AI Overview

Language Log #5: Managing 18 Languages with Phrasing – Low-Priority & Experimental Updates

This episode is a reflective "language log" update focused on the speaker's high number of languages (18 total) using the Phrasing app. After covering main languages (Arabic, Turkish, Croatian, Japanese) and maintenance ones (French, Dutch, Italian, Portuguese) in prior episodes, this installment discusses the remaining ~10 low-priority ("tier 5" or lower) languages, while exploring broader themes of polyglot learning strategy.

• Why learn so many languages? (The core rationale)

  • Admits partial rage bait motivation — choosing 18 creates clickable, outrageous appeal, but emphasizes genuine effort and no deception.

  • Primary driver: Testing Phrasing across extreme linguistic diversity (right-to-left scripts, Chinese characters, tonal systems, agglutinative morphology, case systems, genders, consonant mutation, ergativity, Cyrillic, Devanagari, minority languages).

  • Secondary (and most important) reason: Parallelization of learning. Fluency requires years of exposure with no real shortcuts → Phrasing enables tiny daily doses across many languages simultaneously → builds long-term passive foundation → potential to "cycle up" low-tier languages later with surprising speed.

• Low-priority language highlights & progress notes

  • Maltese (favorite): Semitic (Arabic-like grammar) + heavy Sicilian/Italian vocabulary + Latin script → excellent lens for understanding Arabic patterns more clearly; huge motivation boost when discovering Phrasing already supported it.

  • Cantonese (very enjoyable): More tones (6–9), fun vocal fry/low falling tone; rebellious appeal (Hong Kong, traditional characters, resisting standardization); much more fun to speak than Mandarin.

  • Mandarin Chinese (tier 3, slower progress): Recent switch to character input (no tones) + self-assessing tones → massive improvement in character recognition (≈25% anchor points in sentences).

  • Lithuanian (sleeper hit, tier 2): Most conservative Indo-European language; fascinating links to Sanskrit; unique sounds, very enjoyable despite difficulty.

  • Greek (tier 2, frustrating): High motivation (future Hyperion/Cyprus dialect support goal), but slow grammar progress; many small particles drastically change meaning.

  • Welsh (Celtic): Insane orthography vs. pronunciation mismatch → constant surprise even after months.

  • Macedonian (Balkan Slavic, tier 2): Very close to Croatian → feels like "free" progress; Cyrillic practice; low-resource but usable via Bulgarian transfer.

  • Sanskrit (experimental, limited support): Amazing compound-word formation; Devanagari relearning in progress; AI breakdowns helpful despite occasional errors.

  • Others briefly (mostly passive exposure & sound/word recognition): Estonian (Finno-Ugric, many cases), Spanish (Castilian dialect, tier 5 experiment), European Portuguese (dialect switch experiment).

Overall, these lower-tier languages receive minimal daily attention — just enough exposure to test Phrasing, parse sounds/patterns, and build subconscious familiarity over years. The speaker views this as an long-term experiment in maximizing lifetime languages through parallelism rather than serial deep focus.

Optimized keywords: polyglot learning, learning 18 languages, Phrasing app, parallel language learning, Maltese Arabic connection, Cantonese tones, Lithuanian Sanskrit, low-resource languages.

AI Transcript

[00:00:02] All right. So this is a language log entry. This is the third one of this week where I will be covering the rest of my languages and kind of just talking about the larger number of languages.

[00:00:36] Um, so I've already done a check-in this week for my main languages — that would be Arabic, Turkish, Croatian, and Japanese. And then I did a check-in for the languages that I'm maintaining, relearning, refreshing, etc., which would be French, Dutch, Italian, and Portuguese.

[00:00:40] Now there are still another ten languages to go, and these are all very low-priority languages in my review system. So I just want to discuss the high number of them and kind of where my status is on them. These I'm not learning quick enough to really do in-depth updates like the previous languages.

[00:01:17] So I think the biggest question I want to answer — and I know I've answered this before, but I feel like it needs to be answered several times — is: why am I learning so many languages? Isn't that pretty stupid? And I would say you're not wrong. However, there are three really big reasons that I'm learning eighteen languages.

[00:01:49] The easiest one: it's a little bit of rage bait. I am genuinely learning them — or, as I've liked to say lately, dabbling in a lot of them and learning some quite seriously. But I did want to choose an outrageously high number because it would make for better clicks. So let's acknowledge the elephant in the room — there is a little bit of rage baiting going on, but there's no deception. It's an honest thing that I'm doing and giving genuine effort.

[00:02:28] The second reason — and the one that got me to such a high level — is that I want to test Phrasing in all these different languages. How does Phrasing work for right-to-left? For Chinese characters? Tonal languages? Agglutinative ones? Case systems, genders, consonant mutation, ergativity, different scripts (Cyrillic, Devanagari), minority languages?

[00:03:50] I wanted to learn a language from every major language family to make sure it works and provides a good, pleasant experience. Even if I learn less in the long run, the product gets better.

[00:04:10] The third reason Phrasing's strength really lies in its parallelism. I don't think there's any way to speed up language learning — it takes years of exposure to become fluent. No shortcuts, no "fluent in 30 days." So if you want to learn more languages in your life, you have to parallelize.

[00:05:31] That's what I've tried to do with Phrasing: make it possible to study just a little bit — a couple cards in Estonian per day, maybe not even every day for some languages. These tier-five languages are just things I want to see, to test the app, to get exposed in minimal ways. The brain is predisposed for language, and with audio, visuals, explanations, conjugations — it's saving tons of information passively.

[00:06:52] My theory (yet to be proven, part of the experiment): if I expose myself to these on a small basis while focusing on tier-one languages, after a couple years I can cycle up the tier-fives and make quick progress — the same way a native English speaker picks up Spanish or French easily.

[00:08:18] So that's the plan: test the application, learn in parallel, and hopefully learn substantially more languages over multiple years than focusing on one at a time.

[00:10:26] With all that being said, let's jump into some of the languages.

[00:10:36] First, Maltese. Malta is an island just off Sicily. It was conquered by Arabic speakers, then cut off from the Arabic world but traded heavily with Sicily. Maltese is a Semitic language (very old branch of Arabic) with hundreds or thousands of Sicilian/Italian words.

[00:12:41] It's written in Latin script, so I can read it. About 40% of vocabulary is Italian-influenced, giving anchor points. But grammar is Semitic. This has greatly improved my Arabic — patterns that were foggy in Arabic script become crystal clear in Maltese.

[00:14:38] It has a special place in my heart: I discovered it, thought "how have I never heard this?", checked Phrasing — and it was supported. Went from "this exists" to studying it in an hour. One of my favorites.

[00:16:19] Another fun one is Cantonese. I always told myself I wouldn't learn Chinese until I was thirty. When I turned thirty, I tried Cantonese first (traditional characters, Hong Kong, rebellious vibe), but failed due to lack of resources. Now with Phrasing, both Mandarin and Cantonese work similarly.

[00:18:54] Cantonese is so much fun to speak — more tones, low falling tone with vocal fry. Highly recommend just trying to repeat sentences. It feels accessible pronunciation-wise.

[00:20:54] Maltese and Cantonese are tier two. Mandarin is tier three (probably need to bump it up — not enough reviews for progress). Switched to character input (no tones) and self-assess tones — huge jump in character recognition.

[00:25:03] Lithuanian has been a sleeper hit. Most conservative Indo-European language; fascinating connections to Sanskrit. Unique sound, very enjoyable despite difficulty.

[00:27:33] Greek (tier two) is insanely hard — struggling a lot, but motivated for future dialect support. Many tiny words/particles change everything.

[00:29:45] Welsh is insane — orthography vs. pronunciation mismatch is wild; still surprised constantly.

[00:30:44] Estonian — completely different family; parsing sounds and basic patterns.

[00:31:45] Sanskrit — worst support (fibbed a bit); slow progress relearning Devanagari, but compound words fascinating.

[00:35:25] Spanish (tier five, Castilian dialect) — pure experiment: how much can minimal exposure + Romance overlap give over years?

[00:36:52] European Portuguese — dialect switch experiment; sounds completely different.

[00:37:21] Macedonian — Balkan Slavic, very close to Croatian → feels like free progress; Cyrillic practice.

[00:39:12] That covers most. These low-tier languages are mostly about recognizing words, sounds, basic patterns (verbs vs. adjectives, genders, cases). I'm dabbling, but over years it's building.

[00:40:54] This has been the longest check-in yet. Thanks for listening if you made it this far. More non-language-log episodes coming soon — product, business, Phrasing method. Ciao.