A Note on the Construction of the Index (Unedited AI-slop)

What this index is

The Index view is a curated index in the bookmaker's sense: about 220 entries — roughly 140 concepts and 80 proper names — followed across all 27 seminars (1953–1980) and every edition column on this site: the French critical transcriptions, the official published editions, the unofficial translations, and the machine translations. It is not a full-text search; the search box above the comparison table does that. Every entry was chosen the way an indexer chooses: the canonical Lacanian vocabulary — the terms a reader of Lacan actually looks up — plus the analysts, philosophers, patients, and figures who recur across the teaching. AI systems drafted the entry list and the matching rules described below; the human curator reviewed the list and continues to edit it.

How occurrences are found

The counting itself involves no AI judgment — it is fixed, mechanical matching, the same rules applied the same way every time. Each entry carries a short list of forms per language (its entry page shows them), and the matcher applies these rules:

1. Capitals and accents are ignored. Désir, désir, and DESIR are the same word.
2. Only whole words count. désir never fires inside désirer.
3. French elision is understood. l'objet counts as containing objet.
4. Hyphens count as spaces. Nom-du-Père and nom du père are one and the same form.
5. A word inside quotation marks still counts.
6. Within one entry, the longest listed form wins, so nothing is double-counted: the Möbius-strip entry lists both bande de Mœbius and the bare surname Mœbius, and a single "bande de Mœbius" counts once — not a second time for the surname inside it.
7. But different entries count independently — on purpose. A sentence about the desire of the Other counts for desire of the Other and for desire, because a passage about the desire of the Other is a passage about desire. This is how any good book index behaves.

What is not counted

Two whole columns are deliberately excluded: the audio transcription and the stenotype. They are the noisiest witnesses in the corpus — valuable to read, too unreliable to count. Their mishearings and misreadings would corrupt both the totals and the first-appearance dates, so they contribute nothing to the index. The handful of sessions that appear in two seminars are counted once, in their canonical home. And some strings are simply unmatchable: the bare algebraic letter a, the bare word autre, the bare English need — matching them raw would bury the concepts under thousands of ordinary-language hits. Where a language offers no reliable form, the entry is simply not tracked in that language.

Judgment calls

A curated index is a series of decisions. These are the consequential ones — read the numbers knowing they were made.

1. The bare notation petit a keeps its own entry — because it changes its referent. This is the index's clearest example of a decision it refuses to make for the reader, and it comes in two parts. First: through the 1950s, petit a relates to the other, not to the object — in schema L it writes the little other, the imaginary counterpart, as against grand A, the big Other. From around Seminar X (1962–63), the same little letter increasingly writes something else: objet petit a, the object-cause of desire. An index that folded every bare petit a into the objet petit a entry would silently backdate the object by a decade; one that folded it into the little other would erase the notation's later life. So the bare notation has its own entry, petit a (the notation), cross-linked to both little other and objet petit a, with the shift explained on its page — the three entries are meant to be read together. Where Lacan says only petit a, the index reports the fact and leaves the reading to you. Second: the bare objet a is deliberately not matched at all. Once accents are folded away, "objet a" cannot be told apart from "objet à…" or from "l'objet a…" where a is simply the verb avoir ("the object has…"); matching it would flood the counts with grammar noise. The objet petit a entry therefore counts only explicit "objet petit a" phrasings — a deliberate undercount — while the bare notation's trail lives in the petit a entry. A small residue of noise remains and is accepted: French petit à petit ("little by little," 46 occurrences corpus-wide) and English turns like "so small a result" can be caught by that entry's forms.
2. Everyday words are article-bound. Bare ça, moi, demande, regard, voix are ordinary French speech; counting them raw would drown the concepts in noise and wreck the first-appearance dates. Those entries match only with their article — le ça, le moi, la demande, le regard, la voix — and the same guard protects the other languages: the real and o real rather than the bare adjective, o falo because bare falo is also Portuguese for "I speak." These entries undercount, deliberately.
3. Some people are matched by full name only. Bare Klein would count every Klein bottle toward Melanie Klein; bare Reich would hand Wilhelm Reich the Third Reich; bare Freud belongs to Sigmund, so Anna Freud requires her full name. Passing references — "Mrs Klein" — are missed; that is the price of not counting nonsense.
4. sinthome and symptôme are never merged. The archaic spelling Lacan revives in Seminar XXIII is an event, and the index treats it as one: two entries, cross-linked (in Portuguese likewise sinthoma and sintoma), so the word's 1975 debut stays visible instead of drowning in twenty years of symptom.
5. Freud's German terms have their own entries. Verwerfung is indexed separately from forclusion, its French rendering — cross-linked, so the German term and its translation can each be watched on its own timeline. das Ding is matched only in German: once capitals are ignored, "la Chose" and "the Thing" cannot be told apart from ordinary things, so the entry rides on the German alone and understates the theme.
6. Some famous common words are kept noisy on purpose. amour, lettre, vérité, rêve, sujet — these count every ordinary love, letter, and dream. But an index that could not find the discussions of love, or of the letter, would be worse than one that over-counts them; their entry notes say plainly which way they err.

How to read the numbers

1. Counts are floors, not tallies. Wherever a form was ambiguous, the decision was almost always to undercount rather than to fabricate. Apart from the deliberately noisy entries flagged in their notes, read every total as "at least this many."
2. "Introduced" is a fact about the surviving text, not about doctrine. The date shown for each edition group is the first surviving occurrence of a listed form in that group of witnesses — nothing more. Words have lives before they become concepts: le nom du père is heard in its ordinary liturgical sense years before the Name-of-the-Father is formalized. The index reports what was said and when; the doctrinal birthdate is a judgment it leaves to the reader.
3. A hatched timeline cell means the edition does not exist for that seminar. The witness is missing — which is not the same as the term being absent. Absence of evidence is drawn differently from evidence of absence.

Corrections and additions

The entry list is a living document. If an entry is missing, a matched form seems wrong, or one of the judgment calls above strikes you as the wrong call, write to admin@lacanseminars.com. Adding or repairing an entry is cheap: the whole index is rebuilt from the curated list.