Catalan Crossword Puzzles

French, Spanish, Italian and Portuguese were seeded here before Catalan, and none of them could hold a full 15×15 in the American style at first — the fully checked grid, where every letter belongs to an Across and a Down answer, spends about twenty of its entries at three letters and thirty at four, and a Spanish pool of common words turned out to have roughly 137 three-letter answers to spend. Catalan is the first of the family whose shelf is that shape all the way through: 11×11, 13×13 and 15×15 grids, every letter crossed both ways, an Across and a Down list, and nothing else. If you solve the New York Times, the layout needs no introduction; if you solve the mots encreuats in La Vanguardia or El Punt Avui, the difference is the numbering — a number in the first square of every answer, rather than the rows and columns numbered down the side and along the top.

What Catalan has that its neighbours lacked is a short vocabulary that reads as words rather than abbreviations. The pool the puzzles were filled from holds 300 three-letter and 903 four-letter answers, and the three-letter ones are the language's own — ARA, AVI, BEC, CEL, COR, FOC, LLET, NEU, NIT, PEU, RIU, SOL, ULL, XAI. The two 15×15s were built first, while the pool was whole, and both filled from the most frequent 15% of the vocabulary; the 13×13s followed from the top 15% and 30%, and the 11×11s from what was left. Each finished grid withholds its answers from the ones after it, and the order matters: with the big grids last, as the Danish shelf was built, the second 15×15 could not be filled at all.

Catalan writes with the 26 Latin letters and marks its sounds with signs rather than new letters, and every one of those signs comes off in the grid. À, È, É, Í, Ï, Ò, Ó, Ú and Ü are entered as the plain vowel; Ç is entered as C, so CANÇÓ sits in five squares as CANCO and BARÇA in five as BARCA. That is the convention of every Catalan newspaper grid, and it is why the accent that separates an open from a closed E — DÉU (god) from DEU (ten) — is no help in a crossword, and the crossing letters have to be.

The letter Catalan is known for is the geminate L: col·legi, intel·ligent, paral·lel, written with a raised dot between the two Ls. In a grid the dot is dropped and the word takes two ordinary L squares, COLLEGI, which makes it indistinguishable from the LL of LLET or CASTELL — the pool behind these puzzles holds 5,398 headwords spelled with l·l, and 19,324 word-list entries with LL in them one way or the other. NY does the work Spanish gives to Ñ (ANY, CANYA, MUNTANYA), and appears in 4,181 entries; TX (COTXE, DUTXA) in 1,107; and the -IG that closes BOIG, PUIG and MAIG in 492.

Because the folded alphabet is the plain Latin 26, .puz export works for Catalan without loss: a puzzle survives the round trip and opens in Across Lite.

Auto-fill draws on about 317,000 Catalan entries, from the Softcatalà spelling dictionary and three frequency corpora. Every other language here is ranked by one — the OpenSubtitles corpus, which says how often a form is spoken in film dialogue — and for Catalan that corpus is a tenth the size of the others: 71,000 forms against Danish's 689,000. Ranked from it alone, 17,263 of the 26,529 words that ca.wiktionary defines had never once been said in a subtitle and sat at the bottom of the list — MALT, BULB, GURU, TUBA and ORAR among them. Adding the Leipzig Corpora Collection's Catalan news and Wikipedia word tables, each scaled to its own size before the counts are summed, raised the frequency-scored vocabulary from 39,588 words to 177,792 and doubled the pool the puzzles could be filled from.

The list peaks at eight letters, with 43,780 entries, and is nearly as thick at seven (40,436) and nine (42,265); it still holds 5,332 entries at fifteen. The clues have a shape of their own. ca.wiktionary is sixty percent conjugation table — 'tercera persona del plural del condicional de disfressar' — and those rows were filtered out before any grid was filled, so the answers are lemmas: a verb appears as its infinitive, a noun in the singular, and the clue is a definition rather than a grammar label. Five and a half thousand of the definitions were written fresh for words the Catalan Wiktionary had never glossed, from the English Wiktionary's Catalan entries, in central Catalan rather than translated from the English.