Phonetic variability of the Greek rhotic sound

1
Phonetic variability of the Greek rhotic sound
Mary Baltazani University of Ioannina, Greece
ltmbaltaz_at_cc.uoi.grgt
Results Experiment 1 VrV
Introduction

• 2.1 What factors influence the appearance of the
  epenthetic V fragment?
• An effect of the position of C was found (Cr-rC)
• All Cr clusters are realized with epenthetic V
  fragment
• Most, but not all, rC clusters have an
  epenthetic V fragment (Table 4)

• Rhotics exhibit considerable phonetic
• variety cross-linguistically (Lindau 1985,
• Inouye 1995, Ladefoged and Maddieson 1996).
• Different prosodic and phonetic contexts
• have been found to affect their realization
• across languages (cf Recasens 1991,
• Recasens et al 1993).
• The situation for Greek rhotics
• There is no exclusive acoustic study on the
• characteristics of the Greek rhotic. General
• studies on Greek consonants describe the
• rhotic in passing as a tap (Arvaniti 1999,
• Nicolaidis 2001). However, no details of its
• phonetic characteristics are provided.
• We still ignore its duration and spectral
• characteristics in different contexts. As a
• result, the phonetic variability of this
• segment is unknown.
• This paper reports on a systematic

1. Variability in rhotic quality In intervocalic
position, r in Greek is a tap predominantly.
However, a lot of variability in production was
found (Fig 1)
Figure 1. Three repetitions of the same word, by
the same speaker show the variability in r
production (a) complete closure (b) partial
closure (c) weak, glide-like constriction
Epenthetic V 82
No epenth V 18
Table 4 Makeup of rC taps
(a) (b) (c)
The percentages of different types of rhotic
produced are shown in Table 1. Table 2 shows
further variability in the amount of constriction
for the tap.

• The quality of the vowel or consonant adjacent
  to the tap did not have any effect on its
  realization.
• No speaker effect was found

Discussion
Complete constriction 48
Incomplete constriction frication 52
tap 64
approximant 34
trill 2
Greek speakers are not aware of the presence of
the epenthetic V fragment. Why is it there? It
has been suggested in the literature that the
epenthetic V fragment serves the purpose of
making the tap perceptible when it is found next
to an obstruent. This of course is a matter for
empirical verification through a perception
experiment. However there are two considerations
which lessen the likelihood for the
perceptibility explanation on the one hand, the
existence of tokens in Greek lacking a V fragment
which are perfectly perceptible on the other
hand, the existence of languages without such
epenthetic vowels. The obvious alternative
(complementary?) explanation for the existence of
these fragments is articulatory ease and the
universal dispreference against clusters. One
experimental finding gives some weight to this
explanation Cr clusters in this experiment were
word initial (and homo-syllabic) and they always
had a V fragment, while rC clusters were word
medial (and hetero-syllabic) and approximately
20 of these tokens lacked such a fragment. The
effect of these prosodic parameters on the
realization of r in clusters is left open.

• 2. What factors influence rhotic quality?
• The gradience in tap production correlated with
  duration Short duration correlated with
  weakening in constriction.
• Stress, prosodic position and flanking vowel had
  no effects on the nature of rhotic that was
  produced.
• A speaker effect was found males showed greater
  percentages of approximant production than female
  speakers.

Aim

• Explore
• whether the Greek rhotic displays variability in
  production
• factors affecting its phonetic realization.
• The parameters examined
• the phonetic environment (V or C context)
• the prosodic environment (word initial, word
  medial, stress).
• Two experiments
• In Experiment 1, the rhotic is investigated in
  intervocalic position (VrV), to determine whether
  any other types of rhotic are attested.
• In Experiment 2, the phonetic context is Cr and
  rC clusters to determine the influence of
  phonetic context.

Results Experiment 2, Cr and rC clusters

• Rhotic quality
• In consonant clusters, just like in intervocalic
  position, Greek most frequently uses a tap.
  Approximants in C-clusters are fewer than in
  intervocalic position.
• 2. Characteristics of the tap in C-clusters
• An epenthetic, short vowel fragment was found
  between the tap and the stop, similar to that
  reported for Spanish (Navarro Tomás 1918,Quillis
  1993, Bradley and Schmeiser 2003).
• The formant structure of this vowel fragment is
  
• similar to the vowel adjacent to the tap (Fig
  2).
• Its duration is very short, on average 30 ms.

CR RC
tap 100 95
approximant 0 5
Table 3 General variability in clusters
Conclusion
In summary, the detailed phonetic examination of
the Greek rhotic sound has revealed that it is a
tap. The single contact of this tap is realized
with various degrees of constriction along a
contunuumfrom complete closure to a glide-like
approximation. The phonetic implementation
of this segment is influenced by pressures in the
domains of time, perceptibility, and
articulation. Cross-linguistically, such
processes have led to reanalysis and
lexicalization of the epenthetic
vowel. Examination of the acoustic details of
segments like the Greek tap are important for the
better understanding of processes that shape the
phonological systems of languages.
Method

• Experiment 1 Intervocalic position (VrV)
• 15 words embedded in I leksi ___ ine apli The
  word ___ is simple. 6 speakers, 6 repetitions
  each
• Varying parameters
• vowel quality a, e, i, o, u, same vowel
  flanking r
• prosodic position of r (word-initial or medial)
• stress (in stressed or unstressed syllable).
• Experiment 2 Consonant clusters Cr and rC
• 36 words embedded in O oros ___ silavizete efkola
  The term ___ is easily syllabified. 5 speakers,
  3 repetitions
• Varying parameters
• segmental makeup of C (p, t, k, b, d, g)
• quality of flanking vowels (i, a, u)
• position of C (Cr or rC)

Figure 2. These spectrograms illustrate the
V fragments (shown with an arrow) in (a) kru,
(b) kri, (c) kra
Acknowledgements
I would like to thank the University of Ioannina
first-year linguistics students of 2004-2005 for
their help.