Wikipedia:Reference desk/Archives/Language/2018 May 28

= May 28 =

Software for teaching pronunciation?
Is speech recognition so far developed so that it could be helpful for pronouncing correctly foreign words?--Doroletho (talk) 13:32, 28 May 2018 (UTC)
 * "There's an app for that"™ (no clue as to how good it is):.  Thus-far, there doesn't seem to be an app where you speak in one language, and it translates it to speech (with correct pronunciation) in another language. Siri is remarkably good at understanding speech (at least in my experience) -- combine that with a good language translator and you've got something. This doesn't seem out-of-reach with today's technology.  Links: Speech recognition & Natural-language processing —2606:A000:1126:4CA:0:98F2:CFF6:1782 (talk) 19:15, 28 May 2018 (UTC)
 * That's in the right direction. Only problem is that I don't have an iPhone. I see it promises help for learners of foreign languages, and includes not only the pronunciation of single words, but also intonation of sentences. The latter is for me quite important not only for sounding natural, but to be understood properly.--Doroletho (talk) 13:37, 29 May 2018 (UTC)
 * When I learned Mandarin, it wasn't hearing pronunciation that helped. It was hearing MY pronunciation that helped. I thought I was pronouncing things correctly, but when I heard my speech played back alongside proper speech, I realized that I was very far off. Therefore, a good tool (in my opinion) should have a record/playback feature. 209.149.113.5 (talk) 14:04, 29 May 2018 (UTC)
 * LingoDeer and MangoLanguages are language-learning tools. Both of them have a record/playback feature. MangoLanguages is a service offered by libraries all across the United States of America. You just need your library card to register for an account on MangoLanguages. LingoDeer originally was made for East Asian languages (Korean, Mandarin, Japanese), but then it expanded to other languages. HelloTalk is a social chat app that allows users to chat with native speakers from all over the globe. You can record your voice and let the native speaker make judgments on it. I think Mandarin-as-second-language learners are lucky. The Mandarin tones provide a guideline in pronouncing the words. Try to learn a non-tonal language from a tonal language background. Personally, my experience with learning Korean on LingoDeer is that the Korean version provides 2-3 different recordings of the same word; they are different, because the tone is different. The non-tonal nature of Korean makes the language hard, because for no apparent reason, the tone would be raised or lowered. Most of the time, I couldn't trust my own recordings, because I might pronounce it one way based on one recording but then suddenly hear a different pronunciation in a different context. It's not a big deal. Over time, I eventually recognized the two different pronunciations as variations of the same word, even though I'm not completely sure of the context or the reason why the tone must be raised/lowered. I think a record/playback feature would be useful, but you still need a real native speaker to judge your pronunciation. I think the main struggle that tonal-language speakers have when learning non-tonal languages is that non-tonal languages tend to have unpredictable tones that vary depending on context. Someone on YouTube did a hypothetical video, wondering what it would be like if English was taught with Mandarin tones. LOL. SSS (talk) 19:14, 29 May 2018 (UTC)
 * There are (or once were) software that allows you to visually compare your waveform with a pattern. The idea is shown here. Шурбур (talk) 14:54, 29 May 2018 (UTC)