reading a spectrogram in praat-french word

[katie333]

Hi
I am looking at a spectrogram in praat.
Part of the sentence includes the words "symptoms"
I am wondering how to distinguish the p from the t
Secondly, I'm not even sure I hear the t.
Any thoughts?

[Daniel]

I am wondering how to distinguish the p from the t

It would depend on the recording, but in fast (even normal) speech, they'd run together without any release of the P before the stop for the T, so it would just be one break in the signal. Basically they share a closure, so you could divide it anywhere (the midpoint? or the end of the closure?) and just make sure that you have the part before the closure as the transition to the P (from M) and then the time after the closure as the transition from the T (to the O).

Secondly, I'm not even sure I hear the t.

Hmmm.... they might not sound separate (see above) or it might be sort of articulated under the already closed P. The P makes the lips close, then the T is a constriction inside the mouth, then maybe they're released at similar times. Certainly sometimes you could hear a distinction between "sympoms" and "symptoms" but that could be a little variable.

[katie333]

hearing it again it's the p I don't think I can hear.
I hear symtoms
is it possible it's not being pronounced or is it just there as you describe

[ibarrere]

That word is a little tough because you have a closure starting with a bilabial nasal [m], proceeding into a bilabial stop [p], followed by the alveolar stop [t]. Most people don't noticeably articulate the [p] in normal speech, and since the first two segments share the same place of articulation it's easy to lose it and go straight into the [t].

Is the exercise aimed at identifying the segments on the spectrogram or at transcribing what you see/hear? It may be sort of a trick question, aimed at getting you to dispose of the written bias. Since we're so used to writing symptom as such, it's hard to let go of the fact that the [p] may not even show up in speech. 

[katie333]

it's an exercise to identify the segments that we hear and I just don't hear the p in this voice

[ibarrere]

Then don't transcribe it.

They probably gave you that word specifically because it's written with a letter that's not found articulated in the sample, so it's a bit of a challenge. Just remember that when you're transcribing words with which you're familiar, transcribe them as you hear them, rather than as you might say them yourself, or as somebody might say them in a different environment.

A transcription of myself saying the word in normal/fast speech would be something like [ˈsɪmtəm].

[katie333]

I think I can hear it.  There is a slight p there I believe.  How do I distinguish the sound p and t on the spectrogram?
I have to try to segment the phrase. First time so no clue!

[ibarrere]

The best cues for distinguishing between places of articulation for voiceless stops is by looking for frication noise on the obstruents and formant shaping in surrounding voiced segments.

You can get a lot of information on surrounding obstruents by looking at how the voicing formants shape themselves in preparation for articulatory closure. Bilabials tend to pinch formants two and three, so you'll see both formant lines coming together right before the closure. In a word like this though that may not help a ton since we've already got a bilabial nasal beforehand messing with the formants.

In addition, bilabial stops tend to have a lot less frication than alveolar stops, you can see this by making the sounds one after another and listening carefully. With a [t] it is almost realized as an affricate whereas with the [p] is a clearer stop. Also, since this [p] will be lacking in any aspiration (due to its location in the word) it should have almost zero frication noise associated with it. It won't be super easy to spot, but look for frication noise after each suspected stop, that should give you enough information to distinguish between them. And remember, the [p] may not be there at all, depending on the sample.

This document has a lot of good information on stop cues in spectrograms.

[lx]

I'd be very surprised to see the a [p] show up on a spectrum of the word symptoms. If it was an isolated word reading, in citation form, really slow, then that's where you have people speaking more like the written form but in any normal speech I would be really surprised if people actually used it. Very often in these courses, they throw words like this at you to get you to challenge what you think you know about words and what constitutes them. This seems to be a pretty clear case.

[katie333]

thanks for all the thoughts
I guess the prof is really wanting us to think. And I am!  this is just an exercise and I"m thinking this is going to be complicated!
I'll attach the sound file.
Just curious if you hear the p.
Is it possible to hear the p then and have it not show on the spectrogram?
It's in french but symptoms is the same word.

[lx]

It's in french but symptoms is the same word.

Wow.
I'm really taken aback you neglected to mention it was someone speaking in French. You can't ask a question about the English word 'symptoms', get a few responses about it, with people thinking within the system of English phonetics and phonology, and then just plug in that advice for a language with a cognate word where the word is said with a significant phonetic difference.

The English word 'symptom' and the French word 'symptôme' in the study of phonetics and acoustics are to be classified as completely independent. I know you said you were new to this, but you just can't be under the impression that you can ask about a word, leading people to believe it's said within the system of English phonetics/phonology and what you're actually referring to, isn't.

The difference between English and French in producing [mp] is huge. In the French word, the p devoices the nasal and the nasal has the effect of nasalising the vowel and then the 'p' is (canonically) implied by the devoiced nasal, but could also be heard in some speakers (my knowledge on French phonetics/phonology in general is limited). Compared to English, the corresponding phonetic description is very, very different in a word like 'symptom'.

I cannot highlight enough the importance of clearly understanding that you can't link words like this across languages at all.

[katie333]

my apologies
I do understand there are differences but I guess didn't think of them as hugely as you say I should
I am truly new to this.
Sorry but I do thank you for stressing this to me

[lx]

my apologies
I do understand there are differences but I guess didn't think of them as hugely as you say I should
I am truly new to this.
Sorry but I do thank you for stressing this to me

I read over my post and I realise it seemed a bit strongly worded and I didn't intend for that at all. Please don't read anything into it except an urgency to correct what I viewed as an important mistake to correct. But, now we have the information, the right analysis can be found now we know we're not dealing with how an English speaker would say this word.

As I said before, what happens is the sounds like 'p' have an effect on surrounding words, and the brain hears the EFFECTS of this sound and then makes an association to that sound, but if the sound is then removed, then you can hear its typical effects, but not the sound itself. That happens a lot in French. In words like 'bon' there is effectively no end nasal (n) at the end of the word, but nasals coming after vowels often have the effect of nasalising the vowel, and that's why the word doesn't sound like 'bo', because the mind hears the nasal quality of the vowel and therefore 'hears' it, even though on a normal spectrogram there is more than likely not going to be any trace of an 'n' at the end of the word.

I believe that's what's happening in your case with the 'm' and 'p'.

[katie333]

I understand about bon.
So you think the p being unvoiced is making the m unvoiced as well?
or vice versa?

[lx]

I understand about bon.
So you think the p being unvoiced is making the m unvoiced as well?
or vice versa?

The [p] (voiceless bilabial plosive) shares the same place of articulation as [m] (voiced bilabial nasal) and the nasal clearly passes its nasal quality to the preceding vowel (regressive assimilation) and then the voiceless nature of the [p] - being similar in articulatory features to the [m] - then regressively assimilates with the feature of voicelessness, having the effect of devoicing the [m].

Then you also have the effect of this happening at the end of the first syllable, with the following syllable onset also being a plosive, this clash is also a factor in causing other sounds in words to drop out. So, this is probably the main cause of the [p] not being heard, but its effects are present in how it affected the [m].

So, as is very unintuitive to new learners of phonetics, a sound is understood not solely by the production of the sound itself, but also by how it affects the surrounding sounds. So, thinking you can hear a 'p' and not actually hearing a 'p' don't have to be contradictory statements. A 'p' might not be there, but its traces might be, which are often very highly associated with such a sound.