Verbavia

Most language apps pick vocabulary by theme. You learn twenty animals before you learn "although". Verbavia orders vocabulary by how often words actually occur, inside a real A1 to C2 syllabus with grammar, reading and listening. verbavia.com #langsky

'Thank you', ranked by how often film and TV say it: gracias #70 in Spanish, grazie #74 in Italian, 谢谢 #92 in Mandarin, merci #95 in French, danke #98 in German, obrigado #112 in Portuguese. English 'thank' only makes #131.

Happy German Unity Day. German in numbers: ich is the most common word on German screens, 91 word forms cover half of everything said, and doch (#68) is the 'yes, actually' English never got.

Plural "you", ranked in film and TV subtitles: vous #8 in French, ihr #37 in German, vocês #93 in Brazilian Portuguese, voi #152 in Italian, ustedes #258 in Spanish. English has nothing, so "guys" sits at #193. Y'all is filling a real gap. #langsky

Happy European Day of Languages! Share of film and TV dialogue covered by each language's 1,000 most common words: French 79% German 78% Portuguese 77% Spanish 75% Italian 75% A thousand words gets you three quarters of the words in any conversation. #EDL2026 #langsky

Coffee or tea in film/TV dialogue? Coffee, everywhere we checked (rank, 1 = most common): Mandarin 咖啡 904, 茶 1,901 Spanish 617 vs 1,127 French 598 vs 1,203 German 706 vs 1,306 Italian 1,299 vs 2,242 Portuguese 501 vs 1,107 Esperanto 1,285 vs 2,286 Subtitles skew Hollywood, mind. #langsky

Dogs vs cats, ranked by frequency in film/TV dialogue: English: dog 538, cat 1,041 Spanish: perro 622, gato 1,531 French: chien 594, chat 1,442 German: Hund 676, Katze 1,772 Italian: cane 730, gatto 1,752 Mandarin: 狗 634, 猫 1,471 Dogs win everywhere. Cats could not care less. #langsky

Money or love? Rank in film/TV dialogue: Spanish: dinero 164, amor 250 Portuguese (BR): dinheiro 180, amor 233 French: argent 205, amour 299 Italian: soldi 232, amore 345 German: Liebe 188, Geld 193 English: love 123, money 186 Love wins only in English and German, where it's also a verb. #langsky

Esperanto was designed to be the easiest language to learn. Count how much film and TV dialogue its 1,000 commonest words cover, and it comes second from last. The reason: its greatest strength, regular endings. Counted as written, -o, -a, -j, -n turn one root into many forms. #Esperanto #langsky

The 100 most common words make up about half of everything said on screen, in all six of these languages. The next 900 add about a quarter; the next 4,000, about a tenth. From OpenSubtitles counts. Words as written, so hablo and habla count separately, which pulls the Romance languages down.

Mandarin has four tones, Cantonese six (nine if you count checked syllables the traditional way), Shanghainese five. But in Shanghainese the count matters less than you'd think. In a word, the first syllable's tone spreads across the whole thing and the rest lose theirs. #langsky #Shanghainese

Esperanto builds 45 of its commonest words from a grid. Five openings (ki- what, ti- that, i- some, ĉi- every, neni- no) times nine endings (-u person, -o thing, -e place, -am time…). So kiam is when, tiam then, neniam never. Learn 14 pieces, get 45 words. #Esperanto #langsky

In Mandarin 爹 is dad. In Shanghainese 老爹 lau tia is your father's father. And 阿娘 a nyan isn't mother, as a Mandarin speaker would read it. It's grandmother. Kinship terms are where a shared script misleads you most. #Shanghainese #langsky

Many Shanghainese characters have two readings, one literary and one colloquial. 人 is nyin in everyday words like 男人 'man', but zen in learned ones like 人民 'the people'. The literary layer sits closer to Mandarin. Which reading you reach for tells people which register you're in. #langsky

The first 1,000 word families cover roughly 78 to 81 percent of written English. Getting to comfortable unassisted reading takes something like 8,000 to 9,000. So the order you learn words in matters enormously at the start and barely at all later. Most courses have this backwards. #langsky

Esperanto speaker counts range from 63,000 to over two million depending on who is counting. Lindstedt's breakdown is more useful than any single figure: ~1,000 native speakers, ~10,000 fluent, ~100,000 who can use it actively, perhaps a million with passive knowledge. #langsky

In Mandarin 只 zhī classifies animals and one of a pair. In Shanghainese 只 tsah has spread into something close to a general classifier for things. So a Mandarin speaker defaulting to 个 for everything will be understood — and will instantly sound like a Mandarin speaker. #langsky

Wong Kar-wai's Blossoms Shanghai premiered in December 2023 in two versions on the same night: one in Mandarin, one in Shanghainese. After years of the language being kept off broadcast television, that was an event. It's still the best listening practice a learner can get. #langsky

Mandarin lost the Middle Chinese entering tone — the syllables that ended in a stop. Shanghainese kept them. 一 ih, 十 zeh, 学 ghoh: they stop abruptly, on a catch in the throat. It's part of why classical verse still rhymes in southern varieties where Mandarin has lost the rhyme. #langsky

Shanghainese has a classifier Mandarin doesn't: 爿 bae, for shops. 一爿店 ih bae tie, a shop. 一爿饭店 ih bae vae tie, a restaurant. Mandarin uses 家 here. A small word, but one of the tells — you can hear whether someone learned the language or converted it from Mandarin. #langsky

Hangzhou is Wu, but it doesn't sound like the rest of Wu. When the Song court moved there in 1138 it brought a northern-speaking population with it, and the dialect still carries that layer. Wu speakers elsewhere hear it as the odd one out. #langsky

阿拉 ala, the Shanghainese word for "we", isn't originally from Shanghai. It came in with Ningbo migrants and displaced the older local form. The city's most recognisable pronoun is a loan from down the coast. #Shanghainese #langsky

Shanghai was never the prestige variety of Wu. Suzhou was — for centuries it was the cultured form, the one opera was sung in. Shanghai only displaced it as the city grew through the 19th and 20th centuries. Prestige followed the money. #langsky

There is no agreed way to write Shanghainese in the Latin alphabet. Wu Association romanisation and Wugniu spell about a third of common characters differently. 一 is ih or iq. 人 is nyin or gnin. Pick one and stay with it. Mixing sources is the real beginner trap. #langsky

Mandarin has a two-way stop contrast: 波 vs 坡, unaspirated vs aspirated. Shanghainese has three. 波, 坡, 婆 — the third is voiced. Mandarin speakers learning it reliably collapse 婆 into 波, because their own system has no slot for it. #Shanghainese #langsky

Shanghainese fuses its negatives into single characters. 覅 viau = 勿 + 要, "don't" 甮 = "need not" 朆 = "never did" Mandarin writes each of these as two words. Wu compressed them into one syllable and gave each its own character. #Shanghainese #langsky

The FSI puts Mandarin and Cantonese at roughly 2,200 hours for English speakers. It doesn't list Shanghainese at all. So any number you're quoted for it is an extrapolation. Worth saying that out loud rather than inventing a figure. #Shanghainese #langsky

Cantonese and Shanghainese both preserve Middle Chinese — different halves of it. Cantonese kept the final consonants and the tone categories. Shanghainese kept the voiced initials that Mandarin lost. Neither is the older one. They're two different survivals. #langsky

Shanghainese doesn't give each syllable its own tone the way Mandarin does. One contour runs across a whole word or phrase — the first syllable sets it, the rest follow. So drilling a tone per character, the right instinct for Mandarin, actively gets in the way here. #Shanghainese #langsky

FSI hour estimates, for planning realistically: ~600–750 — Spanish, French, Italian, Portuguese, Dutch, Norwegian ~900 — German ~1,100 — Russian, Greek, Hindi, Polish ~2,200 — Mandarin, Japanese, Arabic, Korean That's full-time study with a teacher. A floor, not a forecast. #langsky

The cruellest part of the vocabulary curve is the top. 1,000 word families ≈ 80% of ordinary text 3,000 ≈ 90% 5,000 ≈ 95% 8,000–9,000 ≈ 98%, where you read unaided The last three points cost more words than the first eighty. Hence years at "I get the gist". #langsky

Why vocabulary-size claims never agree: people count different things. "Run, runs, ran, running, runner" is one word family. Count it as five and every figure inflates three- or fourfold. When you see a number, ask whether it's families, lemmas or forms. Most articles don't say. #langsky

Italian makes you choose which auxiliary a verb takes, and then agree the participle: Ho mangiato — I ate Sono andato / sono andata — I went (m/f) Spanish uses haber for everything and never moves the participle. It's the single biggest reason Italian feels harder in month one. #Italian #langsky

Esperanto gets dismissed as a hobby, but it has a literature taken seriously. William Auld was nominated for the Nobel Prize in Literature more than once, for work written in it. Esperanto PEN joined International PEN in 1993 — the body that recognises national literatures. #Esperanto #langsky

"How many Chinese characters do I need?" is the wrong question. Characters aren't words. 电 diàn and 脑 nǎo are two characters; 电脑 "computer" is one word. Most modern Chinese vocabulary is two-character compounds. Count words, not characters. #Mandarin #langsky

The vocabulary coverage curve, which more courses should be built around: 1,000 word families ≈ 80% of running text 3,000 ≈ 90% 5,000 ≈ 95% 8,000–9,000 ≈ 98%, the point you read unaided The first thousand does roughly as much work as the next seven combined. Order matters more than total. #langsky

A thing that surprises people learning Shanghainese: the pronouns aren't Mandarin pronouns with a different accent. They're different words. 侬 nong — you 阿拉 ala — we 伊 yi — he/she You can't guess any of them from 你 我们 他. And they're in your first sentence. #Shanghainese #langsky

Tullio De Mauro's vocabolario di base maps Italian by how much work each word does: about 2,000 fundamental, 2,750 high-use, 2,300 high-availability. Around 7,000 total, covering the overwhelming majority of ordinary Italian. Built in the 1980s. Still the best map of the language. #Italian #langsky

The Goethe-Institut's own word lists: A1 about 650 words, A2 about 1,300, B1 about 2,400. People imagine B1 German needs tens of thousands. It needs roughly the vocabulary of a six-year-old — met in the right order, and actually retrievable. #German #langsky

Cantonese and Shanghainese both have tens of millions of speakers. The resource gap is striking. Cantonese: Jyutping, its own written characters (嘅 咗 佢 喺 冇), dictionaries, subtitled film. Shanghainese: two competing romanisations, no settled written form, almost no graded reading. #langsky

Spanish learners are taught ser = permanent, estar = temporary. Then they meet "está muerto." Death is not temporary. The real split is closer to essence vs. state. Estar marks a condition you arrived at. Being dead is a state you ended up in — so, estar. #Spanish #langsky

The single letter beginners drop in Esperanto is the one doing the most work. The accusative -n marks the object, so word order stops carrying that job. La knabo vidas la hundon and La hundon vidas la knabo mean the same thing. Drop the -n and the sentence collapses. #langsky

Shanghainese keeps a three-way contrast in its initials: voiceless, aspirated, and genuinely voiced. Mandarin only has the first two. It's why Shanghainese sounds softer to Mandarin ears, and why Wu is useful for reconstructing older stages of Chinese. #langsky

The Italian passato remoto is described as literary, and in Milan that's true. In Sicily and much of the south it is ordinary spoken past tense, used for things that happened this morning. Same language, and the tense is alive in one half and archaic in the other. #langsky

Mandarin makes you sort every countable noun into a category before you can count it. 一条鱼 for the fish, 一张纸 for the paper, 一只猫 for the cat. Long and thin, flat, animal. The logic is visible afterwards and almost never predictable beforehand. #langsky

German makes you hold the sentence in memory. The finite verb goes second, then everything piles up in the middle, and the part that tells you what actually happened arrives at the very end. Ich habe das Buch, das mir mein Bruder geschenkt hat, gelesen. #langsky

France says soixante-dix, quatre-vingts, quatre-vingt-dix. Belgium and Switzerland say septante, nonante, and Switzerland adds huitante. The base-twenty counting isn't older French being preserved in Paris. It's an oddity Paris kept and the others dropped. #langsky

Spanish ser and estar are not permanent versus temporary. That rule breaks immediately: está muerto is about as permanent as it gets. Closer: ser classifies what something is, estar reports the state or place it's currently in. #langsky

Portuguese has something no other Romance language kept: an infinitive that conjugates. para eu fazer, para tu fazeres, para nós fazermos. An infinitive with a subject built into it. Spanish, French and Italian all lost it. #langsky

的 is the most common character in written Chinese by a wide margin — somewhere around one in every twenty-five you'll meet on a page. Getting comfortable with everything it does buys you more reading ability than the next twenty characters combined. #langsky

What makes French hard to listen to isn't the grammar, it's liaison. les amis runs together. les héros doesn't. Nothing in the spelling tells you which, so h aspiré is a list of a few hundred words you simply have to know. #langsky