English: A vs. An

A funny quirk of English I never noticed until I started learning Spanish:

In Spanish, to say for example, "he is a doctor," you would say "él es médico." Which more literally translates to "he is doctor," which feels unintuitive coming from English. Spanish DOES have an equivalent for "a" - "un/una", so the natural thing for an English speaker to say would be "él es UN médico", but I understand that to be awkward/unnatural.

But then I realized, when you say the same sentence in plural - "they are doctors" - you don't have an article. You don't say "they are some doctors" (I had to Google, apparently "some" isn't an article anyway?)

So Spanish ends up more consistent - Singular: "Él es médico." Plural: "Ellos son medicos."

Unlike English - Singular: "He is a doctor." Plural: "They are doctors."

As a native speaker of a language without articles, I find the English article system by far the most intuitive and easiest to learn of all the languages I have studied. Most other Indo-European languages have articles that you simply need to memorize, as there is no way to derive them from the word itself. Which I never manage to do.

What gets me as a non-native speaker far more than "a vs an" is the "a vs the" conundrum. I still don't quite get when something needs to be "indefinite" and I've been speaking English for close to forty years. My native tongue lacks articles altogether and it's soo bloody hard I feel hopeless I'll ever master them fully.

It's avoiding two distinct vowel sounds in succession.

Consider other variations of this rule.

In French, one would say "l'hôpital" rather than "le hôpital" because the 'h' is an aspirated breath (try saying "a aspirated" rather than "an aspirated") leading to an initial vowel sound.

In Irish, the word for father is "athair". To say "her father" it becomes "a hathair" https://en.wikipedia.org/wiki/Prothesis_(linguistics)

In Turkish, it can be seen with araba (car). "I saw the car" -> "Arabayı gördüm." (note the 'y' to avoid two vowels). Book is kitap and "I saw the book" is "Kitabı gördüm." Which doesn't need the additional consonant sound between the final sound and the "ı".

Whatever language you speak, it likely has some form of this rule.

And other languages have the quirk too (like the "a unicorn" or "a university"). For example, in French there's l’hôtel and l'hôpital. But if you've got the hedge... it's la haie... or even more egregiously, the marmoset is le ouistiti.

This is exactly what language models are good at. It's not a problem that could be solved purely deterministically, because it requires knowledge of the actual pronunciation in that particular language or dialect. There can be so many exceptions to the rule, so it's just easier to train a small language model on a sufficiently large dataset.

> if word.startswith("ub"): return False

An uber-rich man riding in an Uber installed an Ubuntu package.

I'm actually a bit surprised by this rule. Is this line there entirely due to the word "ubiquitous"? I think it is.

The most problematic for me as a non-native has been "the" vs "nothing", i.e. when one can/should put the definite article.

Pre-LLM I was joking that real AI is whatever can make that decision correctly 100% of times.

I would take a/an any day over ein/einen/einem/einer/eines in German

There's also "the" vs "the". As in "the boy ate the apple". First 'the' ends with a schwa sound. Second 'the' ends with 'ee' sound.

What about "An honorable man" spoken with an American accent? British accent?

Dialects and syntax are weirdly fascinating .

I'm always curious why sometimes I pronounce the as /ðə/ ("thuh") and sometimes as /ði/ ("thee"). I don't really know what governs this but I expect it's the presence of a vowel at the start of the following word. However, only some vowels. Does anyone have an answer?

Fun fact an adder (snake) was once a nadder, and an apron was on a napron. This fun fact brought to you by The History of English Podcast.

https://historyofenglishpodcast.com/

Happily sourced in Wikipedia so I don't have to figure out which episode I learned that from:

https://en.wikipedia.org/wiki/Adder <- see below the table in Taxonomy.

Starting earlier today, I've been using Firefox's local translation feature to auto-translate every page to Italian (to learn). I just had to remove English from my language list and check an option to translate English to Italian.

Firefox's local translation is pretty decent but not amazing. It did render this article's title as "Italiano: A vs. An", which I found pretty funny.

Non-native speaker here. To me in spoken language both "unicorn" and "hour" have a combination of 2 vowels in the beginning. Let me transcribe them as "ee-oo" and "aa-oo" here.

Why is the aa-oo combination at the beginning of "hour" considered a vowel but the ee-oo in "uniform" is a consonant?

Interesting analysis

> [LLM It doesn’t need to be clean or maintainable. It only needs to be correct

But correctness is precisely the issue with LLMs?

But also

> I would’ve spent more time on the trie simplification algorithm and less time on parsing cmudict and re-learning d3.js.]

You could've spent nothing on D3 because the result is bad/poorly readable - much worse than a simple table alphabetically-aligned interception table, which I could use to answer curious questions such as "how can I compare "A" and "O" in terms of exceptions? or "which letter has the most exceptions"?")

(and the blue color is out of place, vowel/consonant split should visually be the same "category", not black/colord. Red for exceptions is fine)