I am currently trying to get my head around the pitch accent based on the following dialog and especially the marked sentence:


女 いいわね、広々としていて。気持ちまでゆったりするわ。遠くに一本だけ見える木がまたいい。 …

I've loaded the audio file (it's from a JLPT practice book: 新日檢N2模擬試題大全) into Praat and displayed the pitch of this phrase:

Praat output

In the OJAD dictionary, on the other hand, it's displayed like this:


For me, this is quite a difference. Can this be because of some regional differences or am I reading the Praat output wrong? Especially I'm wondering why there is such a huge drop before the ち. Can this be because of the affricate sound? Same thing with the drop at the っ in ゆったり.

Also, according to the OJAD, the ゆ should still be low, but it's clearly rising already inside the ゆ.

Edit: Here's the audio file: https://voca.ro/11eMLIDQKTL0

  • Could you include audio for those that dont have the book?
    – firuvi
    Commented Jun 1, 2022 at 19:42
  • Could you include the audio? This is going to be impossible to answer otherwise. Commented Jun 1, 2022 at 19:43
  • Okay, I added the audio file (Unfortunately Stackexchange doesn't seem to have a good audio upload option :D).
    – Hellstorm
    Commented Jun 1, 2022 at 20:05
  • I love how so many people worry about pitch, when my Japanese teacher told us not to worry about it, and it's never been an issue for me.
    – istrasci
    Commented Jun 2, 2022 at 18:45
  • @istrasci Seems a bit rude to go around commenting on pitch accent questions obliquely suggesting that people are being too concerned about it. Commented Jun 2, 2022 at 19:07

2 Answers 2


It looks as though the highest part (between ち and ゆ) got slided down in the chart for some reason. The long vertical line after ゆ indicates that.

EDIT: Per @morhetb's comment below, it looks like the pitch range settings affect this. The range can be configured in: Menu > Pitch > Pitch settings > Pitch range. (75-500 Hz by default)

I tried using your audio file and a fresh install of Praat with default configuration to produce the same chart. (Praat 6.2.09 on Ubuntu Linux) The result doesn't contain the oddity. It basically matches the OJAD contour except for the pitch rise at the end of わ.

Praat result

I'm not sure if this matters, but I used Audacity to extract the part in question from your MP3 file and saved it as a WAV before feeding it to Praat.

https://linguistics.stackexchange.com/ might be a better place for getting help with Praat. At least I'm certainly not an expert.

  • 3
    @Hellstorm The problem seems to be caused by the pitch range in Pitch settings.
    – morhetb
    Commented Jun 2, 2022 at 16:38
  • 1
    Thanks a lot! I think it's really a problem with the pitch setting. When I set the upper bound to 500 Hz, the output looks correct (the highest pitch was 350 Hz, which was outside my previous range). I thought these pitch settings are just for display, but apparently it's also used for calculating the curve.
    – Hellstorm
    Commented Jun 2, 2022 at 18:50

It looks like this is just confusing/wrong output from Praat.

The audio almost exactly matches OJAD's suggested line/accents, with the exception that the audio has a drop before わ followed by an intonation rise on the end of the わ (like Praat captures) which OJAD doesn't capture.

  • 1
    Maybe someone else can write an answer about why Praat gets it wrong, since you seem to be interested in that. I do imagine it has something to do with how dominant the consonant is in ち and the fact that っ is silent, and it looks like it messes up when the lines are disconnected from each other. But I'm no expert on the tool so I can't really give a well-founded explanation as to why its analysis method is going awry. Commented Jun 1, 2022 at 22:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .