Chapter 4.   Prosodic analysis with Praat   

 

Table of  Contents      

 

4.1 Measuring Duration

 4.1.1 Voice onset time

 4.1.2 Linking duration                  

 4.1.3 Pause duration

 4.1.4 Speech rate

4.2 Measuring Lexical Stress

 4.2.1 What is lexical stress?

 4.2.2 Stress analysis in Praat

4.3 Measuring Intonation

 4.3.1 What is intonation?

 4.3.2 Cantonese vs English

 4.3.3 Common types of tones in English

 4.3.4 Communicative functions of English intonation

4.4 Measuring Tone

  4.4.1 What is tone?

  4.4.2 Tones in Putonghua

  4.4.3 Tones in Cantonese

  4.4.4 Analyzing tones

 

Questions for Chapter 4

 

Back to Praat Home

 

 

 

 

 

 

 

 

   

 

 

 

4.1 Measuring Duration                                                                           Top

 

Once the sound file is open in the Editor window, you can read the total duration easily from the duration bar, but if you want to know the duration of your selected part, you can follow the following steps (Styler, 2012:13)

 

1. Select the portion of the file you’d like to measure with the cursor

2. Read the duration of the selection (in seconds) from the duration bar along the bottom of the Editor window

3. Editor à Query à Get selection length and read your selection in the "Info" window

 

If you’d like to know the duration of an entire file, just select the file in the Objects window and select

Objectsà Queryà Query Time Domainà Get Total Duration

 

4.1.1 Voice Onset Time (VOT)                                                                             Top

 

"Voice Onset Time" (VOT) is the time between when the stop is released and when the voicing of the following vowel begins. Measuring this time, which can be positive (say, for the English voiceless aspirated stop [tH]), around zero (for the English "voiced" stop /d/, or, more commonly, the voiceless unaspirated [t]), or negative (for fully voiced stops, where voicing starts before the stop is released, as found in most non-English languages). Languages classify their stops largely based on voice onset time, and it’s an excellent way of proving the "voiced/voiceless" phonological distinction.

 

Measuring voice onset time (VOT) is very easy to do in Praat, as it’s just a duration measurement between two set points, the release of the stop and the start of voicing.

 

1. Find the stop release

2. Find the start of voicing

3. Select the span between these two points

4. Read the duration of the selection (in seconds) from the duration bar along the bottom of the Editor window

 

If the start of voicing came before the stop release, the VOT is negative. Otherwise, the VOT is positive. In general, voiced sounds (in languages other than English) will have a VOT which is negative. Voiceless unaspirated sounds will have a VOT which is around 0, and aspirated sounds will have a positive VOT. (Styler, 2012:14)

 

4.1.2 Linking duration                                                                                            Top

As Heike (1987) observed, linking can occur in English between two consonants, between a consonant and a vowel, or between two vowels.

  • a. Consonant-to-consonant (CC) linking occurs when the final consonant of the former word and the initial consonant of the latter word are identical, (e.g., “that time”),   only one consonant is manifest and may be slightly prolonged. 

  • b. Consonant-to-vowel (CV) linking (e.g. “kind of”) involves the assignment of the final consonant of a word to the following, vowel-initial syllable.

  • c. Vowel-to-vowel (VV) linking (e.g. “say it”) occurs when a word-final tense vowel is followed by a word-initial vowel.  This kind of linking often involves glide attraction. 

We can decide whether there is linking by measuring the duration of CC, CV, and VV in Praat.

 

4.1.3 Pause duration                                                                                                  Top

Pause is defined as “any interval of the oscillographic trace where the amplitude is indistinguishable from that of the background noise” (Duez, 1982, p.13).  Intensity can be considered the acoustic cue to measure pause (Chen, 2005).  Pauses could be analyzed by measuring pause duration, pause distribution, and pause location.  As a common practice,  only those pauses greater than 100 ms, the cut-off for pauses (Griffths, 1991), were considered to be a  “pause”. Therefore, we can decide whether there is a pause by measuring the duration between two adjacent words.

 

4.1.4. Speech rate                                                                                                      Top

 

Speech rate is defined as the total duration of a sentence (including pauses). The most common measurements of speech rate are syllable per second (sps) and words per minute (wpm) (Buck, 2001).  The total duration and SPS were calculated.  Since measurement of syllable duration is the major focus in this dissertation and the selected speech samples from the four groups were short and identical, the measurement of WPM was not employed here.

 

4.2  Lexical Stress                                                                                                  Top

 

In this chapter, we will discuss the application of Praat in the analysis of the word stress and intonation. Both of stress and intonation belong to the suprasegmental features. There are two kinds of stress: word stress and sentence stress. Word stress is concerned with the stressing of individual words of two or more syllables when they are pronounced in isolation. By intonation we mean the rise and fall of the pitch-change of the voice which take place on the nucleus.

 

4.2.1 What is lexical stress?                                                                                       Top

 

In English we usually find that simple words consisting of two or more syllables have a strong stress on one of these syllables and a weak stress on the remaining syllable or syllables. Stress plays an important role in communication, for it distinguishes different meanings when the stress is put in different positions.

 

“All stressed syllables in words have one characteristic in common, and that is prominence.

Prominence is produced by four main factor: (i) loudness(intensity), (ii) length, (iii) pitch(Fundimental Frequency/F0), (iv)quality. Generally these four factors work together in combination, but experimental work has shown that these factors are not equally important; the strongest effect is produced by pitch, and length is also a powerful factor. Loudness and quality have much less effect.” (Roach 2009:74)

 

Since stress is manifested as rise in pitch, greater intensity (loudness) or greater vowel length, the analysis of stress must include various elements of acoustic analysis, like waveform, spectrogram, and pitch contour.(See Figure 4. 1 )

 

Figure 4.1 The waveform, spectrogram, and pitch contour of sound

 

(retrieved from Stonham’s handout: p3, http://stonham.dyndns.org/phonetics/handouts/prosod_hndt.pdf)

 

 

4.2.2 Stress analysis in Praat                                                                                        Top

 

Since stress is manifested as rise in pitch, greater intensity (loudness) or greater vowel length, the analysis of stress must include various elements of acoustic analysis, like waveform, spectrogram, and pitch contour. You can follow the following standard procedure in Praat to analyze the stress.

 

1. Open the sound file in Praat ( Openread from files…)

2. Select the file and click "View and Edit” on the right side

3. In the Edit window, select SpectrumShow spectrogram (the default setting is showing the spectrogram)

 and then select Pitch→Show pitch, and you will see the following window Figure 4.2:  the pitch analysis is laid over on the spectrogram, but when you export them to the Praat objects list they will be treated as separate objects

 

Figure 4.2

4.  To export different windows, you can go to Spectrum and then choose Extract visible spectrogram, then a new file will appear in your Praat objects list as 'Spectrogram untitled' in Figure 4.3

Figure 4.3

 

5. Then go to Pitch and choose Extract visible pitch contour, and then a new file will appear in your Praat objects list as 'Pitch untitled'as shown in Figure 4. 4

Figure 4.4

 

6. You can also draw the graphics of the waveform by selecting Fileà draw visible sound

7. You can also draw the graphics of pitch track by selecting Pitchà draw visible pitch contour

8. You can also draw the graphics of the spectrogram by selecting Spectrogram à paint visible spectrogram

 

Please see the output in Figure 4.5, which shows waveform, pitch track, and spectrogram

 

Figure 4. 5

See details of drawing and painting in Praat in 1.3.4

 

Let’s look at two examples of analysis of lexical stress in Praat:

 

Example 1                  Figure 4.6  Record  /rI Ûk??ùd/

                       

 

                                      Figure 4.7 Pitch contour of reCORD

                            

 

 

Example 2

Figure 4. 8 Record  / Ûrek«d/

 

 

 

 

Figure 4.9 Pitch contour of REcord

 

 

The word ‘record’ is divided into two parts –‘re-’ and ‘-cord’.

 

1. In the first one, re’cord, the stress is put on the second syllable.

(1). In “re-”, the pitch is falling down, while in “-cord”, the pitch is rising, especially in the part of vowel.

(2) The stressed syllable is higher than unstressed one. And the duration of the second syllable “-cord” is longer than the first.

 

2. In the second one, the stress is put on the first syllable.

(1). It is obvious that the duration of the first syllable “re-” is much longer than the one in the first example. And the pitch is rising in the first syllable.

(2). in the second syllable, the duration is shorter and the pitch is falling down.

 

The table below shows this more clearly;

 

Pitch (HZ)

Duration(S)

 

Pitch(HZ)

Duration(S)

re-

160

0.132

’-cord

172

0.615

’re-

230

0.240

-cord

137

0.356

 

 

 

 

Let's look at another two examples: 

Example 3

                    

                            Figure 4.10  Contrast   /kɒn'trAùst/

 

                      Figure 4. 11 Pitch contour of  ConTRAST

 

 

          Example 4

Figure 4.12  Contrast   /'kɒn trAùst/

                          Figure 4. 13 Pitch contour of CONtrast

 

 

 

Pitch (HZ)

Duration(S)

 

Pitch(HZ)

Duration(S)

con-

129

0.193

’-trast

125

0.567

’con-

146

0.278

-trast

87

0.507

 

4.3 Measuring Intonation                                                                                   Top

 

4.3.1 What is intonation?                                                                                         Top

 

Roach (2000, p.150)

à No definition is completely satisfactory

à The pitch of the voice plays the most important part

Kenworthy (1987, p.11)

à intonation = the melody of speech

Teschner & Whitley (2004, p.64)

 

à Intonation is a series of pitches sung over a whole utterance

Celce-Murcia, Brinton & Goodwin (1996, p.184)

 

à If pitch = individual tones of speech

       then intonation = the entire melodic line

à Intonation involves the rising and falling of the voice to various pitch levels during the articulation of an utterance.

 

4.3.2 Cantonese vs English intonation                                                                     Top

 

Chinese

(including Cantonese, Putonghua & other regional languages in China

English

tone language

 

intonation language

The tone of each character contributes to lexical meanings

The tones vary over a stretch of utterance, and contribute to accentual features, or convey attitudinal meanings.

 

Intonation is determined by many elements like pitch, energy, duration, tempo and sound quality. Pitch plays a more important role than the remaining elements. As English is an intonational language, the intonation expresses speakers’ emotion and attitude. Native speakers use different intonations in different sentence types.

In Praat, we mainly use pitch contour to indicate different tones, like falling and rising tones.

 

1. Select the sound in object list

2. Click View&Edit

3. Pitchàtick show pitch

 

4.3.3 Common types of tones in English                                                              Top

 

a. Fall      

Figure 4.14   

  yes                             no                   well                         four

 

 

b. Rise    

       

Figure 4.15 

         yes                         no                    well                         four     

 

 

c. Fall-rise      

 

Figure 4.16  

            yes                            no                        well                       four    

 

 

d. Rise-fall     

 

Figure 4. 17  

     yes                          no                      well                        four

 

 

e. Level        

 

Figure 4.18

            yes                                 no                         well                         four

(Roach 2009, p.192   Audio Unit 15, Ex 1)

 

4.3.4 Communicative functions of English intonation                                          Top

 

1) The accentual function

       to indicate the focus of the information

2) The attitudinal function

       to indicate the speaker’s attitude

 

The falling tone suggests:

 

    finality, completion

    certainty, belief in the content of the utterance

    speaker-dominance

The rising tone suggests:

 

    more to follow

    querying, uncertainty

    encouraging

    speaker-deference

  the speaker does not know à asks

  does not have authority à requests

The level tone suggests:

 

    routine

    boredom, no interest

 

 

3) The grammatical function

 

a. Statements: falling tone

 

Most native speakers of English can, for example, recognize people from France by their French accents.

Figure 4. 19

 

b. Wh- Questions:  normally falling tone

 

Why do people usually have an accent when they speak a second language?

Figure 4.20

 

c. imperative sentence: falling tone

 

Please sit down.

Figure 4.21

 

d. Yes/No Questions: normally rising tone

Does this mean that accents can’t be changed?

 

Figure 4.22

 

e. Listing: combination of rise & fall

I  like  apple, cherry,↗ and banana.

 

Figure 4.23

 

f. Tag questions: rising and falling tone

 

Falling tone à the speaker is quite certain, only seeking confirmation / agreement

 Rising tone à the speaker is unsure, a genuine request for more information

 

But old habits won’t change without a lot of hard work, will they?

Figure 4.24

 

Another possibility:

 

But old habits won’t change without a lot of hard work, will they?

Figure 4.25

 

g. Closed-choice alternative questions: combination of rise & fall

Will you manage to make progress, or will you just give up?

Figure 4. 26

Figure 4. 27

 

4.4 Measuring Tone                                                                                            Top

 

4.4.1. What is tone?                                                                                            Top

As mentioned in 4.3, tone is the pitch of our voice. In Chinese, a syllable or morpheme may have its own pitch, and the tone of each character contributes to lexical meaning while in English, a series of pitches sung over a whole utterance (intonation) is the major prosodic features to indicate the meaning (Teschner & Whitley, 2004, p.64).   Therefore, ‘pitch’ is mainly used to analyze the tone.

 

4.4.2 Tones in Putonghua                                                                                  Top

Tone is a contrastive lexical property of Standard Chinese. As a tone language, there are five tones in mandarin Chinese --the first tone, the second tone, the third tone, the fourth tone and the neutral tone. We can use the “pitch” in Praat to analyze the tones in Chinese.

 

The diagram below shows four tones in Putonghua.

Figure 4.28

 

Excerpt from http://web.mit.edu/jinzhang/www/pinyin/tones/

 

Let’s take “ma” in Putonghua for example, five tones represents five different characters which have different meanings

Figure 4. 29

Excerpt from http://www.misspandachinese.com/resources/tones-in-chinese/

 

Figure 4.30 Four Chinese tones

 

4.4.3 Tones in Cantonese                                                                                  Top

 

As we know that Cantonese is another tone language which has 9 tones (the last three tones share the same pitch with first, third and sixth tone respectively)

you can listen to the sounds via the following website http://www.fyan8.com/yuepin.htm

 

We can also use Praat to show the six tones in Cantonese.

Figure 4.31 Six Cantonese tones

Excerpt from http://bbs.cantonese.asia/thread-15985-1-1.html

 

 

4.4.4 Analyzing tones                                                                                          Top

 

Now let's discuss how we analyze tone using Praat. We will use a recording of the four Standard Chinese words as an example.

 

1. Created a WAV file with four Chinese words mā `mother', má `hemp', mǎ `horse', and mà `scold'

2. Selected the file and click View& Edit,

3. Pitchàtick Show pitch

 

The window is like this:

Figure 4.32

 

 

This window provides us with a graphic presentation of the variation in fundamental frequency (pitch contours) of the four Chinese tones

 

4. Adjust the pitch settings to make the display clearer.

 

I just briefed two important points here:

  • a. First, adjust the range of analysis as men have pitch ranges in the 50-180 Hz range while women's pitch tends to be higher, 80-250 Hz

  • b. Second, autocorrelation is preferred for intonation studies and cross-correlation for tone studies

(Please see 1.4.3 for more details on measuring pitch)

 

Figure 4.33

 

 

(Retrieved from Stoneham, 2011: p2, http://stonham.dyndns.org/phonetics/handouts/prosod_hndt.pdf)

 

5. Draw the pitch contours

  •  Select an area in Picture window using the pink frame

  •  Select Pitchàdraw visible pitch contour

  • See the results in Praat picture window

 Figure 4.34

(Please see 1.3.4 for more details on drawing pitch contours)

 

 

Questions for Chapter 4:                                                                              Top

 

1.      What stands for “stress” in Praat?

2.      Is the duration of the stressed syllable longer or shorter than the unstressed?

3.      Put the stress marker in “record” according the following two pictures.

 

 

record                                               record

 

4.      Which one is the noun of “contrast”?

 

 

                      contrast                                              contrast

 

5.  Mark the stressed syllables in each word with capital letters according to the pitch contour in the spectrogram and tell the reason in simple words.

 

  

                     Creativity                                                       Practical

 

 Japanese                                                   Beautiful

 

6.      Which sentence does the following picture belong to?

A.    I will go to the supermarket tomorrow.                        B. What’s your name?

C.    I like apples, cherries and bananas.                              D. Do you like this one?

 

 

7. How do you measure the VOT in Praat?

8. List the major types of linking and explain how you measure them in Praat.

9. What stands for “intonation” in Praat? How do you observe the intonation in Praat?

10. What’s the difference between rising and falling tone of tag questions?

 

(See the suggested answers at the end of the manual)

 

References

 

Styler, Will. (2014). Using Praat for Linguistic Research.

Stonham's lecture notes (2011), retrieved from http://stonham.dyndns.org/phonetics/handouts/prosod_hndt.pdf

http://www.fyan8.com/yuepin.htm

http://bbs.cantonese.asia/thread-15985-1-1.html

http://www.misspandachinese.com/resources/tones-in-chinese/

http://web.mit.edu/jinzhang/www/pinyin/tones/

 

 

Praat Manual Home