2

Do IMEs offer gibberish non-words when they make suggestions?

I know that there's many words that have the same pronunciation, and therefore users of IMEs have to choose the correct one. I think 感じ (feeling) versus 漢字 (kanji) would be an example of that.

I'm also aware of different kanji for related, but different concepts, that have the same pronunciation and are presumably true cognates. I think お祖父さん (grandfather) and お爺さん (elderly man) would be an example of this.

But do IMEs offer words that are nonsensical gibberish in Japanese? For example, if I type in "げいこ" into Google Translate's IME, the second option I get is "芸姑", which isn't found in goo.ne.jp nor jisho.org. Is it a real word, or a word that only makes sense in a language other than Japanese, or gibberish? I once came across someone typing 芸鼓, which similarly seems to be a doubtful word.

hippietrail
  • 15,025
  • 29
  • 94
  • 143
Andrew Grimm
  • 16,067
  • 14
  • 70
  • 179
  • I realise this question talks about IMEs, but I asked this on the main site rather than meta because it isn't asking about how to use a specific IME. – Andrew Grimm Jan 30 '14 at 11:55
  • Googling for 芸姑, I find quite a few hits, and they all seem to be referring to more or less the same concept as 芸子, so I would call this a "non-standard choice of characters" for 芸子. – dainichi Jan 31 '14 at 00:06
  • @dainichi Google redirected your search, had the same thing happen to me. 姑 means mother-in-law and is read as しゅうとめ.　芸姑　might possibly mean an old geisha or the woman running the business? I don't know, can anyone give some hints? – razorramon Feb 12 '14 at 01:57
  • IMEs may or may not offer wrong candidates due to bugs or due to loopholes in their candidate selection logic. There is no yes or no answer. The best you could find is some examples of IMEs offering candidates everybody agrees are wrong. Then you will know there is at least one bug in at least that one IME. – hippietrail Mar 09 '14 at 09:58

2 Answers2

1

The wikipedia article for 芸妓 says that, in Kyoto, 芸妓 is pronounced げいこ and sometimes written as 芸姑:

芸妓を「芸妓（げいこ）」（「芸姑」という表記もあり）、見習を「舞妓（まいこ）」と呼ぶ。

So yes, this seems to be a real but rare word/spelling.

Darius Jahandarie
  • 15,498
  • 4
  • 43
  • 108
0

I can't replicate that result using mozc, which is a Google-originated IME I use on Linux, or on the standard Windows IME.

In both cases they give 芸子 and then 稽古{けいこ} - the latter probably because of words like 朝稽古{あさげいこ}.

I think the Google algorithm at some point has managed to confuse 妓 (芸妓) and 姑, perhaps by using some badly OCRed text as input.

nkjt
  • 5,742
  • 2
  • 18
  • 40
  • 2
    I think it's more likely that Google analyzes automatic text completion from their web search. Searching for 芸姑 gives quite a few results in Japanese. Whether they be typos or not, the Google Translate IME probably knows about them. – Earthliŋ Jan 30 '14 at 18:05