Docstoc

The role of F0 peaks in the identification of lexical stress in

Document Sample
The role of F0 peaks in the identification of lexical stress in Powered By Docstoc
					LLISTERRI, J.- MACHUCA, M. J.- de la MOTA, C.- RIERA,
M.- RÍOS, A. (2002) "The role of F0 peaks in the identification
  of lexical stress in Spanish", in BRAUN, A.- MASTHOFF,
H.R. (Eds.) Phonetics and its Applications. Festschrift for Jens-
 Peter Köster on the Occasion of his 60th Birthday. Stuttgart:
    Franz Steiner Verlag (Zeitschrrift für Dialektologie und
 Linguistik, Beiheft 121). pp. 350-361. ISBN: 3-515-08094-5

http://liceu.uab.es/~joaquim/publicacions/
        Llisterri_et_al_Koester.pdf
JOAQUIM LLISTERRI / MARÍA JESÚS M ACHUCA / C ARME DE LA M OTA / MONTSERRAT RIERA /
ANTONIO RÍOS

THE ROLE OF F0 PEAKS IN THE IDENTIFICATION OF LEXICAL STRESS IN SPANISH


1.- Introduction

It is a well documented fact that in several languages (see, for example, Silverman and
Pierrehumbert 1990 for American English) the fundamental frequency (F0) peak associated to a
lexically stressed syllable is not always coincident in time with this syllable. This phenomenon,
described as "F0 shift", "F0 peak displacement" or "peak delay" has been studied for Spanish
from an acoustical perspective (Garrido et al. 1993; Prieto et al. 1995; Llisterri et al. 1995), but
there seems to be a lack of experimental evidence on the perceptual effects of the left to right
shift of the F0 peak – "overshooting" - to syllables which are not lexically stressed.
      It is generally acknowledged that, as in other languages, stress in Spanish is signalled by
three simultaneous acoustic cues: fundamental frequency, duration and intensity. Some acoustic
studies have emphasized the importance of F0 (Quilis 1971), while others have argued for
duration as a highly relevant parameter in production (Canellada – Madsen 1987; Garrido et al.
1993; Mora et al. 1997; Díaz Campos 2000; Riera et al. 2001). As far as perception is
concerned, Enríquez et al. (1989) conclude that F0 is the main parameter systematically related
to the identification of the stressed syllable of a word, while duration is a secondary cue, also
conditioned by the stress pattern. Effects of other factors on the perception of accent such as the
phonological structure have also been observed (Face 2000).
      Taking advantage of the fact that Spanish is a free accent language – i.e. lexical stress can
appear in any syllable of the word – a perceptual experiment has been designed to assess the
role of F0 peaks in the identification of the stressed syllable, using words with the same
segmental form but with differences in stress placement, such as [»numeRo] (número – 'the
number', n.), [nu»meRo] (numero, 'I number', v.), [nume»Ro] (numeró, 'he/she numbered', v.). The
stimuli have been manipulated to observe the effects of the alignment of F0 peaks either with
the stressed syllable or with the post-stressed one.
      To assess the role of lexical knowledge phonologically acceptable but non-existent words
have also been included in the experiment. Contextual effects have been analysed by
considering the words both in isolation and in carrier sentences.

2.   Experimental design

A corpus for analysis has been built up and recorded by a native speaker of Castilian Spanish.
The recordings have been manipulated to obtain the stimuli for a perceptual test, as explained in
the next sections.




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
2                                                   Joaquim Llisterri et al.



2.1. Acoustic data
2.1.1. Corpus
The initial corpus consisted of four actual three syllable words allowing the stress to be placed
on the first (proparoxytone), second (paroxytone) and final (oxytone) syllable –
número/numero/numeró; límite/limite/limité; médico/medico/medicó; válido/valido/validó - and
two non-existent words in which the position of the stress has also been varied –
*núlibo/*nulibo/*nulibó; *ládebo/*ladebo/*ladebó . These words have been chosen to maintain
a constant CV syllable structure and to minimize the number of voiceless consonants to improve
F0 detection.
      A set of carrier sentences was also designed considering the stress pattern of the elements
preceding and following the target words, so as to obtain a constant number of two unstressed
syllables between lexical stressed ones; with this design, the same stress pattern is obtained for
every sentence, as shown in figure 1. For actual words, carrier sentences have been chosen
according to the lexical category of the target word, while non-existent words were combined
with all possible carrier sentences.
      Proparoxytone target words – e.g. [»numeRo] número – were preceded by four syllables
with stress on the second syllable – e.g. [le»DiTesel] le dices el ('you tell him the…') – and
followed by a stressed monosyllabic word – e.g. [»mal] mal ('wrongly')-. Paroxytone target
words – e.g. [nu»meRo] numero – were preceded by a sequence of stressed plus unstressed
syllable – e.g. [»nole] no le ('you don't/didn't…')– and followed by a 3 syllable noun phrase with
stress on the second one – e.g. [las»filas] las filas ('the rows')-. Finally, oxytone target words –
e.g. [nume»Ro] numeró – were preceded by a stressed monosyllabic word – e.g. [»no] no (no) –
and followed by a 4 syllable noun phrase with stress on the penultimate syllable – e.g.
[loz»Boletos] los boletos ('the tickets').

Words with stress on the first syllable: número
U    S      U        U    S        U       U    S
Le   di     ces      el   nú       me      ro   mal
                                Noun

          Verb                Object                Adverb

'You tell him the number wrongly'

Words with stress on the second syllable: numero
S               U              U       S    U   U       S U
No              le             nu me ro las fi las
Negation         Pronoun            Verb              Object

'I don’t number the rows for him'



LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                        The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                         3



Words with stress on the final syllable: numeró
S             U     U      S     U       U     S     U
No            nu    me     ró    los     bo    le tos
 Negation          Verb                  Object

'He didn't number the tickets'

Figure 1: Stress pattern and syntactic structure of the carrier sentences. S stands for "stressed
syllable" and U for "unstressed syllables".

Both the isolated words and those embedded in the carrier sentences were recorded in
acoustically controlled conditions by a native speaker of Castilian Spanish. Each word was
recorded 10 times in isolation and carrier sentences with embedded target words were recorded
only once. Isolated words were used to obtain mean F0 values for the creation of the set of basic
stimuli for the perceptual test.

2.1.2. Acoustic analysis
The F0 in the 180 target words (10 repetitions of 4 actual words x 3 stress patterns + 2 non-
existent words x 3 stress patterns) was measured at the beginning, centre and end of each of the
three vowels in the word using the Praat software. Results, presented in Appendix 1, show that
for isolated existent and non-existent words higher F0 values are regularly associated with the
lexically stressed syllable.

2.2. Perceptual tests
2.2.1. Stimuli
The F0 of the isolated target words was manipulated using the Praat programme. First of all, for
each of the target words as recorded by the speaker, F0 original values in the vowels have been
replaced by mean values obtained from the 10 repetitions of the same word (see Appendix 1).
Duration and intensity were maintained as in the original recordings.
     Two further modifications were performed to obtain stimuli with F0 displacement. In
words with lexical stress in the first syllable (proparoxytones), the initial, medial and final F0
values for each vowel were replaced by the F0 values found in words with stress on the second
syllable (paroxytones). Moreover, in words with lexical stress on the second syllable, the F0
values for each vowel were changed by the values found in words with lexical stress on the final
syllable (oxytones). Duration and intensity parameters were not modified. To create the new
stimuli, each word was resynthesised with the new pitch contour using PSOLA as implemented
in Praat. Figures 2 and 3 show an example of this manipulation. Oxytone words were not
manipulated to avoid shifting an F0 peak across a word boundary.




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
4                                         Joaquim Llisterri et al.




    (a) Válido [»baliDo] with original F0                (b) Válido [»baliDo] with F0 contour
        contour                                              extracted from valido[»baliDo]


Figure 2: Waveform, F0 (black line) and intensity (grey line) for the word válido (a) and
waveform, F0 and intensity obtained after manipulation superimposing the F0 pattern of the
word valido (b).




    (a) Valido [ba»liD o] with original F0                  (b) Valido [ba»liD o] with F0 contour
        contour                                                 extracted from validó [bali»Do]


Figure 3: Waveform, F0 (black line) and intensity (grey line) for the word valido (a) and
waveform, F0 and intensity obtained after manipulation superimposing the F0 pattern of the
word validó (b).


2.2.2. Subjects and method
Three different tests were prepared. In the first one, subjects were asked to identify the syllable
in which the stress was perceived – first, second or last syllable – for a total of 30 isolated
words. The second test consisted of 24 pairs of isolated words, and subjects were asked wether
the words were equal or different in their stress pattern. In the third test, consisting on 84
stimuli, the target words were embedded in the carrier sentences described in 2.1.1 and subjects




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                          The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                                          5

had to indicated in which syllable – first, second or last – of the target word the stress was
perceived. Within each test, stimuli were presented in random order.
      The test was administered through individual headphones at the Language Laboratory of
the Department of French and Romance Philology at the Universitat Autònoma de Barcelona.
Subjects were given written instructions on paper as well as an oral briefing, and they were
warned of the presence of existent and non-existent words on the test. A set of five training
stimuli was included at the beginning of each test, and questions on the procedure were taken
after each training period. No blank replies were allowed.
      Twenty first year students of Spanish Philology at the Universitat Autònoma de Barcelona,
speakers of the Castilian variety of Spanish, with ages between 18 and 22 years old, responded
to the test. A total of 2760 replies were obtained.

3.        Results

3.1. Stimuli without F0 peak shift
As explained in 2.2.2. , in test 1 subjects were asked to identify the syllable in which the stress
was perceived in a series of 30 isolated words. In test 3, these same words were embedded in
the carrier sentences described in 2.1.1. Stimuli were prepared using the mean F0 values
obtained from the recordings shown in appendix 1. Note that in those cases F0 peaks are
simultaneous with the syllable bearing the lexical stress. In all cases, existent and non-existent
words were included.
     Tables 1 to 4 describe the average percentage of identification of the stress pattern of the
stimuli.

                                                                            % of replies
                                              Stress on the first          Stress on the            Stress on the final
                                                   syllable               second syllable                syllable
          Stress on the first syllable              97.37                      2.63                          0
          (e.g. [»baliDo])                          (3.04)                    (3.04)
Stimuli




          Stress on the second syllable              1.32                      96.05                       2.63
          (e.g. [ba»liDo])                          (2.63)                    (5.04)                      (5.26)
          Stress on the final syllable                 0                         0                         100
          (e.g. [bali»Do])

Table 1: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in isolated existent words




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
6                                           Joaquim Llisterri et al.



                                                                         % of replies
                                          Stress on the first           Stress on the    Stress on the final
                                               syllable                second syllable        syllable
          Stress on the first syllable           100                          0                   0
          (e.g. [»laDeBo])
Stimuli




          Stress on the second syllable          2.63                      97.37                   0
          (e.g. [la»DeBo])                      (3.72)                     (3.72)
          Stress on the final syllable            0                          0                    100
          (e.g. [laDe»Bo])

Table 2: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in isolated non-existent words


                                                                         % of replies
                                          Stress on the first           Stress on the    Stress on the final
                                               syllable                second syllable        syllable
          Stress on the first syllable           100                          0                   0
          (e.g. [»baliDo])
Stimuli




          Stress on the second syllable          1.32                      98.68                  0
          (e.g. [ba»liDo])                      (2.44)                     (2.44)
          Stress on the final syllable            0                         7.89               92.11
          (e.g. [bali»Do])                                                (18.45)             (18.45)

Table 3: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in existent words embedded in carrier sentences


                                                                         % of replies
                                         Stress on the first            Stress on the    Stress on the final
                                              syllable                 second syllable        syllable
          Stress on the first syllable         99.34                        0.66                  0
          (e.g. [»laDeBo])                     (1.86)                      (1.86)
Stimuli




          Stress on the second syllable         3.62                        96.38                  0
          (e.g. [la»DeBo])                     (3.71)                      (3.71)
          Stress on the final syllable 1.32                  0                           98.68
          (e.g. [laDe»Bo])              (2.35)                                           (2.35)

Table 4: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in non-existent words embedded in carrier sentences




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                      The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                                 7

As can be observed in Tables 1 to 4, no difficulties in the correct identification of the lexically
stressed syllable are appreciated for the stimuli in which the F0 peak is simultaneous with the
lexically stressed syllable. Results are very similar for actual and for nonsense words, and do
not seem to be influenced by the presence or absence of the carrier sentence.

3.2. Stimuli with manipulated F0 peak shift
For stimuli with manipulation of the F0 contour two types of tests were prepared: one in which
subjects were asked to identify the syllable bearing the lexical stress, both in words in isolation
(Test 1) and in context (Test 3), and another one in which pairs of words were compared (Test
2).

3.2.1. Identification of lexical stress
Stimuli used for these two tests were produced as described and exemplified in 2.2.1: the F0
contour of a paroxytone word (e.g. [ba»liDo]) was superimposed to a proparoxytone word
([»baliDo]) and the F0 contour of an oxytone word ([bali»Do]) was superimposed to a
paroxytone one ([ba»liDo]), without modifications in duration and intensity. The results obtained
for the perceptual identification of the stressed syllable are presented in tables 5 to 8.

                                                                               % of replies
                                                        Stress on the         Stress on the     Stress on the
                                                        first syllable       second syllable    final syllable
     Lexical stress on the first syllable and               61.84                 36.84              1.32
     F0 contour with peak on the second                     (30.8)               (31.87)            (2.63)
Stimuli




     syllable
     Lexical stress on the second syllable             1.32      60.53         38.16
     and F0 contour with peak on the final            (2.63)     (34.8)        (33.7)
     syllable
Table 5: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in existent words in isolation


                                                                               % of replies
                                                        Stress on the         Stress on the     Stress on the
                                                        first syllable       second syllable    final syllable
     Lexical stress on the first syllable and               72.37                 27.63                0
     F0 contour with peak on the second                    (20.78)               (20.78)
Stimuli




     syllable
     Lexical stress on the second syllable             1.32     67.11          31.58
     with F0 contour with peak on the final           (2.63)   (45.35)        (46.08)
     syllable
Table 6: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in existent words in context




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
8                                             Joaquim Llisterri et al.



For existent words, isolated or in context, the syllable perceived as stressed is the one bearing
the lexical stress before the manipulation, with percentages of identification ranging from
72.37% to 60.5%. Nevertheless, a percentage of replies between 27.63% and 38.16% shows that
the stress pattern that has been perceived is influenced by the F0 contour which has been
superimposed, i.e. that the syllable identified as stressed is the one bearing the F0 peak.

                                                                           % of replies
                                                        Stress on the     Stress on the    Stress on the
                                                        first syllable   second syllable   final syllable
          Lexical stress on the first syllable and          23.68             68.42             7.89
          F0 contour with peak on the second               (33.49)           (37.21)           (3.72)
Stimuli




          syllable
          Lexical stress on the second syllable                 0            63.16            36.84
          and F0 contour with peak on the final                              (44.7)           (44.7)
          syllable

Table 7: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in non-existent words in isolation


                                                                           % of replies
                                                        Stress on the     Stress on the    Stress on the
                                                        first syllable   second syllable   final syllable
          Lexical stress on the first syllable and          61.18             38.82               0
          F0 contour with peak on the second               (21.41)           (21.41)
Stimuli




          syllable
          Lexical stress on the second syllable               1.97           67.11             30.92
          and F0 contour with peak on the final              (3.92)         (31.67)           (33.58)
          syllable

Table 8: Mean percentage and standard deviation (in parentheses) of identification of the
position of the stressed syllable in non-existent words in context


For non-existent words, the behaviour is quite similar to the one found for actual words except
in the case of isolated words which were originally stressed on the first syllable (proparoxytone)
and have been modified with an F0 contour corresponding to a word with the stress on the
second syllable (paroxytone). In this case, the manipulated stimulus is perceived with the
original lexical stress in 23,68% of cases, while it is perceived with stress in the syllable in
which the F0 peak actually occurs in 68,42% of the cases; in 7,89% of the replies, the stress has
been perceived on the last syllable.
     A closer analysis of the stimuli reveals certain differences between the two non-existent
words [»laDeBo] and [»nuliBo] in the cases in which the paroxytone F0 contour has been


LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                      The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                              9

superimposed over the proparoxytone contour: [»laDeBo] in isolation is perceived as
proparoxytone ([»laDeBo]) in 47.37% of cases and as paroxytone ([la»DeBo]) in 42.11% of cases,
while the same word included in a carrier sentence it is mainly interpreted as a proparoxytone
([»laDeBo]) (78.94% of cases). The word [»nuliBo] in isolation is clearly identified as paroxytone
([un»liBo]) (94.74% of cases), but results when embedded in carrier sentences range from
68.42% to 42.11% of cases in which the word has been perceived as paroxytone ([nu»liBo]).

3.2.2. Comparison of paired words
Pairs of isolated words with the manipulations described in 2.2.1 were prepared as stimuli for
test 2. Subjects were asked to label the pairs as "same" or "different" taking into consideration
the position of the stress in the word. The results obtained are summarised in tables 9 and 10.

                            Lexical stress on the           Lexical stress on the  Lexical stress on the
                             first syllable and F0         second syllable and F0 final syllable and F0
                            contour with peak on            contour with peak on contour with peak on
                               the first syllable             the second syllable   the final syllable
                               (proparoxytone)                   (paroxytone)           (oxytone)
                           Same         Different          Same        Different  Same       Different
Lexical stress on the        53.95         46.05             22.37        77.63
first syllable              (18.4)         (18.4)            (21.2)       (21.2)
(proparoxytone) and
F0 contour with peak
on the second syllable
(paroxytone)
Lexical stress on the                                        65.79           34.21           9.21    90.79
second syllable                                              (20.4)          (20.4)         (5.04)   (5.04)
(paroxytone) and F0
contour with stress on
the final syllable
(oxytone)

Table 9: Mean percentage and standard deviation (in parentheses) of replies for existent words


When proparoxytone words are compared with proparoxytone ones with a superimposed F0
contour of a paroxytone, they do not seem to be clearly identified as words with the stress on the
first syllable (53.95%) or as words with the stress on the second syllable (46.05%). In the case
of paroxytone words, when compared with proparoxytones with the superimposed contour of a
paroxytone one, they are interpreted as having different stress patterns in 77.63% of cases. If
these same stimuli are compared with paroxytone words with the superimposed contour of
oxytone ones, they are perceived as having the stress on the second syllable in 65.79% of cases.
Finally, when oxytone words are compared with paroxytone ones with a superimposed oxytone
contour, in 90.70% of cases they are perceived as different.
      It seems that the paroxytone stress pattern is better perceived as such in cases where an
oxytone F0 contour has been superimposed over a paroxytone word (65.79%) than in cases

LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
10                                        Joaquim Llisterri et al.



where a paroxytone F0 contour has been superimposed over a proparoxytone word (22.37%).
Even oxytones compared with paroxytones with an oxytone F0 contour are not perceived as
being the same stimulus in 90.79% of cases. Thus, the original stress pattern of existent words
seems to be prevalent over the superimposed F0 contour.

                           Lexical stress on the   Lexical stress on the  Lexical stress on the
                            first syllable and F0 second syllable and F0 final syllable and F0
                           contour with peak on    contour with peak on   contour with peak on
                               the first syllable   the second syllable     the final syllable
                              (proparoxytone)          (paroxytone)             (oxytone)
                          Same          Different Same       Different   Same        Different
Lexical stress on the       42.11          57.89    55.26        44.75
first syllable              (52.1)         (52.1)  (33.49)      (33.49)
(proparoxytone) and
F0 contour with peak
on the second syllable
(paroxytone)
Lexical stress on the                                   52.63        47.37         15.79        84.21
second syllable                                         (59.5)       (59.5)       (22.33)      (22.33)
(paroxytone) and F0
contour with stress on
the final syllable
(oxytone)

Table 10: Mean percentage and standard deviation (in parentheses) of replies for non-existent
words


For non-existent words, proparoxytone words compared with proparoxytone ones with a
superimposed F0 contour of a paroxytone do not seem to be clearly identified as proparoxytones
(42.11%) or as paroxytones (57.89%). In the case of paroxytone words, when compared with
proparoxytones with the superimposed F0 contour of a paroxytone one, results are not
conclusive again, since they are identified as having the stress on the first syllable (55.26%) or
in the second one (44.75%). If these same stimuli are compared with paroxytone words with the
superimposed contour of oxytone ones, a similar behaviour is found: they are not clearly
interpreted as paroxytones (52.63%) or oxytones (47.37%). Finally, when oxytone words are
compared with paroxytone ones with a superimposed oxytone contour, in 84.21% of cases they
are perceived as different.
     Non-existent words only seem to follow the same trend as existent ones in two cases: (a)
comparisons between proparoxytones and proparoxytones with a superimposed paroxytone F0
contour, and (b) comparisons between oxytones and paroxytones with a superimposed oxytone
F0 contour.



LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                      The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                         11

4.   Conclusions

As results presented in 3.1. clearly show, when the F0 peak is aligned with the lexically stressed
syllable, the position of the stress is correctly identified with percentages of replies ranging from
100% to 92.11%; this is found for existent and for non-existent words in isolation as well as in
carrier sentences. These percentages are in agreement with the 89.59% of correct perception of
lexical stress in a corpus of spontaneous speech reported by Mora et al. (1997) in an experiment
involving Spanish speaking subjects.
     In those cases in which the F0 contour has been replaced by a different one (3.2.1), the
general trend seems to be the identification as stressed of the syllable originally bearing the
lexical stress, despite the presence a superimposed F0 contour with a peak displaced to the right.
Percentages for those cases range from 72.37% to 60.53%. This general trend is not followed by
isolated proparoxytone non-existent words with a paroxytone F0 contour, where an important
variability in the replies to different stimuli has been observed.
     When pairs of words are compared (3.2.2), paroxytone words to which an F0 contour with
a peak on the final syllable has been superimposed are clearly not perceived as being the same
as an oxytone word. In existent words, the result of this manipulation is interpreted as a word
bearing the stress on the second syllable (65.79% of cases) – i.e. the originally stressed one -,
while in non-existent words a clear tendency is not found.
     For the stimuli with superimposed F0 contours no systematic effects of the context and of
the differences between actual and non-existent words has been observed, except in isolated
cases discussed in 3.2.1 and 3.2.2.
     In this experiment the role of F0 peak displacement in the perception of lexical stress in
Spanish has been considered. The stimuli with shifted peaks included in the tests were obtained
by replacing the initial, medial and final F0 values of all the vowels in the target words, while
intensity contours and segmental durations were kept as in the original corpus produced by a
native speaker. The general tendency observed in the results indicates that a replacement of the
F0 contour is not enough to induce the identification as stressed of the syllable aligned with the
F0 peak.
     Further research will apply the same strategy to intensity contours and to segmental
durations, first considering each acoustic correlate in isolation and secondly taking into account
different combinations of F0, duration and intensity.


References

C ANELLADA , María Josefa / M ADSEN , John Kuhlmann (1987): Pronunciación del español.
     Lengua hablada y literaria. Madrid.
D ÍAZ C AMPOS, Manuel (2000): The phonetic manifestation of secondary stress in Spanish. In:
     C AMPOS , H. / H ERBURGER , E. / MORALES F RONT, A. / WA L S H, T.J. (eds.): Hispanic
     Linguistics at the Turn of the Millennium. Papers from the 3rd Hispanic Linguistics
     Symposium. Somerville, MA, pp. 49-65.
E NRÍQUEZ, Emilia V. / CASADO, Celia / SANTOS, Andrés (1989): La percepción del acento en
     español. In: Lingüística Española Actual 11, pp. 241-269.


LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
12                                        Joaquim Llisterri et al.



F ACE , Timothy (2000): The role of syllable weight in the perception of Spanish stress. In:
     C AMPOS , H. / HERBURGER , E. / MORALES FRONT, A. / WALSH, T.J (eds.): Hispanic
     Linguistics at the Turn of the Millenium. Somerville, MA, pp. 1-13.
GARRIDO, Juan M. / LLISTERRI, Joaquim / DE LA MOTA, Carme / RÍOS, Antonio (1993):
     Prosodic differences in reading style: Isolated vs. Contextualized Sentences. In:
     EuroSpeech '93. 3rd European Conference on Speech Communication and Technology.
     Berlin, vol. 1, pp. 573-576.
     http://liceu.uab.es/~joaquim/publicacions/Berlin_93/berlin_93.html
LLISTERRI, Joaquim / M ARÍN , Rafael / DE LA M OTA , Carme / RÍOS, Antonio (1995): Factors
     affecting F0 peak displacement in Spanish. In: EuroSpeech'95. 4th European Conference
     on Speech Communication and Technology. Madrid, vol. 3, pp. 2061-2064.
     http://liceu.uab.es/~joaquim/publicacions/eurospeech_95.html
M ORA , Elsa / COURTOIS , Fabienne / CAVÉ , Christian (1997): Étude comparative de la
     perception par des sujets francophones et hispanophones de l'accent lexical en espagnol. In:
     Revue Parole 1, 75-86.
PRIETO, Pilar / VAN S ANTEN , Jan / HIRSCHBERG, Julia (1995): Tonal Alignment Patterns in
     Spanish. In: Journal of Phonetics 23, 429-451.
QUILIS, Antonio (1971): Caracterización fonética del acento español. In: Travaux de
     Linguistique et de Littérature 9, 53-72.
RIERA, Montserrat / RÍOS, Antonio / DE LA M OTA, Carme / CARBÓ, Carme / MACHUCA, María
     J. (2001): Acento secundario y complejidad morfológica. In: II Congreso de Fonética
     Experimental. Sevilla, pp.
S ILVERMAN , im. / PIERREHUMBERT, Janet (1990): The timing of prenuclear high accents in
     English. In K I N G S T O N , J. / B ECKMAN , M. (eds.) Papers in Laboratory Phonology.
     Cambridge, pp. 72-106.

Appendix 1

Mean F0 values in Hz (standard deviation in parenthesis) as measured for the reference speaker
in 10 repetitions of the isolated target words and used in the non manipulated stimuli included in
the perceptual test. V1, V2 and V3 refer to the first, second and third vowel in the word; I, M
and F stand for the initial, medial and final portions of the vowel where F0 has been measured.
Values for the lexically stressed syllable in each word, as well as the syllable in the orthographic
form, appear in italics.

          V1_I       V1_M       V1_F      V2_I        V2_M           V2_F     V3_I     V3_M     V3_F
Limite    80.60      80.50      78.80     88.60       89.90          84.70    82.60    80.80    80.60
          (2.46)     (2.99)     (3.12)    (2.50)      (2.02)         (4.37)   (3.50)   (4.44)   (4.03)
Limité    79.50      79.00      77.80     77.20       77.90          79.40    89.60    80.10    87.90
          (3.41)     (3.02)     (2.10)    (2.57)      (2.92)         (3.57)   (5.21)   (4.86)   (10.47)
Límite    91.10      92.80      87.60     77.20       76.00          75.80    79.00    77.20    77.80
          (5.45)     (4.71)     (3.20)    (2.44)      (1.70)         (1.03)   (3.46)   (2.10)   (2.53)



LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf
                      The role of F0 peaks in the identification of lexical stress in Spanish
                                                                                                                   13

Medico 83.00         80.30       77.60       89.90       92.00        86.20       82.90         82.20    81.60
       (3.94)        (2.63)      (2.37)      (2.51)      (3.37)       (4.57)      (6.28)        (6.05)   (6.13)
Medicó 80.80         77.20       76.20       77.00       76.50        77.50       86.10         81.20    88.40
       (3.58)        (1.81)      (0.92)      (2.05)      (2.12)       (2.27)      (3.98)        (5.05)   (13.54)
Médico 91.70         91.60       82.90       78.60       78.10        76.70       77.50         77.70    79.40
       (4.22)        (3.60)      (3.81)      (2.84)      (2.18)       (2.06)      (2.17)        (2.26)   (2.84)
Numero 84.89         83.11       80.56       87.44       87.56        80.78       77.56         77.00    79.78
       (2.85)        (2.71)      (2.46)      (2.55)      (2.01)       (2.05)      (2.19)        (2.45)   (3.49)
Numeró 83.80         82.90       79.80       78.90       77.30        76.80       82.80         81.00    88.40
       (3.46)        (3.18)      (2.53)      (2.13)      (1.57)       (1.23)      (2.53)        (4.71)   (12.58)
Número 96.80         96.60       92.70       79.60       77.20        76.40       76.70         76.50    80.20
       (4.94)        (4.22)      (3.83)      (2.41)      (2.39)       (1.58)      (1.06)        (1.27)   (7.18)
Valido 80.10         79.80       79.00       87.60       89.90        84.40       76.50         76.90    77.80
       (3.11)        (1.48)      (1.70)      (2.91)      (2.51)       (1.58)      (1.35)        (1.66)   (1.93)
Validó 79.30         76.70       75.90       75.90       75.90        76.10       81.90         81.30    89.50
       (3.95)        (1.95)      (1.37)      (0.88)      (0.57)       (0.74)      (3.11)        (5.06)   (12.97)
Válido 84.60         89.60       84.10       76.20       76.40        75.80       77.30         77.50    78.30
       (2.72)        (4.79)      (3.93)      (1.03)      (1.17)       (1.03)      (1.89)        (2.42)   (2.50)
Nulibo    84.50      82.60       80.60       90.00       92.10        86.70       76.80         77.50    78.10
          (2.46)     (2.32)      (1.65)      (2.98)      (3.31)       (3.77)      (0.92)        (0.71)   (1.73)
Nulibó    82.80      80.20       76.50       76.30       76.10        76.30       79.70         82.20    91.10
          (1.32)     (1.40)      (0.85)      (1.77)      (1.45)       (1.57)      (2.63)        (6.36)   (15.32)
Núlibo    97.40      97.20       91.40       77.00       76.40        75.60       75.80         76.00    78.70
          (3.66)     (3.55)      (4.43)      (2.36)      (1.43)       (0.97)      (1.62)        (1.89)   (4.90)
Ladebo    80.30      79.10       78.00       87.30       86.00        81.50       77.80         78.40    79.80
          (2.16)     (2.51)      (1.49)      (1.49)      (2.26)       (2.92)      (2.82)        (3.89)   (5.27)
Ladebó    78.70      76.50       78.60       78.10       77.40        77.30       77.70         78.70    82.30
          (1.70)     (1.08)      (2.88)      (2.88)      (2.95)       (2.95)      (2.75)        (6.65)   (13.16)
Ládebo    88.30      87.20       84.20       80.00       75.80        75.80       76.40         77.20    78.20
          (4.40)     (4.52)      (3.74)      (2.87)      (1.03)       (0.63)      (1.43)        (1.62)   (2.74)




LLISTERRI, J.- MACHUCA, M.J.- de la MOTA, C.- RIERA, M.- RÍOS, A. (2002) "The role of F0
peaks in the identification of lexical stress in Spanish", in BRAUN, A.- MASTHOFF, H.R. (Eds.)
Phonetics and its Applications. Festschrift for Jens-Peter Köster on the Occasion of his 60th Birthday.
Stuttgart: Franz Steiner Verlag (Zeitschrrift für Dialektologie und Linguistik, Beiheft 121). pp. 350-361.
http://liceu.uab.es/~joaquim/publicacions/Llisterri_et_al_Koester.pdf

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:6
posted:6/3/2011
language:English
pages:14