Phonetics, Phonology, Praat, Spectrogram

Flap vs. Trill (ɾ vs. r)

In Spanish [r] and [ɾ] have phonemic status: they provide a difference meaning forming a minimal pair. During my studies, I was told that Spanish ‘but’ <pero> [peɾo] and ‘dog’ <perro>[pero] form such a minimal pair (please correct me if I’m wrong). Since I am able to pronounce the rolled/trilled /r/ as well as the flapped one, I recorded myself (but I hope to deliver a recording of a native speaker of Spanish some day soon). This is the very spectrogram:

Praat Spectrogram <pero> <perro>

You can clearly see that in <pero> there is just one single flap (or tap) of the tongue against the alveolar ridge, whereas as in <perro> the tongue flaps against it several times (5 times to be precise). Also perfectly visible: the plosive /p/ with an opening of the lips (before the dark vertical line) and the “explosion”: airstream bursts out at once (dark line itself), and finally the aspiration (/p, t, k/ tend to be unaspirated in Spanish; since I am a native German speaker, who aspirates voiceless plosives as well as English native speakers do, the aspiration does occur in the spectrogram.) The vowels are clearly detectable as well.

Stay curious and have a nice weekend!


Phonetics, Phonology, Praat, Spectrogram

Praat: Consonants and the Vocal Tract

Finally, I have spectrogram proof for consonants blocking the airstream at different places in the mouth. I mentioned this problem in another post: I tried to see the difference between the pronunciation of /d/ and /b/. I recorded myself, but the spectrogram did not really show significant differences. Hence, I recorded my boyfriend to see if that might change the results…and it did! I let him say <mo> and <no>. So. first of all consider the following information and take a look at the spectrogram:

  1. Keep in mind that /m/ is a bilabial nasal, hence the airstream is blocked at the lips. /n/ is an alveolar nasal, the airstream is therefore blocked earlier between the back of your teeth and the alveolar ridge. In turn /m/ “makes more use” of the vocal tract than the /n/. (Help: Take a look at the vocal tract below to visualize the place of articulation!)
  2. F2 gives us information about the place of articulation, as we already saw in the spectrograms of the different vowel (front/back). This also applies for consonants as follows: the more the consonants makes use of the vocal tract (bilabials make the most use) the lower is F2. If the airstream is blocked earlier, F2 starts at a higher point.
  3. You can observe that F2 of /m/ is lower than that of /n/, proofing the assumption that F2 also indicates the place of articulation of consonants. What is striking as well is that nasals strongly resemble vowels, except the formants are not as strong.

taken from FRH 2011: 216.

I also recorded the sequences /baː/ /daː/ /gaː/ and the same perfectly visible result: they are all plosive but have a different place of articulation: /b/-bilabial, /d/-alveolar and /g/-velar. These three stops block the airstream at three different points in the vocal tract, from the front to the back. As you can see again in the spectrogram: the more the consonants makes use of the vocal tract (bilabials make the most use, alveolars follow, then velars) the lower is F2. If the airstream is blocked earlier, F2 starts at a higher point:

Praat Spectrogram of <ba> <da> <ga>

Stay curious, motivated and interested and of course: stay tuned!


P.S. A special thank you goes out to Dennis who does not mind being recorded and provides great sound samples!

FRH 2011: Fromkin, V., Rodman, R., Hyams, N. An Introduction to Language . Boston: Wadsworth.

Phonetics, Phonology, Praat, Spectrogram

Praat: Consonants – The Big Voiced/Voiceless Test

Well, I sat down and tried to categorize the consonants cleverly to get good, comprehensible and analyzable spectrogram results. I decided to sort them into the main voiced [p, t, k, tʃ, f, θ, s, ʃ] and voiceless ones [b, d, ɡ, dʒ, v, ð, z, ʒ] of the English. It was a lot of (fun) work to record, extract and analyze and of course finally getting a god result:

Spectrogram of Voiced Consonants (of English that have voiceless counterparts)

Spectrogram of Voiceless Consonants

Now, what are the striking differences?

It is obvious that the voiced consonants, contrary to the voiceless, show strong acoustic energy at a very low frequency (F1 is low): this corresponds to the vibrating of the vocal cords (UCL). Since the pronunciation of voiceless consonants does not involve vibrating of the vocal cords, we do not find this strong band at the bottom of their spectrogram. Another observation made is that the formants of the voiceless consonants seem to be more and more confused as the ones of the voiced consonants. This could prove that voiceless consonants are pronounced with more acoustic power or to put it unscientifically simple: they sound hard and more pressurized than their voiced counterparts: they sound softer and relaxed due to the vibration of the vocal cords. In German, we even say “scharfes s/sharp s ” and “weiches s/soft s” to refer to the voiceless and voiced /s/.

Feel voiced/voiceless consonants:

If you are not sure whether you are dealing with the voiced or voiceless variant of a sound, touch your neck/throat and make the sounds, e.g. /b/ and /p/: if you feel the vibration, it is a voiced consonant

Stay tuned because the next topic I’m preparing is Morphology!

Phonetics, Phonology, Praat, Spectrogram

Praat: Diphthongs

Today I recorded and analyzed the spectrograms of diphthongs which was quite successful. Consonants on the other hand, are very hard to distinguish from each other and sometimes analyzing the spectrogram seems more like interpreting it or “trying to see want you want to see or should presumably see”. Therefore, I am still trying to work out the differences between e.g., the plosives /b/ and /d/ (the F2 should indicate the length of the vocal tract: /b/ is bilabial and /d/ is alveolar = /b/ has a longer vowel tract than /d/). The same accounts for the nasals /m/ and /n/.

The following two spectrograms show the diphthongs /ɔɪ̬/ and in <boy> and /aɪ̬/ as in <eye>:

Praat Spectrogram of <boy>


Praat Spectrogram of <eye>

You can clearly see how one vowel shifts to the other: in <boy> you have a open-mid, back vowel /ɔ/ that shifts to a quite closed, front vowel ɪ: F1 of is /ɪ/ lower than that of /ɔ/, referring to the vowel being more closed; F2 of /ɪ/ is much higher than that of /ɔ/, since it is a front vowel. In short: We can see the shift from open-mid to closed (F2 getting lower) and from back to front (F2 raising high). Almost the same applies for the other diphthong in <eye>, since /a/ and /ɔ/’s spectrograms look very similar (except for the F1 of /a/ starting on a higher frequency than /ɔ/ due to it being more open).

There is still more to come…hopefully spectrograms of different consonants, so stay tuned 🙂

Phonetics, Phonology, Praat, Spectrogram

Phonetics/Phonology: Praat is so much fun!

Dear readers

at the moment I am preparing for my Linguistic Analysis class next semester. I restructured the course to make it more fun and related to practice; hence, I decided to work with Praat, a free scientific software program for the analysis of speech in phonetics, to make Phonetics/Phonology more interesting and fun! I wondered how I could use this program to teach the different phones and phonemes in English and German (and further languages like Spanish where the trilled/rolled and tapped r form a minimal pair). Analyzing the spectrogram: Do vowels look different from consonants? Do vowels among each other differ? Do voiced and voiceless consonants differ? Can you extract, see and hear the aspiration of voiceless plosives [pʰ, tʰ, kʰ]? How do fricatives or plosives look like?

  1. I browsed the Internet to gather some information of how, e.g. spectrograms of different vowels (should) look like.
  2. I recorded the different vowels myself to prove that the information is right.
  3. You can find short and comprehensible descriptions on how to “read” a spectrogram here:
  4. Look at my cut spectrograms for the vowels [iː (beat), uː (boot), ɑː (BE bath)]

Short description of the spectrograms above:

F1 (Formant[1] 1, the first (dark) line you see) always shows the lowest frequency: If it starts at a higher frequency we have an open vowel like [ɑ]; if it starts at a lower frequency we have a closed vowel like [i, u]. You can perfectly see this in the spectrograms!

F2 (Formant 2, the second (dark) line): If it shows a higher frequency we can expect a front vowel like [i]; if it shows a lower frequency (hence is closer to F1), we can expect a back vowel like [u]. Again, perfectly visible in the spectrograms.

As you can see there are further formants to each vowel, but usually the first two suffice to determine the vowel.

Vowel Chart

More spectrograms will follow, so stay tuned and check out Praat yourself: it is fun!


[1] in short: “distinguishing or meaningful frequency components of human speech” (EN-Wikipedia: Formant); concentration of acustic energy in a certain frequence range (DE-Wikipedia: Formant)