05.08.2013 Views

visas tekstas - VPU biblioteka - Vilniaus pedagoginis universitetas

visas tekstas - VPU biblioteka - Vilniaus pedagoginis universitetas

visas tekstas - VPU biblioteka - Vilniaus pedagoginis universitetas

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

8<br />

Differential Acoustical Cues for Palatalized vs<br />

Nonpalatalized Prevocalic Sonants in Lithuanian<br />

The lowest values for the F2 locus in nonpalatalized<br />

cases were found for [m], followed by [l], [n],<br />

and [r]. This is not surprising since this succession<br />

corresponds to the step-by-step posterior movement<br />

of the constriction; from labials to alveolars or postalveolars,<br />

i.e., in a sense, a plain [r] is more ‘palatalized’<br />

than a plain [m]. What is more interesting is that<br />

dental [l] shows a statistically negligible difference<br />

from labial [m]. A possible explanation is that, at least<br />

for the speaker examined, the corresponding places<br />

of constriction do not differ much, and, in the case<br />

of [l], the tongue makes a significant excursion to<br />

the uvula or lower, thus lowering F2 (refer to Figure<br />

1). This articulatory difference is also reflected in<br />

the significant rise of F3 for [l] for the same reasons<br />

(see the rise of F3 when the (secondary) constriction<br />

moves to the uvula or lower, Figure 1).<br />

In addition, plain (nonpalatalized) [n] seems more<br />

retracted, i.e., moved into the direction of apicoalveolar<br />

position, compared to [l].<br />

For palatalized consonants, F2 loci show the<br />

highest values for [n’], followed by [r’], [l’], and<br />

[m’]. Thus [n’] is the most ‘soft’, and this can be<br />

ascertained simply audially. It results from the natural<br />

‘comfortable’ configurations of the tongue, which<br />

differ according to the different consonants (here we<br />

will not discuss this issue in detail).<br />

More information could be extracted from Figure<br />

2. Here we will limit ourselves to two additional observations.<br />

First, for [m’], a certain separate subclass<br />

characteristic of high F2s and F3s is formed. It is<br />

mostly associated with a distinct change of articulation<br />

in [m’ė] or [m’i]-type syllables ([m’] becomes<br />

extremely palatalized – at least, in the case of the<br />

speaker examined). Second, F2s for [r] and [r’] are<br />

not separated as much as for other consonants; even<br />

a slight overlap is registered. Inter alia, it seems that<br />

sometimes the speaker tends to ‘strengthen’ [r’]s,<br />

making them harder.<br />

One can expect that the impact of the place of<br />

articulation onto loci should be similar for different<br />

homorganic sounds. For instance, most probably,<br />

F2 loci for homorganic sonants and stops should<br />

coincide. Loci for stops have been examined in<br />

numerous studies; Figure 3 presents a comparison<br />

with the evaluations for stops made by Kewley-Port<br />

(1983).<br />

At first glance, it might seem that the F2 values<br />

for nonpalatalized sonants in the present study are<br />

unnaturally low. However, we should remember that<br />

Kewley-Port examined English consonants. In the<br />

Lithuanian case, the distinction between palatalized<br />

and nonpalatalized consonants is made stronger, thus<br />

Diferenciniai akustiniai palatalizuotų ir<br />

nepalatalizuotų prevokalinių lietuvių kalbos sonantų<br />

požymiai<br />

the F2 loci for plain consonants are moved to lower<br />

values, in a sense, to keep a distance from their palatalized<br />

counterparts. In English, palatalization does<br />

not form opposite consonant pairs. For the Lithuanian<br />

ear, English alveolar stops sound semipalatalized,<br />

and Figure 3 visualizes this phenomenon.<br />

Figure 3. Left: F2 loci for prevocalic sonants (higher:<br />

palatalized; lower: nonpalatalized). Medians (diamonds)<br />

Figure 3. Left: F2 loci for prevocalic sonants (higher: palata<br />

and interquartiles (vertical lines) are depicted. Right:<br />

nonpalatalized). rough evaluations Medians for stops; (diamonds) after Kewley-Port, and interquartiles 1983. (vertical lines)<br />

Right: rough evaluations for stops; after Kewley-Port, 1983.<br />

Effect of vowels (regressive<br />

coarticulation)<br />

At first glance, it might seem that the F2 values for nonpalatalized<br />

present study are unnaturally low. However, we should remember tha<br />

examined It is known English that loci consonants. are far from In the invariant; Lithuanian they case, the distinc<br />

depend palatalized on and the nonpalatalized phonetic context consonants (for instance, is made stronger, see thus the F2<br />

the typical work by Öhman (1966)). One can derive<br />

consonants are moved to lower values, in a sense, to keep a distan<br />

the loci frequencies as the functions ‘of the vowel’s<br />

own<br />

palatalized<br />

intrinsic,<br />

counterparts.<br />

target formant<br />

In English,<br />

frequency’<br />

palatalization<br />

(‘locus<br />

does not form oppo<br />

equations’; pairs. For the Howie, Lithuanian 2001, ear, 2). English Figure alveolar 4, in the stops manner sound semipalatalized<br />

similar visualizes to the this one phenomenon. presented by Öhman, demonstrates<br />

this dependence (on the succeeding vowels) extracted<br />

from the data examined in the present study. The<br />

tense (‘long’) Effect and of vowels lax (‘short’) (regressive vowels coarticulation) are combined<br />

into joint classes. It should be noted that, generally,<br />

back vowels It is known result that in loci lower are far F2 from loci invariant; compared they to depend on the ph<br />

front<br />

(for<br />

vowels.<br />

instance,<br />

However,<br />

see the typical<br />

some<br />

work<br />

permutations<br />

by Öhman<br />

among<br />

(1966)). One can d<br />

vowels are also noticeable. This has several explana-<br />

frequencies as the functions ‘of the vowel’s own intrinsic, target form<br />

tions; here we will mention only two of them. First,<br />

certain (‘locus Lithuanian equations’; Howie, vowels 2001, have 2). only Figure tense 4, in versions. the manner similar to the<br />

This explains why [ė] (tense) frequently results in<br />

higher F2s compared to [i] (averaged tense and lax<br />

versions), and [o] (tense) frequently results in lower<br />

F2s compared to [u] (averaged tense and lax versions).<br />

Second, the influence of the specific speaker<br />

and the limited statistical sample could be at work.<br />

Rytis AMBRAZEVIČIUS

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!