subscribers herself. When Petr Sgall visited from Prague he asked who their managing editor was — and was startled to be shown the lab assistant.

They studied like graduate students of a department that didn't exist: math-analysis lectures with Gavrilova and Shikhanovich, control papers, Vietnamese with Nguyen-Hai-Zyong, Turkish by textbook, Latin terminology and Hausa from Shcheglov — and the junior lab assistant quietly extracting the French verb system from Martemyanov. Zholkovsky and Shcheglov — "the Konstantinoviches" — analyzed Eisenstein films aloud and turned the ideas straight into articles; it was the apprenticeship that would make them famous as literary scholars in America. Out of the exhaustion came the jokes: returning from a brutal calculus exam, comparing how each had computed the sines, Zholkovsky produced a spontaneous verb-coinage so absurd that Ira and Nina could not stop laughing — and he was offended: "When I really say something witty, you don't laugh!"

Novosibirsk: trees drawn on sand

One episode captures the LMP's science, its hierarchy, and her future heresy in a single frame. A research trip was planned to Novosibirsk Akademgorodok, to make contact with the only other collective working on semantic trees — Aleksey Gladky's group at the Institute of Mathematics. Zholkovsky, Martemyanov, and Nina were to go; the vice-rector for science, Kolshansky — no friend of the lab — struck out her name immediately: lab assistants do not get research trips. Martemyanov refused to go without her, and so it was Alik and Nina.

On the shore of the Ob reservoir — a strip of sand between the artificial sea and the railway — they sat drawing semantic structures, trees, erasing them, explaining: literally with a stick, on sand. Waves washed the structures away; passing trains interrupted the arguments; later Ivanov, perhaps not by chance, read them his poem with the refrain "They say nothing is stronger than a house built on sand." And it was there, walking with Gladky through the half-dead reeds, that Zholkovsky — unable to bear the Siberian mathematician's solemnity — tore out a reed taller than a man, knelt before her, and declaimed: "Nina, I give you this flower." She answered in kind: "Thank you, Alik, I shall return to the hotel and put it in a glass." Gladky himself later recalled the scene — and corrected the landscape: in those days both sides of the road (later built up into Morskoy Prospekt) were real dense forest with great trees.

But underneath the comedy, something had gone "mute" inside her — her word — that never fully found words: "Semantics is BIGGER than syntax; it cannot fit inside a tree structure." Gladky's Siberian trees pleased her more than LMP's "first-order predicate" notation, but the protest was already lodged. Zholkovsky recorded it in his samizdat notebook with a wit that stung for sixty years: "Nina — she's from the tundra; she doesn't understand trees."^[22]

Nine years later, when she needed an official opponent for her dissertation defense, she wrote to Gladky. His answer came by telegram, as terse as a proof: "Who else?"

1968: the defense nobody planned

Her kandidat dissertation exists because of an administrative misunderstanding — she says so herself. She had no intention of defending: career motives were alien to her and to her scientific supervisor (Melchuk, at the Institute of Linguistics' correspondence aspirantura); she was on leave with a newborn second son. But sector head Reformatsky asked her to assemble her papers so the sector's report could show a completed aspirantship — without a defense. She handed in a folder of loose sheets; the aspirantura office took it for a finished dissertation, scheduled the defense for the last session of the Scientific Council in June 1968, and informed her that cancellation was impossible.

So the dissertation Semantic Analysis of Text and Semantic Incompleteness was assembled in one month of madness — and one accident made it coherent. Sorting her examples on ellipsis (a topic Melchuk had given her, stipulating it be solved by purely syntactic means), she was passed in the corridor by Reformatsky, who glanced at her papers and remarked: "That's not ellipsis — that's a syntactic zero." The passing comment of a great man, she says, re-lit the whole material: she understood that zero, emptiness itself, has gradations; she separated syntactic ellipsis from semantic compression and derived several formally distinct kinds of semantic incompleteness — the phenomenon she would call the cornerstone of text semantics for the rest of her life.^[23]

The defense itself was a comedy with a knife in it:

• The recommendation letter. The official character reference from In-Yaz was half a page listing courses taken — no mention of her unpaid organizational work, nothing about the science. Director Filin kept raising his eyebrows: "First time I see such a recommendation." The truth: Kolshansky had given her a template and made her write her own reference; she had deleted every stock praise except "enjoys the respect of comrades" — and Kolshansky, signing it, crossed even that out.
• The two strangest reviews ever filed. One came from Tbilisi, from M. P. Chkhaidze — a man who had "sat out his seventeen years" in the camps and joined machine translation on return — and it argued, in effect: Nina L. is a good person. Sima Nikitina is also a good person. They wrote a good article together. Therefore... The second was from a random geologist who had read her note on the semantics of Time and declared that geology needed exactly this for dating rock strata. The hall was delighted; Filin's eyebrows went up again. A perfectly serious review by Apresyan apparently went unmentioned.
• The collapsing charts. Told only the night before that a defense requires demonstration diagrams, she stayed up building them; her husband rigged the watman sheets with rails and string — and during her talk they fell, one by one, each precisely as she finished displaying it. Her opponent Gladky, an impromptu speaker, had planned to work from chart one; it fell; he switched to chart two; it fell; when the third fell he "lost heart" — and was called to speak.
• Zvegintsev's question. Vladimir Zvegintsev — founder of MSU's own applied-linguistics department — asked her to write two lines on the board, Lermontov's "Harun ran swifter than a doe" and Pushkin's "now it is you who run from my caresses" — and state what valencies she would enter for run in her dictionary. Her answer satisfied him; he still announced he would vote against, on principle: she claimed a method for analyzing the whole Text, yet had demonstrated only a sentence. She answered that the method was the same for small and large stretches — and a full-text parse could not fit in a defense. History proved her right about the demonstration, if not yet about the method.
• Akhmanova's half-vote. As the vote began — doomed, everyone thought — Olga Akhmanova, head of an MSU department and council member, burst in late (she had been at her grandchildren's bedside: "I had to abandon... the enemas"), signed in, dropped her ballot, and tipped the result: a margin of half a vote FOR. In the corridor she said it plainly: "It was I who saved this girl." A decade later Zvegintsev, walking the corridor with her during another defense, said the thing himself: "We didn't yet understand the importance of such work, back then" — and years after that he was sending his best TiPL graduates to her laboratory at the VCP, calling its work "nearly the most interesting in Moscow."^[24]

The Melchuk question

No honest account of Leontyeva can skip her scientific divorce from Igor Melchuk — it structures everything after 1968 — but it must be told as she tells it: with love, grievance, and eventual one-sided forgiveness.

Melchuk gave her everything at the start: the first task, the red pen, the aspirantura, the topic of ellipsis-as-incompleteness — with the stipulation that it be solved by syntactic means. She obeyed, produced the rules for simple and complex sentences for the big sentence-analysis system he was then building with Lida Iordanskaya — and then quietly broke the terms. The phenomena "behaved indecently," she says: they crawled out beyond the sentence boundary and demanded the whole text. His next assigned topic, the analysis of "weak" prepositions, came with the same stipulation — call the relations whatever you like, but everything stays inside the sentence tree — and produced the same breach. Her deep prepositions were already textual.

Behind the technical quarrel stood a deeper one about what machine translation was for. Melchuk's circle — she is emphatic about their greatness: Melchuk, Apresyan, Iordanskaya, "the strongest men" who carried the new discipline — were building an elite theoretical science, Meaning⇄Text, whose object was the ideal well-formed sentence. She, the self-described serednyak (middling talent), kept glancing sideways at the homely, already-working "peasant" systems — the information-retrieval services that indexed and searched real, ugly texts. Her heresy, stated simply: the two are sisters, and neither survives alone — syntagmatic MT needed the paradigmatic world-knowledge that domain thesauri carried, and the thesaurus-driven retrieval systems needed linguistics. When she worried aloud at seminars, Melchuk answered with a joke that wounded: "Go to the States — there you can buy a ready-made retrieval system in any drugstore." Apresyan, seeing her interest in IPS meetings, said more roughly what he thought of the faces at them. And Melchuk's eventual verdict — "You are more an inventor of bicycles and an engineer than a linguist" — she records without anger, adding only: yes, and she was an engineer the Moscow school never needed. She asked him once for leave to go deeper into Meaning⇄Text itself; nothing came of it. She stayed, as she puts it, alone.^[25]

There was also a colder draught. Melchuk, emigrating via Vienna in the mid-1970s, somehow let a remark about her — drawn from her private letters to his wife Lida, her dearest friend, written precisely to keep an emigrant family connected to home — slip into print, and it was reprinted in the Soviet press. She learned of it through a phone call from Apresyan, relaying Melchuk's apologies, taken on the laboratory's single shared telephone in a silent room full of lab heads. She answered in monosyllables and says she understood little — but the VCP's own bosses had all read the note by then, and she spent the next days waiting for the summons to the carpet. It never came. The rift with her teachers, though — that lasted a decade, and it was she who later repaired it, "one-sidedly and late," in the manner of her own model: by deciding which parts of the content carry the meaning and forgiving the rest.^[26]

Gosplan: her first information system

When the American ALPAC report (1966) froze MT funding on both sides of the ocean, the LMP survived on khozdogovor work — contract research for state customers — and Leontyeva volunteered to carry the first and heaviest contract: with the Main Computing Center of the USSR State Planning Committee, Gosplan.

It is hardly a stretch to call this the first Soviet experiment in practical text understanding. Gosplan's torrent of incoming requests — "enterprise X requests two kilograms of steel nails," "allocate thirty pairs of felt boots to the Union of Composers" — had to be classified, indexed, retrieved. Her group wrote them in her semantic metalanguage, built semantic representations, applied transformation rules — all convinced, she insists, that they worked inside the Melchuk–Zholkovsky–Apresyan model. The arrangement inverted the institute's food chain in a way the theorists never quite forgave: the lab assistant's "alien" contract work brought the laboratory its main income, and under Rozentsveig's wing nobody objected to the heresy — yet.^[27]

The contract was also an education in Soviet economics that no textbook gives: quarterly reports bound as hardcover books, hauled through Gosplan's signature gauntlet; annual ritual dismissal and re-hiring of all contract staff; salaries half a year late; a contract officer — Diliya Stepanchenko — whose personal interests had to be served along with the state's; up to fifteen names on the contract, two or three real workers among them. After six years of it she told Diliya: no new contract by New Year 1971, and they walk. Diliya didn't believe it — they had endured seven years.

The acrostic report

Inside this grind sits the most famous episode of the corpus — famous because Zholkovsky later published it in his own vignettes, though her version is better. Worn down, she once begged Rozentsveig to let someone else write up a completed stage of the work — morphological analysis, implemented on contract by Sannikov and Kodzasov. Rozentsveig walked the corridors with her, arm around her shoulders, cooing his beautiful baritone about talents that must not be loaded with "trifles" — and then asked Zholkovsky to "help Ninochka."

Zholkovsky helped. He produced ten brilliant pages describing the finished algorithmic part — and the typist set them with the first letter of every paragraph in bold. The report, handsomely hardbound as required, was an acrostic, and it read: "Pay for this shoddy work quickly, and without squeaking." Nobody dared redo it — they were late already — and the collective judged it unsafe to let Nina go to Gosplan with it alone: they might beat her up. The entire core of the LMP — Rozentsveig at the head, Erastov and Mushanov now among them — rode over to escort her and waited downstairs while she collected the visas.

Upstairs, the comedy played out exactly as required:

• Every signer opened it: "What? THIS is a report?!" Her heart dropped every time — had they seen?
• The chief bookkeeper hefted the book and screamed: "For these miserable ten pages I'm to transfer fifteen thousand rubles?!" Nina bleated her prepared line — mathematical discoveries fit on one page, and occasionally win a Nobel Prize — and the woman turned the pages: nothing but morphological formulas and abbreviations, which to her looked exactly like mathematics. Diliya was already on Nina's side — the report was her salary too.
• The final boss, without opening it, hurled the precious report onto the dusty top of a cabinet already heaped with similar works — which, she notes, is what an information-retrieval system looked like to Gosplan.
• Downstairs: deliver the stamped papers to Rozentsveig, hold the laughter until the sunny street. Zholkovsky, the hero of the victory, squeezed out through tears of laughter: "After such a victory — only a restaurant. Everybody!"

At the Uzbekistan restaurant came the coda: nobody had any money. Zholkovsky stopped the whole procession and addressed the Chief solemnly — "Viktor Yulievich, it has accidentally emerged that none of us has brought money. And how are you... situated?" Rozentsveig struck an equally solemn pose mid-road, patted down his pockets with ceremony, found one breast pocket non-empty, leafed through the bills, and pronounced — slowly, several times — the sentence that deserves a monument: "While there is a wife, there is money!" (Their Anna Markovna, née Marshak, never left her husband in trouble.) There was beshbarmak, real Uzbek plov, a bottle of good wine — the boss treating, which especially delighted Shcheglov. And she learned her lesson: every subsequent report she wrote herself.^[28]

Stroymash: the customer becomes the scientist

On the first of January she moved — as she had threatened — to NIITEIStroymash, a construction-industry research institute, as head of the Laboratory of Information Retrieval Languages — and she took the LMP contract with her. The move inverted her position exactly as designed: she was now the customer, commissioning the LMP's continued semantic work, and for nearly twenty years — through Stroymash and then the VCP — the contracts she wrote kept her old laboratory fed. Department head Y. L. Pokras had courted her with a tripled salary and one honest sentence — "You're worth more than a senior researcher" — and she had warned him just as honestly: the day machine translation reopens anywhere, I leave. He said two promised years would already be valuable; he kept his word.

The Stroymash years (roughly 1971–73) produced real systems, not papers: a new type of domain thesaurus as an instrument of automatic indexing, a working retrieval system on top of it, and — with Lena Uryson — semantic analysis of technical headlines against the shared semantic dictionary begun back at the LMP. The programming was done by a collective in distant Gatchina, for which she wrote algorithms in such detail that the transfer to machine code was nearly mechanical — remote software engineering, in 1972, by post. A side project for the Instrument-Making Institute distilled Melchuk's lexical functions down to a corpus of "encyclopedic functions" — domain knowledge rather than linguistic knowledge — and implemented it by hand against classical indexing; the write-up lay unmoving for seven years at Berg's Cybernetics Council before she withdrew it and published elsewhere.^[29]

She left for the oldest of reasons: boredom and a ceiling. The indexing system worked and drew praise; there was nothing further to do; and Pokras's hints — "it costs you nothing to write a doctoral dissertation" — were aimed at his glory, not hers. She began, with a homesickness she fully owned, to arrange her return "to her linguistic family."

The road not taken: Apresyan's lab, and the Ne-Lebed

The negotiations were real, and she recounts them with guilt. Yuri Apresyan — "the kindest Yura" — invited her into what she calls the finest collective she ever knew, at InformElektro: three days a week, creative freedom, one small rider — "as under Melchuk." That rider caught. She had spent years arguing that semantics is older and larger than syntax; to work "as under Melchuk" was to un-say her whole objection. Still, it was better than "as under Pokras"; they walked the Arbat lanes and agreed; director Malinin signed her application at once — the first candidate, he said, without "difficulties."

Then the personnel department did not sleep. Its head, N. E. Lebed — known to the entire institute as "Ne-Lebed" ("Not-a-Swan," and the vignette is also Zholkovsky's, whose own hiring produced the man's immortal dictum: "We, of course, hire everyone — but we hire not ONLY everyone") — stalled her for four months. His interview questions are preserved like specimen slides:

1. Why do you accept such a drop in salary? — From lab head to senior researcher cost her some 1,500. Answer: work in her specialty mattered more. He didn't believe her.
2. And what interesting nationality does your husband have? — (The surname on file: Berlyand.) Answer: "That interesting nationality is Russian — his mother is a Russian woman, from Ryazan." — "And your husband's father's interesting nationality?" — "That interesting nationality is Jewish — and what has my father-in-law to do with it?" — "Mm, you know, there's such a resettlement of peoples going on now..." She answered, she says, that she had tried to get her husband to move — Yerevan wanted her — and he always answered: if you want, go alone; I'm not leaving here.

While the file stalled, her own reasons ripened. She knew herself: she would enter Apresyan's monolithic collective and begin to bazaar — why this way, not that? — carrying her idea-fixe into a tuned orchestra that hadn't asked for it. She knew the other side too: an "alien" collective inside an information institute survives badly (her own lab at Stroymash was nicknamed "the linguinists"); she had failed to push two candidates through her own personnel department — Lida Iordanskaya, fired from the Institute of Linguistics as one of the disgraced circle (Apresyan succeeded where she failed, for her and others), and her student Anya Gurevich, over whom she fought Pokras for six months until he shouted: "I am Pokras myself — and you accuse ME of anti—...!" To move to Apresyan now was to abandon those she had failed, for a freedom with a rider.

Three days before Ne-Lebed finally signed, the new institute appeared — as she puts it, "like the devil out of a snuffbox." When Apresyan called to say "all is in order," she answered not with joy but a mumble — "perhaps I've started to think" — and he hung up. She counts the hurt she did him the one real debt in the whole affair; his collective remained, always, warm to her.^[30]

The All-Union Centre of Translation: the only woman

The devil out of the snuffbox was the All-Union Centre of Translation — Vsesoyuzny Tsentr Perevodov, VCP — a brand-new institute under double subordination to the USSR Academy of Sciences and the State Committee for Science and Technology, gathering collectives for a new MT Department. A colleague she had once examined mentioned it in passing; she rode over to look; the English and German groups were already staffed, and the department head — Yuri Nikolayevich Marchuk — offered her the French group on astonishing terms: head of laboratory, free choice of model (Melchuk's, Kulagina's, Marchuk's own, or her own), free hiring, and his blessing on her cherished LMP contract.

She did not accept at once — she went to her teachers. Melchuk answered evasively and was plainly displeased (she was even bothered, she admits, by the echo of Marchuk and Melchuk). Kulagina approved — Nina had promised to begin by re-implementing her FR-2 model — and Apresyan supported Kulagina's view, before her own change of heart. She took the job: a jump into a cold well with many unknowns, and the beginning of the decade that made her.

The scale of the anomaly needs stating flatly. In the entire VCP — an academic institute of several laboratories under a state committee — she was the only woman laboratory head and the only non-Party member in all the management, down to the petty bosses. Marchuk's reasons remain mysterious even to her: she came from that laboratory, the one that had recently sunk his first doctoral defense. (The notebooks preserve the mechanics: Tolya Shaykevich, as scientific secretary, had shown at the board that "Doctor" Marchuk's computations contained errors below tenth-grade level; a year later Marchuk defended anyway — the MT problem had been, she was told, temporarily classified to let him pass — and in time he even invited his former wreckers to work for him.) Whatever his reasons, the arrangement held: his name shielded them from the Committee, and they were shielded from most of his world. When the FRAP group reached sixteen people, it still did not contain a single Party member — in a system where that was not an accident but an achievement.^[31]

FRAP: translation through understanding

FRAP — Frantsuzsko-Russkiy Avtomaticheskiy Perevod, French–Russian automatic translation — was the third Moscow attempt at the language pair, after Kulagina–Melchuk's FR-1/FR-2 experiments and Apresyan's group. She took it without knowing French, and defended that choice with a precedent: Bernard Vauquois in Grenoble knew no Russian, yet had run a Russian–French system to industrial scale. Not knowing the source language, she argued, actually modeled the dumb machine — the system could never rely on the linguist's intuition filling gaps the algorithm hadn't covered. It is either a defensible engineering stance or a rationalization; either way, it produced a different system.

FRAP differed from the standard Soviet MT model in ways that were her whole career in miniature:^[32]

• Graphematic analysis first. A language-independent pre-pass — dates, formulas, abbreviations, document furniture — before any linguistics began.
• Deliberately imperfect syntax. After a year they abandoned Kulagina's morphology for a simpler, sturdier one (participles decoupled from verb classes; later, at ISK RAN, Zhanna Anoshkina implemented Russian morphological analysis directly on Zaliznyak's Grammatical Dictionary — the first such implementation). The syntactic analyzer allowed breaks and ambiguous links — because for real technical texts the perfect tree does not exist; in Grenoble, Vauquois's own group could fully parse only ~30% of real input. Where the standard model rejected what it couldn't resolve, FRAP graded the residue and passed it upward.
• Semantics top-down, not bottom-up. Instead of completing syntax and then attempting a semantic pass that inherited all its failures, FRAP's semantic stage — working on her own metalanguage, not the Meaning⇄Text notation — could repair missing links using domain knowledge. She drew the direction of travel explicitly: not Meaning→Text, but Text→Meaning.
• The text, not the sentence. The semantic record she built was a continuous, partly formalized structure over the whole document — the "primary textual semantic representation." Translation was conceived as one function of a larger information system: summarize, index, abstract, infer, translate. Analysis and synthesis were declared asymmetric: compressing a text is a semantic operation, not a syntactic one.
• Melchuk's principle, kept. Dictionaries were fully separated from algorithms — the contract dictionary teams at the LMP and MSU worked without ever knowing which programs would consume their output. Even heresy had commandments.

The lab she built was itself an argument about how science should run. The staff: Sergey Nikogosov, the lead programmer — a mathematician (semirings) squeezed out of a closed institute, rejected by Rozentsveig's personnel filter for writing his biography "non-standardly" ("family of Russified Armenians"), snapped up by Marchuk in a signature; he read the whole MT literature and pronounced it "unexpectedly empirical," then ran seminars on PL/1 and Assembler for her linguists. Olga Malevich, defecting from Tikhomirov's lab out of boredom with word-for-word translation, created the internal language OLYA on which Leontyeva's linguists programmed their own analysis modules — Sokolova the syntax, Kudryashova the semantics — and debugged them themselves: a computational-linguistics lab in which the linguists were the engineers. Sveta Elnitskaya, who opened mornings by reading Winnie-the-Pooh in beautiful English for ten minutes — "drop the nonsense, Nina, let's do work." And through the personnel sieve — Zvegintsev himself interceded — the last-minute substitutions Masha Zhitomirskaya and Mira Bergelson, two of the best TiPL graduates, in place of the officially distributed "good surnames."^[33]

There are two famous episodes of the building years. The first is small: at the founding meeting, asked who would program the French system when Tikhomirov's programmers served only English and German, she answered that newly hired N. Rebetskaya wanted to learn. Ubin asked, to general hilarity, "And where were you looking when you hired her?" — "In the eyes; I liked her." Two months after giving birth, per the rules of the time, Rebetskaya was back writing simple programs. The second is the Pericles–Michelet affair, which deserves its own paragraph: presenting FRAP at an LMP seminar, she took as her example an authentic subheading from a real arrived-for-translation text on microelectronics — "L'alimentation de Périclès-Michelet." Rozentsveig shouted from his seat: "Nina, that's not French!" She answered that whatever it was, it stood in a real article and they were obliged to translate it. The room proposed many renderings — nutrition, supply, the feeding of... — and the truth surfaced only six months later, when one of her French staff finally read the entire article to the end: it concerned the transfer of electrical power from a place called Périclès to a place called Michelet. No sentence-level model could solve it. Only the whole text could. She could not have invented a better vindication of text semantics if she had forged it.^[34]

The communique: a quiet campaign

The war on FRAP is documented in the notebooks with a care that suggests she knew exactly what she was recording. The short version: an official evaluation — she calls it the "political communique" — was prepared in advance of the seminar it purported to summarize; a closed meeting at the State Committee, organized by Kulagina together with Rozentsveig, decided FRAP's closure before the discussion she was invited to had even occurred. She learned of the hidden seminar only because Zhenya Lovtsky — head of the AMPAR system and Marchuk's press secretary, well-disposed to FRAP — told her.

Her offenses, itemized across the notebooks, were: bringing informatics into MT; not accepting Melchuk's definition of Meaning; doing morphology not by Kulagina (and finally by Zaliznyak); not joining Apresyan; inventing "ugly words" like databases and artificial intelligence; and proposing translation "through some incomprehensible databases." The communique asserted flatly that no semantics was happening at the VCP at all — which meant, as she points out, that its signatories had neither heard their reports nor read their publications. A senior official banned the very term "information-translation system" — after which Marchuk applied it to his own word-for-word systems.

Two small pushes against the avalanche deserve recording. Grigory Tseytin — the Leningrad mathematical-linguistics head, a professor at twenty-five, himself wired into the Committee through programming — learned of the intrigue and managed to postpone the rout by a year. And Vauquois, visiting the VCP and hearing of FRAP's design, said — "ça reste dans la famille" — that he liked even its name: frapper, Russian udar, a blow.

And then the episode that shows who she was, better than any defense. After FRAP was delivered and closed, the Committee sent Apresyan's group's report — covering the period to 1985 — to the VCP for review. At the Scientific Council, Marchuk asked who would write the response; silence; then Vlasov — the informer who counted the whole LMP "enemies of the people" — leapt up: "I am ready, Yuri Nikolayevich!" There was, she says, no choice: "No, Albin Nikolayevich — you do not know that collective's work. I do — I took part in the delivery of their system. Let me." A Vlasov review would have been a rout and, signed by Marchuk and Ubin, a firing offense for the old friends; three of them — Martemyanov, Shaykevich, herself — would likely have walked out of VCP the same week. Her review was positive, honest, and signed with her name: it praised the work, noted the renamed-syntax character of its semantic structures and that ETAP-1 was not an industrial system (true, and known — no industrial MT existed), and omitted the word theory, which the Committee hated. Apresyan's collective worked on undisturbed; his people's warmth to her never wavered. She later phoned him to claim a duelist's right of reply — the second participant answers the shot — and came to his group and simply talked; at Sangaste, later, they drank together and she danced with all of them, including Yura.^[35]

The smaller persecutions

The notebooks also keep the petty ledger, because that is how institutional wars are actually fought:

• The title that never came. Twice her file for senior-researcher rank was submitted; twice the council quorum mysteriously failed when Ubin — twice her superior — presided. She got the professorship eventually from RGGU, after eighteen years of part-time teaching.
• The medal. When the directorate received a stack of Veteran of Labour booklets, they were distributed to everyone — even lab assistants hired the previous week. Ubin came to their corner himself, embracing and kissing each of her "French birds" — and handed her alone nothing. Only at MSU, decades later, did a supervisor quietly obtain the document for her. She understood late that she had been robbed of the attached benefits for years.
• The impossible deadline. After the first version's delivery, the plan allotted seven years to the next; a year before the (five-year) mark, an order arrived demanding FRAP in one year — with Nikogosov dead and only Lyonya Korostelyov left on programming. Tikhomirov offered to save them at the price of amputating "all wings and tails" — syntax, semantics — leaving word-for-word. Her answer was categorical: better an incomplete system than like all the rest. She prepared protest papers for the Committee, needed the director's visa — and then, in the famous episode, preoccupied and hypertense, walked into the wrong washroom and found herself face to face in the mirror with Ubin's widening eyes. By the corridor she understood — men's room, wrong floor, her cropped hair and trousers. The strength to walk into his office with the protest never came; she went home saying nothing. "What 'all is lost' means, I already knew."
• The final acceptance, all the same: a ~30-member State Commission took FRAP-2 in 1985–86. At the demo on borrowed hardware, commission members typed French sentences — Gorodetsky entered a little poem — and received "acceptable" translations; Korostelyov dragged the wide-format printout out with his foot because the battered machine couldn't. To the next day's report she brought a piece of string — to hang on the diagram where the unfinished module belonged: the system's point was that a missing module didn't stop it; it simply called the next one.

INION, ISK RAN, and the lean years

She left the VCP the day she walked out toward Profsoyuznaya metro — and collided, literally, with an old acquaintance: Robert Mdivani (to her, Vakhtang — husband of the pianist Marina Mdivani). Told they were being dispersed, he said: come to my department at INION as chief researcher, run the thesaurus. Two years of salary-and-talon existence — one half-lunch voucher and one vodka voucher monthly (the vouchers survive; her family didn't drink) — trying to teach his thesaurus-linguists her semantic dictionary, with little success.

Then the Institute of US and Canadian Studies — ISK RAN — the "elite institute," where there were suddenly personal computers and she sat at one until the guards learned to call her "the slave of Aura." There, with Kiev programmer Ponomarenko, she built the color-coded multilevel dictionary editor; there her "invisible collective" of students and young enthusiasts assembled; and there the directors dropped in evenings to see the wonder — director Arbatov admiring the bound Zaliznyak volume, deputy minister Kokoshin studying the semantic dictionary's dialogue and crying: "You could build a dialogue between Gorbachev and Bush this way!" — and she bit back the retort about the drawing in the middle where the naked girls dance, because such fantasies were what directors had. Yudina brought Duma and government people; commissions came; and every six months the particular customer was fired, promoted, or vanished with all their documents. The hardest of the lean nineties.

And the doctorate: pressed to write it because her group contained, besides herself, exactly one candidate of sciences, she wrote at night on the institute computer for half a year and went to defend in Kiev — where the council, assembling five extra doctors of philology, announced at the last moment they would accept the work only as technical sciences. She rebuilt it on the spot as the information model of an entire automatic-text-understanding system, and defended as Doctor of Technical Sciences, 1990 — the title whose irony she never stopped savoring: the philologist who refused a technical institute as a girl became, in the end, a doctor of it.^[36]

Moscow State at last: RUSLAN and the taken system

The final seventeen years — NIVTs, the Research Computing Center of Lomonosov Moscow State University, April 1999 to the end of 2016 — began with the thing she had wanted since school: to work at MSU. The "invisible collective" became the ROSSIYA information system — the ROSS general semantic dictionary at its center, fed by Yudina's rich media-text databases — the thing that should have grown into POLITEXT, her full text-understanding model. Instead came the bitterest loss of the whole record: the young programmers they themselves had hired walked away with the system and began selling it to the very customers the collective had found at ISK RAN. The physicist she had chosen to lead them — Yudina's husband, Sergey Zhuravlyov — dismissed the brightest of her trained students one by one, keeping the quiet executors, and consulted her hourly while doing it; when he dismissed chief programmer A. Sokirko after a working dispute, she walked out the same hour into a neighboring laboratory. The tools stayed with the new owners. Requests for her own morphology results came back as raw dumps.

ROSS was renamed — by the system's new proprietors — into RUSLAN, the Russian dictionary for text analysis; under Dobrov and Lukashevich (who had done her student practice inside FRAP, and who put into the IPS the maximum of linguistics the invisible collective had built) the project flourishes to this day, and Leontyeva's recorded feeling is the characteristic one: "I am glad my labors were not lost." Leaving NIVTs she gave away the RUSLAN lexical content — some 12,000 dictionary entries plus the grammar of semantic elements behind them — to outside collectives; her former chief editor S. Yu. Semyonova carries it on at INION.

The MSU years also held her happiest teaching. For five semesters she ran free special courses on text semantics — for structural and applied linguists, then for the applied mathematicians two floors up, who called themselves "the primates" (from prikladnaya matematika): six enrolled the first semester, twelve the second, twenty showed up. Coursework topics were snatched up faster than she could post them; students went on to her systems. The fourth course was over-subscribed enough that she refused — each student needed near-personal work. The semester the timetable simply "forgot" her course, she came on the first of September only to say — or to hear — thank you.^[37]

The account ends in the flat tones of a report: the 300-page textbook Automatic Text Understanding: Systems, Models, Resources (Academia, 2006) — the only textbook on the subject, evaluated by the points system exactly as a one-page abstract; RGNF grants thinning after her seventieth year; the RFBR application that two of three experts — she knew two, both famous — had already scored at the maximum, refused anyway; A. I. Gerd of St. Petersburg learning of the refusal from her in a corridor, darkening, saying "so they don't value me to that degree?!" — the last words she ever heard from him. When even the core NIVTs staff went to half-salary, she resigned altogether. Pension: a little over twenty thousand rubles. "For freedom, I decided, that would suffice — though the head asks for work, even 'for the drawer.'"

The people of the notebooks

Notebook 3 is her gallery — "records and reviews, at first businesslike and even commissioned, but becoming steadily more personal." She warns the reader herself: "of my closest friends I cannot write; I mention them only in passing or in joking little verses." The portraits below are assembled from that notebook, quoted and weighed as her testimony — which is what they are.

The men who opened the door

Vyacheslav Vsevolodovich Ivanov — the great man on the staircase. He set the LMP its task (the elementary meanings) and later, at the family dacha, became part of the corpus's best story — the fire. He appears throughout as a benign remote constellation rather than a daily presence; her gratitude to him is permanent. When she sent him her father's inscribed book — through Panov — it closed a circle: her father the folklore collector, and the scholar who read him with respect.

Viktor Yulievich Rozentsveig — "Vyu." Her "second father," the organizer of everything: the Association for MT with its 200-member card file, the first All-Union conference, the MT Department, the journal, the lab itself — run pro bono, without status or salary, because he knew he could not lead the science himself and chose instead to protect those who could. He had washed bottles in French restaurants and finished the Sorbonne; he could switch to any European language mid-conversation; his motto was "put it on the table" — the algorithm, the article, the program, delivered on time. And yet the same notebooks record the ambiguities: the surname campaign (some must change their names to "more euphonious" ones — Ira Tseytlin reverted overnight to her maiden name Mosalyova; Uryson and Gurevich were saved only because Gindin told Kolshansky the Gosplan contract could not go on without them); the quiet siphoning of her best people; the hidden GKNT seminar; Edik Korolyov driven toward his second, fatal heart attack. And yet — the same man who forgot their appointment one evening thereby sent her home hours early, to find her mother sitting three hours in a cold bath unable to climb out. "His forgetfulness once saved my mother's life," she writes, and means it. They saw him off to his children in America "in the best possible way" — Ira and Lena hauling his books to the post office for a month. The final verdict is hers and it is deliberately unresolved: he was Papa Carlo to giants, and a politician, and her benefactor, and the man who signed the communique. All four at once.

Igor Aleksandrovich Melchuk — "on the dawn of my misty youth he eclipsed the whole white world with the red color of his hair." The idol, the red pen, the hikes, the thesis, the quarrel, the silence. She is painstakingly fair: she notes that he invented the discipline's depth, that his verdict on her ("more an inventor of bicycles and an engineer") may be the correct one, that she argued with him "not understanding his level," and that his shadow ran the LMP's science because Rozentsveig couldn't. The notebooks also preserve her one genuinely "independent" scientific lineage claim: she insists her semantic relations (ESO) are not identical to his elementary meanings (ES) — "at least in the LMP approach, all the ES were objectified, one-place meanings... fundamentally opposed to my two-place ESO." It is, under the affection, a claim of priority — recorded, not resolved.

Yuri Derenikovich Apresyan — "the kindest Yura." The vignettes are physical: leaping off a moving train at Leningrad Station to pick up a fallen old woman, bag and all, and catching the car; cutting snow cubes for the Khibiny fire-pit so precisely that she understood why his and Melchuk's dictionary article on bonfire was written with such love; feeding an entire impoverished Tbilisi household — three grandchildren, parents, two grandfathers, two guests — off two handfuls of hercules he had insisted on packing against her advice; and, when her Soviet passport was challenged at the airport on the grounds that "no such country exists," charging back through passport control: "And how did you let us in — from a nonexistent country?!" Marina Glovinskaya once asked, slyly, which man in the hike she liked best, and guessed "Yura, probably?" — "No," Nina answered, "not him. Too good." Soon after, Marina married him.

Grigory Samuilovich Tseytin — the Leningrad mathematician, professor at twenty-five, her most valued scientific sparring partner and the man who quietly bought FRAP a year of life. He stenographed every MT talk in 1959 — and, she discovered, stenographed tourist songs the same way, prompting lyrics at campfires while visibly waiting for the science to resume. Their argument was perpetual and specific: he folded every level of analysis — morphological through denotational — into one grand graph; she objected that a denotatum cannot live in a graph ("will you tie the frog in by its foot?") and that each level obeys its own laws. His one question — "And what sets your semantic graph in motion?" — drew the answer she built everything on: "The Question. Incompleteness. The Unknown — explicitly marked." When staff starvation finally drove his own people to paid work, he took offense, left for Europe, and — her word — "literally disappeared." She records her debt: it was Tseytin who first explained the frame to her, over a picture of a cow grazing.

Yuri Aleksandrovich Shikhanovich — the conscience. Punctual to the point of pedantry, he kept an eighteen-field grading sheet and examined students until midnight; later he became one of the bravest of the Moscow human-rights circle, arrested, and — she records this herself — a man who once took a call summoning her to Lubyanka at eleven the same morning. She went; nothing came of it; but at RGGU years later he asked her — she was working under Ubin by then — to "give him MY regards," and she never delivered, not understanding the freight. She writes the apology now, decades on: "if it was slander, forgive me in advance." He went, she says, "to the very end" — and she, who knew herself a "middling person," deliberately steered off roads whose ends she could not have survived.

Zoya Maksimovna Shalyapina — arrived as a third-year student on practice, nicknamed "Melchuk-in-a-skirt," and went on to the Japanese–Russian work and to a vision of the discipline far more total than her mentor's — whole-language architecture, tiers upon tiers. Nina, formally her supervisor, admits she read only the morphology chapter closely: "she had a worldview of her own that needed no supervision." She died on 19 October 2020 — the day after Melchuk turned 88, a number Nina notes encodes two infinities.

Sergey Iosifovich Gindin — the kamerton, the tuning fork: the colleague beside whom you hear your own false note. It was Gindin who, on his own initiative, walked into Kolshansky's office and saved Uryson and Gurevich; Gindin who later chaired the RGGU department where she taught her happiest courses; Gindin who wrote her 2006 profile in Russkiy Yazyk — one of this article's chief external sources. The tuning-fork portrait is also her statement of method: a science survives on the quiet people who correct its pitch.

Yuri Semyonovich Martemyanov — "Udubasuka": three shared terms of service (LMP, VCP, RGGU), the corridor-wide graph-paper table, the hazel-branch bows on the Lithuanian hike, the kayak-flotilla captain on Seliger running a worn black blanket up a mast into the storm, the man who dove into the Neman after a swimmer he had begged not to go — "I shall be obliged to swim too, and my heart is not good" — and went. He gave up his own Novosibirsk seat for her without telling her. His verdict on himself, via her: a figure "always belonging to the Future."

Elena Mikhaylovna Smorgunova — secretly on ty with Ulitskaya, Chudakova, Gorbanevskaya; friend of Kovalyov, Velikanova, Babitsky; reader of the names of the Gulag dead at the Stone of Remembrance; mistress of Old Hebrew, Old Slavonic, and every inscription on every old icon; guide to the Kremlin's treasures. A whole humanities faculty in one person, discovered by her friends only by accident, in pieces.

The Kiev circle — Misha Beletsky and Irina Platonovna Beletskaya-Sevbo, whose kitchen-table conversation produced the corpus's most quoted hostile review: her host's father, the elderly Professor Sevbo, surveyed the assembled linguists and pronounced doom — "You have no prospects: your men are all so small." It was the Beletskys' Kiev whose structural-linguistics department was smashed in the purges; Irina went on toward her triad of Time — Information — Consciousness, answering Nina's questions about her current work with magnificent disdain: "My theme is deafly indifferent to you — where meanings come from, where they leave their traces — the text has nothing to do with it."

The Petersburg line — Sergey Yakovlevich Fitialov, of the Kantorovich–Markov lineage, poet and romantic, who asked her to read French texts aloud while he listened very carefully, and who, admiring nothing in her flat, nonetheless sent her two bundles of his iron bookshelf rods — which hung in her bathroom and kitchen for years. And his student Viktor Zakharov, keeper of the lineage's memory.

Natalia Pertsova — "Natasha the redhead," judged frivolous by the Melchuks at a Black Sea gathering — and who became one of the deepest researchers of the group: semantic preprints on the cutting edge, then massive Khlebnikov scholarship, then the rescue of MSU's anthology Poetry of MSU from Lomonosov to..., of which she lived to complete volumes 7 and 8 — outwardly identical books containing, Nina insists, an enormous investigation. Life was wiser than the Melchuks.

The foreigners in the notebook

She kept a separate shelf for encounters with the giants outside:

• Roman Jakobson — heard at the Institute of Linguistics and at a packed In-Yaz lecture: his "shifters," and his example that may be the seed of her life's topic — "Can you imagine a husband and wife speaking to each other at home in complete sentences?" Melchuk's first assignment to her — incomplete utterances — may trace to that room.
• Viktor Shklovsky — heard once, at a Writers' Union memorial evening (Khlebnikov's, she thinks), thundering that the poet was "NOT one of those ready to sleep in a matchbox!" She notes the later, private use she made of him: she read his knotted prose for facts — the hungry Petrograd years — because her generation was "fed only propaganda and tore at any hint of the truth."
• Yuri Lotman — one freezing dawn on the Leningradskaya platform, −30°C, waiting for the Sangaste train: a small figure pacing for warmth, "a copy of Einstein's portrait." She introduced herself mid-stride, praised his latest NTI article, and got his squinted reply — "And isn't it crazy?" — before asking him her one prepared question, about scientific versus literary prose.
• Charles Fillmore — "Chuck" — at the Montreal conference she reached on a Soros grant in the lean years, staying in the Melchuks' famous Montreal basement alongside Ferenc Papp's family. On Fillmore's 60th birthday morning she showed him her semantic-dictionary entry schema, prepared in English with examples; he scanned it and said, "O, it's an ambitious Dictionary!" — and she retreated in shame until Anya Gurevich explained that in English ambitious is praise. "So perhaps I missed my fortune," she writes, and immediately forgives herself: "Perhaps that is right — new things should be proposed by great and tall men, not by some..." — she leaves the sentence where it belongs.
• John Sowa — noted in a single late addendum: he implemented their dictionary in Pittsburgh; she had sent him all their materials; "so he deceived me — like our Duma." One line, and the whole nineties in it.
• Maria Lvovna Mayenowa and Elżbieta Janus — the Polish thread, which deserves its own section below.

Poland, 1980: the street of Okudzhava

Her one trip to a "democratic" country became the political hinge of the whole memoir. On 1 October 1980 — the first day of the academic year, in the Solidarity autumn — she walked to Warsaw University through chains of police vehicles, waving off the megaphones addressed to "Pani!" because she knew no Polish and could answer none. During the conference the University was granted its free status. And then Mayenowa sent her to Kraków — both women knowing it was forbidden — where she walked a street of open-air painters, heard the trumpeter from the four sides of the tower, heard Okudzhava's voice in her head, and wept: "It was the most unforgettable day of my life — there I understood for the first time what Freedom is." She claims, proudly and against fashion, to have heard both her "gods" — Okudzhava and Vysotsky — before her closest friends did, and to have hidden them in her soul even from her sister. Poland "legalized her right to internal freedom" — and she credits it, explicitly, with the resolve to take independent work under Marchuk rather than the beloved clan: the right not to follow the crowd.^[38]

Marriage, family, and the eleven apartments

The fourth notebook is the family record, and it changes the biography's aspect — the career ran on top of a full, difficult domestic life.

Andrey. She met her husband on her own birthday, 21 November 1962, at a party in Alexander Zholkovsky's apartment — Andrey was Alik's cousin, a FizTekh graduate and a Moiseyev-ensemble dancer, known then as "Benya" after Babel's Benya Krik. Crossing the Metrostroyevsky bridge he took her arm — and did not let go for a week. They married; she learned his real surname (Berlyand) at the registry office. The marriage lasted twenty-one years and two sons — and her account of it is by turns tender and merciless, as her accounts are.

The children. Son Petya was born at five in the morning on 1 January 1963 — hours after Melchuk's housewarming party at which the Magnificent Seven had performed the "who here is fully Russian?" scene. A second son followed. She insists, in a move that is pure Leontyeva, that the domestic details built her theory: her children's utterances fed her model of text understanding — the toddler's question she renders as "Information is what you wait in line for," and the still more startling child-logic she records as "Information is non-verbal, but the transmission of information is verbal." She meant it literally: the notebooks claim her model of automatic text understanding grew step by step out of listening to how real people — children — actually compress meaning.

The Troitskoe fire (c. 1975). The single best anecdote in the corpus. At the family dacha, 90-plus-year-old nearly-blind Aunt Anyuta walked out of the burning house "like King Lear," holding only her house-book — and inside burned the manuscript of Vyacheslav Vsevolodovich Ivanov that Nina had been hiding. From it survived a single scorched line: "It was frightening to live — but it was simple to live." If the corpus has an epigraph, it is that.

The divorce and after. After twenty-one years they divorced; there were eleven apartment moves across the family record; and the famous "Four Vs" chapter — the four suitors of later years, filed as V-1 through V-4 — conceals, as its quiet payload, V-0: the school friend she loved platonically, the poet Vitya Shnitke, whom she had refused and never stopped regretting. It is the one admission in the entire corpus where the analyst of incompleteness leaves the most important thing unsaid — and then, being Leontyeva, writes a footnote saying so.

Her ideas

It is worth pulling her scientific program out of the chronology and stating it plainly, because in its stubbornness it turned out to be early, not wrong.

1. Semantic incompleteness is the central phenomenon. A sentence's semantic representation is always incomplete — deficient, ambiguous, wrong — when built in isolation; only the text around it, and the world behind the text, complete it. The 1968 dissertation made this her cornerstone; she never moved off it. Jakobson's joke — husbands and wives do not speak in complete sentences — is its slogan; a child's "information is what you stand in line for" is its proof.
2. The object is the Text, not the sentence. The standard model climbed bottom-up — morphology, syntax, then semantics — and broke against real technical prose (she loved the Grenoble datum: ~30% clean parses). Her method: grade the residue, let the semantic level repair it, and work on a continuous representation of the whole document — the "primary textual semantic representation."
3. Analysis and synthesis are asymmetric. Compressing a text to its informational content is a semantic operation; hence translation proper is just one use of a deeper engine — summarize, index, retrieve, infer, answer. FRAP was conceived as part of an information system, not a dictionary that walks. Hence her lifelong battle to be allowed the very word information in an MT shop — banned to her, later adopted by Marchuk for his own systems.
4. MT is informatics, not only linguistics. Databases, knowledge bases, thesaurus-driven domain models — she insisted on them when they were "ugly words," and insisted the two disciplines were sisters. Her information intermediary language (IYaP), sketched from 1967 on, was deliberately non-ideal: formal in structure, tolerant of ambiguity and incompleteness in content — because only a model that tolerates them can explain why different readers, or one reader at different times, extract different meanings from one text. The meaning of a text is, for her, a user-relative extraction — which is what her November-2020 self was still coding.
5. The experiment is everything. Her repeated indictment of fifty years of "MP i PL" is not that the theory was bad but that the field experiment — whole real texts through a whole model — was never permitted: not by Gosplan, not by the GKNT seminar, not by the institutes that took her system. Her final word on it: "One person cannot build a whole system, as one engineer cannot build a whole airplane."

The notebooks themselves

The corpus is six notebooks deposited at Prozhito — the public archive of personal diaries — which thanks the author herself for donating the files; the text is published in her own edition, prepared for upload by Alexey Senyukhin. She wrote them, she says, in her first retirement years, "instead of a standard boring report," on the principle her programmer friend gave her: writing about yourself is necessary for two readers — you, and your grandchildren.

The architecture is deliberate:

• Notebook 1 — childhood and formation, narrated in the third person through Grandmother Narina, the invented-name persona: a folklorist's frame that lets an eighty-year-old watch the little girl Alya as if she were a character in a collected tale.
• Notebook 2 — the career: the working memoir from 1959 to the last official post, "objective material accompanied by personal evaluations."
• Notebook 3 — the people: commissioned portraits that keep breaking into personal ones.
• Notebook 4 — the family: husband, sons, eleven apartments, the fire, the parents' generation.
• Notebook 5 — the November 2020 diary: her 84th-birthday month under COVID lockdown — Echo of Moscow in the night, blood pressure, the Ossetian-pie delivery farce, son Petya as her support, Lida on Skype, Melchuk raging that Trump had lost, and — still, at 84 — her semantic space and her own conception of artificial intelligence, incorporating Beletskaya's "Universal Knowledge" level. She had declared 84 (21 × 4, her birthday's arithmetic) her last year — "but an order from above said live on; my affairs aren't finished." The little white rat figurine — a 2008 gift from her FRAP team — still hung above her desk.
• Notebook 6 — the verse: dedications to Melchuk, Paducheva, Yudina, Zakharov, Akhmadulina, Kirsanov — and to herself. Its opening quatrain sets the accountant and the poet, her two inheritances, as "the difference of two elements."

The corpus is also, she notes, itself a proof of her theory: a text so "loose," so riddled with deliberately retained incompleteness, that only a reader's own semantic space can complete it — which is what she always said a text was.

Timeline

- 1936 — born 21 November in Naryan-Mar, Nenets autonomous okrug, to folklorist Nikolai Leontyev and accountant Raisa Zarubina
- 1941–42 — war begins; evacuated with mother on the last Litfond train to Stalinabad (Dushanbe); father's legs lost to his wound
- 1947–54 — postwar Moscow province: Burkovo, the "bread sin," the Kostino barracks school beside a German POW camp
- 1954 — enters Moscow State Pedagogical Institute (Potyomkin); Panov's phonology course; the LeSok wall-newspaper articles
- 1956–59 — the staircase: Melchuk recruits her; evening faculty; geological expeditions in between
- 1 May 1959 — becomes employee no. 1 (lab assistant) of the new Laboratory of Machine Translation, MGPIIYa
- 1962 — marries "Benya" (Andrey), met at Zholkovsky's birthday-party apartment
- 1963 — son Petya born at 5 a.m. on New Year's Day
- 1967 — begins leading the Gosplan contract (IYaP work); the seven "Gosplan years" begin
- 1968 — Candidate of Philological Sciences: semantic analysis and the semantic incompleteness of text; second son born
- c. 1970 — leaves LMP for NIITÉIStroymash as lab head — becoming LMP's customer rather than its employee
- mid-1970s — joins the All-Union Centre of Translation (VCP): the only woman heading a lab, and the only non-party-member among the junior bosses; builds FRAP
- 1980 — FRAP running; Solidarity autumn — the Warsaw conference and Kraków; "there I first understood what Freedom is"
- 1983 — FRAP-2 strategy published; the Kommyunike campaign against the system peaks
- 1985 — FRAP-2 delivered after Tseytin's purchased year
- 1990 — Doctor of Technical Sciences; divorce-era years; INION and ISK RAN work: POLITEXT, ROSS
- 1990s — Soros-grant travel; Montreal; the "ambitious Dictionary" shown to Fillmore
- 1999/2000 — joins the Research Computing Center (NIVTs) of Moscow State University; RUSLAN; teaching at RGGU under Gindin
- 2006 — textbook *Automatic Text Understanding* published; Gindin's profile
- 2016 — leaves MSU (31 December, per ISTINA); in the first retirement years begins the six notebooks "instead of a standard boring report"
- 2020 — November diary: her 84th-birthday month in lockdown, still coding her semantic space
- 2024 — still writing records for the corpus at 87

Check your understanding

• What does it mean that Leontyeva was "the first employee" of Soviet machine translation — and why was her actual job title "lab assistant"?
• What is "semantic incompleteness," and why did she consider it the central phenomenon of text — not a defect?
• Why was FRAP's use of semantics "almost secret" inside VCP, and what did the Kommyunike campaign claim?
• What happened in Kraków in 1980, and what does Leontyeva say Poland gave her?
• What are the six notebooks, who are they written for, and why does she narrate her own childhood in the third person?

• On names and the Narina frame. The birth-registered "Narina" story is her family's own account, told in Notebook 1; the change to Nina is likewise hers. We follow her practice: Nina in the narrative, Narina only where the persona itself is the subject.
• On memory. These are memoirs written at 79–85 and later, sixty-plus years after the earliest events; we treat colorful detail as hers, corroborate the load-bearing facts (degrees, dates, institutions, systems) against ISTINA, MT Archive, Vestnik RGGU and colleagues' memoirs where possible, and flag the rest as recollection.
• On the husband's name. The corpus is itself inconsistent about how much to conceal; the personnel-department scene supplies "Berlyand." We print it once, there, and otherwise keep her own habit of "Andrey."
• On unfavorable portraits. Vlasov, Ubin, Ne-Lebed, Pokras and others appear here as she saw them; several are deceased; the episodes are recounted as her testimony, not adjudicated fact — with the exception of the suicide of Olya Frolova, which we include because she included it, and because the denunciation culture it belongs to is part of the record.
• On Marchuk's doctorate. Her account — that MT was briefly "classified" to let his defense through, then declassified — is a reported rumor within her own memoir; we report it as such.

References

• Primary: N. N. Leontyeva, "Zhili v epokhu mashinnogo perevoda — Tetradi odnogo spetsialista epokhi peremen" (We Lived in the Era of Machine Translation — Notebooks of One Specialist of an Era of Change), six notebooks, Prozhito corpus (prozhito.org), author's own edition; prepared for upload by A. Senyukhin.
• Leontyeva, N. N., "Kak ya okazalas' v mashinnom perevode" (How I Ended Up in Machine Translation), Vestnik RGGU no. 6/9, series Yazykoznanie, vol. 11, pp. 28–37, Moscow, 2009 — the published memoir fragment.
• Leontyeva, N. N., "Automatic translation through understanding and summarizing," COLING-82 proceedings — FRAP described in English, signed from the All-Union Centre of Translation (MT Archive).
• Leontyeva, N. N. & Nikogosov, S. L., "Sistema FRAP i problema otsenki kachestva avtomaticheskogo perevoda," MP i PL no. 20 (1980), pp. 57–78.
• Leontyeva, N. N., "Sistema frantsuzsko-russkogo avtomaticheskogo perevoda (FRAP): lingvisticheskie resheniya, sostav, realizatsiya," MP i PL no. 21 (1987), pp. 6–25.
• Leontyeva, N. N. et al., "Obshchaya strategiya lingvisticheskogo analiza v sisteme FRAP-2," International MT Seminar theses, VCP, Moscow (1983).
• Leontyeva, N. N., "ROSS: Semantic Dictionary for Text Understanding and Summarization," Meta: Journal des traducteurs (2002), doi:10.7202/003464ar.
• Leontyeva, N., "Is There Any Understanding in Machine Translation?" in Understanding by Communication, Cambridge Scholars Publishing (2012), pp. 138–160 — her principal English-language statement.
• Leontyeva, N. N., Avtomaticheskoe ponimanie tekstov. Sistemy, modeli, resursy (Automatic Text Understanding: Systems, Models, Resources), Academia, Moscow, 2006 — the only textbook on the subject.
• Rozentsveig, V. Yu., "The work on machine translation in the Soviet Union," MT (1958) — cites Leontyeva & Vavilova on redundancy of Russian grammatical gender (MT Archive).
• Gindin, S., "Na lingvisticheskoy ulitse — prazdnik" (A Holiday on Linguistic Street), Russkiy Yazyk no. 22 (2006), rus.1sept.ru — the contemporaneous profile.
• Melchuk, I. A., "Viktor Yul'evich Rozentsveig (1911–1998)," Vestnik RGGU no. 8(88) (2012), pp. 202–220 — the counter-memoir of her "second father."
• Zholkovsky, A. K., vignettes (samizdat and published) — the Konstantinovich's parallel record of the LMP.
• ISTINA (Intellektual'naya Sistema Tematicheskogo Issledovaniya Naukometricheskikh Dannykh), worker profile and publications list, istina.msu.ru — dates, degrees, publications.
• Uspensky, V. A., Trudy po nematematike (Works on Non-Mathematics), two vols. — the organizational history of the MT Association.

See also

• Machine translation · Computational linguistics · Meaning–Text theory · Natural language processing
• Soviet computer science · History of artificial intelligence · Semantic web
• Alonzo Church · Alan Turing — the other generation that taught machines to mean
• JoAnn H. Morgan — another first: the only woman at a console in Apollo 11's Firing Room