3 ms·
[flagged]
by cool_dude85 2y ago
[flagged]
- fastball 2y agoI'm not really understanding your criticism. It seems reasonable to assume that a more complex alphabet generally results in a more complex language. After all, when teaching language, a decent amount of time is spent just memorizing letters. Almost double the number of letters would make that task alone twice as hard. Of course, you could then make the argument that these vowels are encapsulated in English (as the same vowels in English have different sounds), but I think the article makes it clear in the sentences around your quote that these extra letters exist because Navajo has much more complex tonal sounds than English, not because they are making something implicit in English explicit in their language.
- lolinder 2y agoWhat they're saying is that English doesn't have 21 consonant sounds and 5 vowel sounds, which is 100% accurate. English has 21 letters which we identify as consonants and 5 letters which we identify as vowels, which are used to write ~24 consonants (depending on accent, but generally clustered around there) and a large number of vowels that is impossible to neatly catalog because it varies widely by accent—it's somewhere in the neighborhood of 14-25, depending on accent.
- CorrectHorseBat 2y agoFor the vast majority of history, the vast majority of people could not read or write. We also only learn to read or write long after we learn to speak, so it does not seam reasonable to me that more complex alphabet results in a more complex language. Chinese and Japanese have arguably some of the most complex "alphabets" (yes, Japanese has hiragana/katana, but they are more recent and Chinese only uses pinyin/bopomofo since the last century) and they have some of the least complex sounds of all languages.
- fastball 2y agoHow exactly do you figure that Chinese has some of the least complex sounds? Chinese not only has just as many sounds as other languages, they also have the added dimension of "tones" which can be very easily described as "sounds" and trip up most people learning the language.
- CorrectHorseBat 2y agoFair enough, it depends on what you exactly mean by a "complex sound". The number of phomenes in (Mandarin) Chinese is indeed not low, but it has only ~1200 different syllables (including those differentiated by only tone differences), which is pretty low compared to other languages. English has 26 letters which represents 44 phomenes, modern Chinese has ~8000 characters for 1200 syllables, I don't think you can find a system in there.
- dang 2y ago"Please don't post shallow dismissals, especially of other people's work." "Edit out swipes." https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
- xarope 2y agoI think it's reasonably legit. By way of other languages as examples: - words for half-empty/full - Eskimo-Aleut has 50 (?) words for different forms of snow - tonality adds complexity; mandarin (4 tones + neutral), cantonese (9? tones) - and various patois, not just in other languages, but also english (my exceedingly limited source being Graham Norton's youtube videos with the British/Scottish and Americans haha)
- lolinder 2y ago> Eskimo-Aleut has 50 (?) words for different forms of snow This one is actually a very good example of the problem that OP is describing of non-linguists writing linguistics pieces. The "50 words for snow" story is one of the first linguistics cautionary tales they'll teach you in a college linguistics course these days, on the same level as the Hyatt Regency walkway collapse is for engineering. The short of it is that it's based on a misunderstanding of the word "word". English is a very analytic language [0] on the scale of things, meaning it doesn't tend to augment its words to create new meanings. Instead we add adjectives and adverbs and prepositions to adjust a word's meaning. Inuit languages, meanwhile, are highly agglutinative [0], meaning instead of independent adjectives they attach morphemes (meaningful-chunks-of-speech, think prefixes and suffixes) to the word itself to alter its meaning. That's not really the same thing as having X words for Y, because the words are constructed from parts that have their own, independent meaning—these morphemes are essentially a set of adjectives, adverbs, and prepositions that happen to get attached to a word instead of just placed nearby. [0] https://en.wikipedia.org/wiki/Analytic_language https://en.wikipedia.org/wiki/Analytic_language [1] https://en.wikipedia.org/wiki/Agglutinative_language https://en.wikipedia.org/wiki/Agglutinative_language
- FabHK 2y agoThe excellent Language Log blog has a whole series of articles about the "Eskimo words for snow" myth: https://languagelog.ldc.upenn.edu/nll/?p=3497 https://languagelog.ldc.upenn.edu/nll/?p=3497 https://languagelog.ldc.upenn.edu/nll/?cat=37 https://languagelog.ldc.upenn.edu/nll/?cat=37 https://snowclones.org/2007/05/31/if-eskimos-have-n-words-for-snow-x-have-y-words-for-z/ https://snowclones.org/2007/05/31/if-eskimos-have-n-words-fo... And a book: https://www.amazon.com/Eskimo-Vocabulary-Irreverent-Essays-Language/dp/0226685349 https://www.amazon.com/Eskimo-Vocabulary-Irreverent-Essays-L...
- torstenvl 2y agoAgreed. Standard American English has, conservatively, 12 vowels—[a] [æ] [e] [i] [ɛ] [ɪ] [o] [ɑ] [u] [ʊ] [ʌ] [ə]— plus a couple diphthongs. If they are using AI, it's not a good one. ChatGPT says: "English has about 14 to 20 vowel sounds, depending on the dialect. Standard American English typically has 14 distinct vowel sounds, while Received Pronunciation (RP) in British English has around 20. This variation occurs because different dialects have different vowel inventories."