Groningen 30 januari 2004
Workshop
On the Boundaries of Phonology and Phonetics
Phonetics in Phonology
Maartje Schreuder & Dicky Gilbers
Phonetics vs.
Phonology
Both the phonetician and the phonologist are
interested in speech sounds
Dutch /r/-varieties
• /r/ can be realized as [r] or as [ ]
• phonetics: two different sounds:
• [r] is produced as a trill of the tip of the tongue
against the alveolar ridge
• [ ] on the other hand is produced as a trill of
the uvula in the back of the mouth
Outline
Topic: the role of phonetics in phonology
1 – Phonology without Phonetics:
Schwa-insertion in syllable-final consonant
clusters
2 – Interaction of Phonetics and Phonology:
The realisation of liquids in first language
acquisition data
3 – Problematic Phonetic Analysis:
Rhythmically-variable Patterns
Phonetics vs. Phonology
Phoneticians study the physical properties of sound making
Phonologists are interested in
the sound system of a language
[ala] They want to know
• the function of sounds in the language
(what are the phonemes?)
• the way sounds can be combined
• how the sounds are realized in different
contexts
Dutch /r/-varieties
• /r/ can be realized as [r] or as [ ]
• phonology: same segment; no functional
(semantic) difference in the system between
[r
t] and [
t].
• In the system /r/ is coronal, just like /t/ and
/s/
1
 1 First Process: Example: Schwa insertion in
Schwa insertion in final final consonant clusters
consonant clusters data: harp, darm, durf, borg, hark
data: harp, darm, durf, borg, hark no insertion if the consonants share their
place of articulation: hars, hart, mars
r m/k
r m/k r t

place α place β
place α place β place α

Example: Schwa insertion in

final consonant clusters Conclusion
data: harp, darm, durf, borg, hark Synchronically, the process can only be
described in a phonological way, even though
no insertion if the consonants share their
it may have had a phonetic - articulatory - base
place of articulation: hars, hart, mars originally

From a phonetic point of view uvular [ ] and alveolar [t]

do not share their places of articulation, from a
Probably, the uvular [ ௎] is a later variant of
phonological point of view they do. Yet, schwa Dutch /r/ than coronal [r]
insertion is also impossible between [ ] and [t].

2 Second Process: Feature Changes /l/ → [w]

/l/-substitutes in First /l/ → [w]
Language Acquisition Data +son +son
+cons -cons
+cont +cont
+lat -lat
-lab +lab
+ant -ant
+cor -cor
-high +high
-back +back
/l/ → [w] -round +round

2
 Feature Changes /l/ → [w] Data
/l/ → [w]
+son +son • Smith (1973) label [we
b]
left [wpt]
+cons -cons
rabbit [wæbt]
+cont +cont
+lat -lat
• Fikkert (1994) klok [kwk]
What’s wrong with the
-lab +lab
trein [tjin]
+ant -ant
phonologically-based account
+cor -cor
• Van der Linde (1998) lief [wif]
of the gliding process?
-high +high
krokodil [kokodw]
-back +back
hallo [hjo]
-round +round

The Naturalness of a
Common Process Liquids and Glides Perceptively
(Ainsworth and Paliwal, 1984)
• McCarthy (1988, p.86): A common process (...) is
accomplished by an elementary operation of the typical set of responses obtained from listening to
theory. [An uncommon process is far more complex to glide/liquid-vowel synthetic stimuli (simplified)
state].
3160 Hz w w w l l l l j j j
• Gliding is a widely attested phonological process, but it ↑ w w w l l l l j j j
has to be accounted for phonologically as a very F3 locus freq. w w w r r r l j j j
marked process, because almost all features change
value. ↓ w w w r r r j j j j
1540 Hz w w r r r r r j j j
760 Hz ← F2 locus freq. → 2380 Hz
• What is the phonetically-based account
of gliding?

Sonorant Consonants
Acoustically
Lehiste (1964): F1 and F2 of [l] and [w]
coincide to a large extent. F2 for [w] is
somewhat lower than F2 of [l]. F2 of [j] is
somewhat higher than F2 of [l].
Conclusion
From an acoustic point of view, liquid-glide
alternations can be described as minimal
changes from the target segment
Obviously, phonology needs the phonetic
information to explain this process as an
unmarked (widely attested) one

3
 3 Third Process: Language: Re-/misinterpretation of
rhythm in accelerated or sloppy speaking
Rhythmic Variability
• Topic: Rate adjustments: Data
compression or type 1: stú die tòe la ge → stú die toe là ge
restructuring? ‘study grant’
Rightward Stress Shift

• Experiment: Eliciting fast speech

Phonetic evidence?

Language: Re-/misinterpretation of Language: Re-/misinterpretation of

rhythm in accelerated or sloppy speaking rhythm in accelerated or sloppy speaking

Data Data
type 1: stú die tòe la ge → stú die toe là ge type 1: stú die tòe la ge → stú die toe là ge
‘study grant’ ‘study grant’

type 2: per fèc tio níst → pèr fec tio níst type 2: per fèc tio níst → pèr fec tio níst

Leftward Stress Shift

type 3: zùid à fri káans → zùid a fri káans
‘south african’
Beat Reduction

Observation
Phonetic Evidence?

Is there phonetic evidence for rhythmic

There are different grammars for different
restructuring in allegro speech?
rates and styles of speaking
Phonetic Compression is not the sole explanation
Stress Correlates:

1. Duration
In fast speech it is more important to avoid 2. Pitch
clashes. 3. Intensity
4. Spectral Balance (Sluijter, 1995)
5. Rhythm Factors (Quené & Port, 2002 a.o.)

4
 variable stress patterns Acoustic Analysis
Picture (PRAAT)
1. Duration
stu die toe la ge stu die toe la ge duration rhymes
(1.164 ms) (0.85 ms)
0.16
0.14
multi-plying factor 1.37 0.12 andante c1

seconds
0.1 andante c2
[u] [a] [u] [a] 0.08
0.06 allegro c1
pitch 116.8 Hz 99.5 Hz pitch 117.3 Hz 99.8 Hz 0.04 allegro c2
0.02
duration 0.101 ms 0.169 ms duration 0.056 ms 0.196 ms 0
intensity 85.7 dB 80.7 dB intensity 80.7 dB 78.5 dB

en

it
t

rij

t
s

ei
aa

ite
te
ni

er

lit
tio

ra

l
ik

na

i
is

vir
er

pi

al
c

ba
rfe

am

rb
pe

ve
item
Subject 1 Left Shifts
(from a pre-study by the
authors)

Acoustic Analysis Acoustic Analysis

4. Spectral Balance 4. Spectral Balance
A ndante: unstressed [a] A llegro: stresse d [a]
stressed and u n stressed [a]
25 25

20 20
60

15 15 50
p h on

andante
10 10 40
allegro
5 5
30
0
0
T im e (s)
0.169371
0
0
Time (s)
0.143209 20
13
17
21
25
1
5
9

studietoel[a]ge (Andante) studietoel[a]ge (Allegro) b ark

(from a pre-study by the
authors)

Acoustic Analysis Acoustic Analysis

4. Spectral Balance 4. Spectral Balance

stressed and unstressed [e] in Stressed and unstressed [e] in

Stressed and unstressed [a] in Stressed and unstressed [e] in per[fec]tionist
[per]fectionist
[a]merikaan a[me]rikaan
50 60 60 40
40 30
Phon

andante
Phon

40 andante
40 20
Phon

Phon

30 andante andante 20 allegro allegro

10
20 allegro 20 allegro 0
0
10
10
13

16
19

22
25
1

4
7

4
7
10
13

16
19

22
25

0 0 Bark Bark
10
13
16
19
22
25

10
13
16
19
22
25
1
4
7

1
4
7

Bark Bark

Subject 1 Left Shifts Subject 1 Left Shifts

5
 Acoustic Analysis Acoustic Analysis
5. Rhythm Factors 5. Rhythm Factors
Equal Spacing Constraint: Right Shifts beat intervals andante:allegro

Prominent vowel onsets are attracted to 1.1

1
periodically spaced temporal locations

quotie nt
0.9
0.8
quotient
(Couper-Kuhlen 1993; Cummins & Port 1998; Quené & 0.7
0.6
Port 2002) 0.5

...
g

r
e

g
te
e
studietoelage (Andante): 0.358 sec

...

lin
ag

f in
ch
im
itt
s

de
el

ef
ns
u
an

or
to

kr
rin

ph
ei
pa

vo
ie

er

nt
te

lo
studietoelage (Allegro): 0.328 sec factor: 1.091

ud

er

er

yw

ke
ee
es
st

m
w

in
m

bb
ka
eg

tr i

ge

w
(from a pre-study by the

ho
w
authors) Subject 1 Right Shifts

Acoustic Analysis Conclusion

5. Rhythm Factors
0.9 R igh t S h ifts B e at Inte rv als P 1 • The rhythmic differences as a result of different speech
0.8 rates could not be demonstrated acoustically
beat intervals in sec

0.7
0.6 andante short I
0.5 andante long I
0.4
• A systematic correlate of secondary stress still needs to
allegro long I be found
0.3
0.2 allegro short I
0.1
0 • Apparently, not all phonological processes can
be tested empirically by phonetic analysis
e

ng

te

g
er

Expected combinations
g
itte
lag

fin
in
uim
ek

eli

rijv
rz
oe

ef
ind
st

kr

in words 1, 4 and 7
oo

ph
ch
an
iet

er
er

rv

elo
ud

pa

ins
st

yw
me
st

me

ink
er

te
bb
ka

en

Attested combination
w

tri

w
ho
eg

me
w

ge

in word 2 and 6