Slovenski jezik – Slovene Linguistic Studies 6 (2007): 195−207
Peter Jurgec
Center for Advanced Study in Theoretical Linguistics, Universitetet i Tromsø
in/and
Inštitut za slovenski jezik Frana Ramovša ZRC SAZU, Ljubljana
Acoustic Analysis of Tones in Contemporary Standard
Slovene: Preliminary Findings
V članku so predstavljeni predhodni rezultati obsežnejše akustične študije o prozodiji slovenščine.
Pričujoča raziskava o akustičnih lastnostih leksikalnega tona v slovenščini obravnava trajanje samo-
glasnikov, jakost in osnovno frekvenco. Statistično značilnih razlik ni bilo med prvima, medtem
ko so bile razlike v F0 visoko statistično značilne. Rezultati se precej razlikujejo od predhodnjih.
Dvozložnice se razlikujejo v relativnem F0 posameznih samoglasnikov. V trizložnicah z naglasom
na prvem zlogu se F0 najbolj razlikuje v zadnjem zlogu.
The author presents preliminary findings of a larger acoustic study into prosodic structure of
Slovene. This investigation in acoustic properties of tones in Slovene addresses vowel duration, in-
tensity and fundamental frequency. Although no statistical differences between both lexical tones
(or pitch-accents) were found in the first and second variable, the third was found highly signifi-
cant. The results differ greatly from what was previously established. Disyllables differ in F0 of
the individual vowels. In trisyllables with antepenultimate stress, the F0 difference is the highest
in the final syllable.
1 Introduction
It is a general consensus in the field of Slovene studies that Slovene has two lexical
tones, or pitch accents, historically termed acute and circumflex.1 The tones only appear
in some (mainly central) Slovene dialects, and as an option also in the standard language.
Traditionally, these tones have been indicated by a diacritic mark on the stressed vowel
only (or a consonant when vowel is not present in the spelling), and termed tonemes.
This suggests that tone was seen as a property of the stress, further motivated by di-
achronic findings, cf. Toporišič, 2000. However, stress is an independent prosodic feature
in Slovene, realized as increased duration (Srebot Rejec, 1988), different vowel quality,
spectral tilt (Jurgec, 2005b, 2006) and other characteristics (e.g. phonation, cf. Jurgec,
2005c). Also, in the “non-tonal” dialects, stress is indicated primarily by a rise in fun-
damental frequency. As this investigation is not one into phonological properties of the
lexical tones in Slovene, the existence of the two lexical classes shall not be questioned.2
1 The author would like to thank Michael Becker, Martin Krämer, Grant Lundberg, Bruce Morén, David
Odden, Curt Rice, Dragana Šurkalović, Christian Uffmann, the anonymous reviewer, and the participants of the
38th National Convention of the American Association for the Advancement of Slavic Studies (Washington,
D.C., November 18, 2006) for useful suggestions and comments. All remaining errors, if any, are the author’s.
– Author’s address: Universitetet i Tromsø, N-9037 Tromsø, Norge.
2 For phonological interpretation of the present acoustic analysis, see Becker & Jurgec (in preparation).
196 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
Although Slovene tones have been investigated acoustically since the 1930’s, the re-
sults of the studies were inconclusive, often contradictory and generally not representa-
tive. The studies so far limited the scope of analysis to fundamental frequency alone. Du-
ration was also mentioned, although not considered systematically (Srebot Rejec, 1988).
The first acoustic description of Slovene tones on the basis of actual measurements
can be traced to Bezlaj (1939).3 It was not until another two decades, that a proper acous-
tic investigation was conducted (Vodušek, 1961). Vodušek recorded a limited number of
tokens and investigated speakers of different dialectal background, chiefly linguists. No
actual numeric data was presented, however, pitch diagrams were also published and
provide an empirical evidence of author’s claims. Vodušek found two different realiza-
tions of both tones, dependent on the background (dialect) of the informant. Feature,
common to both variants, is a general Low of the acute tone and a general High of the
circumflex tone (stressed vowels only). Author also notes that tones may be realized over
the stressed and the post-tonic syllable, and that monosyllables are distinctive in Central
Upper Carniolan, Horjul and Zilja dialects.
A later study by Toporišič (1968) offered a more in-depth view, with more tokens
analyzed and discussed in detail, although only one speaker was analyzed. In order to
capture the F0 contours, Toporišič resorted to five variables, the most important of them
is the relative height of F0 of the stressed and the post-stonic syllable(s). (However,
typically only the immediately post-tonic syllable was analyzed, and in words with final
stress at the end of the Prosodic Phrase, the difference between the first and the second
part of the syllable.) This parameter was found the most consistent, and resistant to in-
fluences of sentence intonation. This led to the phonological interpretation in the later
works by the author, that both tones differ in F0 of the stressed (tonic) and the immedi-
ately post-stressed (post-tonic) vowel only. In acutes, F0 is Low on the stressed vowel,
and High on the post-stressed; in circumlexes the opposite is true. In the final position of
the Prosodic Word, the post-tonic syllable is omitted, such that acute is realized as Low
tone, and circumflex as High (1).
(1) Slovene lexical tones after Toporišič (1968, 2000)
Stressed σ Post-stressed σ (if any)
Acute Low High
Circumflex High Low
It is also worth mentioning that Toporišič was the first to report that the distinction be-
tween the stressed syllable and the final syllable was actually the greatest in terms or
relative F0 values (Toporišič, 1968: 11). As we will see, this has unnecessarily disre-
garded as trivial.
Neweklowsky (1973) analyzed tone in Carinthian dialects. He also found F0 differ-
ences significant (as well as duration). Roughly, F0 peak was found the most indicative
of both tones, acute having it in the post-tonic and circumflex in the tonic syllable. The
most important finding of Neweklowsky was, that dialects in the same dialectal group
may differ dramatically, acoustically and phonologically.4
3 A general overview of early auditory judgments of phonetic characteristics of Slovene tones is available in
Toporišič (1967). This paper is intended to present only acoustically relevant data, such that non-experimental
data is disregarded.
4 The tones in Ovčja vas (Valbruna) were analyzed as a part of a phonetic description of the speech by
Jurgec (2005a).
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 197
Toporišič’s findings remain the most widely accepted view of Slovene tones, although
a more systematic and quantitatively better analysis was subsequently conducted by Sre-
bot Rejec (1988). She investigated both duration and F0, and compiled an extensive
corpus of sentences by three male speakers from Ljubljana. As regards the prosody, she
concludes the following: (1) quantity is no longer distinctive in Standard Slovene, (2)
tone is marginally, but still significantly distinctive, although probably not in the final
syllable Srebot Rejec (2000), (3) tones are typically realized as contours. Both tones are
ideally realized with a peak in F0 at about two thirds of the duration of the stressed vowel
(circumflex) or at the end of the post-tonic syllable (acute). The domain of tone is a foot,
consisting of two syllables or a single syllable in case of final stress.5
Srebot Rejec’s analysis is phonetically the most accurate so far. Similar to Newek-
lowsky, she took three points of measurements, in the beginning, middle and end of the
vowel.6 Coupled with the well documented loss of vowel quantity contrast (Srebot Re-
jec, 1988; Petek et al., 1996; Jurgec, 2005b), prosodic system of contemporary Slovene
that has been described by methods of acoustic phonetics, has been completely ignored
by Slovene linguists, and has not received proper description so far.
Although more recent sources indicate progressive loss of tone in Slovene dialects
(Lundberg, 2001; Vera Smole and Roberto Dapit, p.c.), it is still believed that tonal
Slovene is widely spoken. Recent perceptual study (Jurgec, in preparation) concluded,
that tones are perceived well above the chance levels by the native speakers, contrary to
what was reported by Šuštaršič & Tivadar (2005).
In this paper, we try to answer some of the outstanding issues as regards acoustic
characteristics of Slovene tones. Three acoustic parameters are investigated: fundamental
frequency, duration and intensity.7 The article is structured as follows. First, an overview
of the experiment method is described. Then the results of the experiments are presented.
The final section interprets and compars the results with previous studies and concludes
the article.
2 Experiment
In order to properly observe tonal phenomena, segmental influences, sentence intonation
and speaker variability should be minimized already when preparing the data-set. In a
controlled experiment such as this, the prepared list of words should be compiled care-
fully. It is known that non-vocalic segments influence F0, duration and intensity. For ex-
ample, while nasals influence F0 minimally, other sonorants have a greater influence, and
obstruents (particulary stops) a considerable one, that should be avoided. On the other
hand, highly sonorous segments are the most difficult to delimitate from the preceding or
the following vowel, while obstruents are not. This leads to a paradoxical situation. Al-
ternatively, minimal pairs in tone can be found, and thus segmental influence minimized.
In Slovene, fewer than 100 non-morphologically related minimal pairs in tone exist, and
the majority of them is rather peripheral. This differs from other typologically similar
languages considerably (cf. more than 500 such pairs in Swedish). For this experiment,
5 As in all the previous studies, the measurements are presented in great detail, which makes it harder to get
a proper overview, abstracted from individual utterances and their acoustic measurements.
6 Slovene tones were also investigated by Woznicki (2006). She confirmed Srebot-Rejec’s observation that
Slovene as spoken in Ljubljana is in the process of tone loss. However, she also found speakers that had
contrastive tone. Generally, her findings were inconclusive.
7 As this is a preliminary report of a larger study of Slovene prosody, only the main results are presented.
198 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
only a very limited amount of minimal pairs is required. From the dictionaries available,
16 minimal pairs of mono- to trisyllables were selected (a few of them are morphologi-
cally related). The following consonant and vowel patterns were observed: CV.CV as a
basic tamplate, CVC.CV and CV.CVC as templates with closed syllables, monosyllable
CVC and tri-syllables CV.CV.CV. The complete list of words with both tones8 is listed
in Table 1.
Form Class I Class II
"kila ‘hernia’ ‘kilogram’
"kuRa ‘chicken’ ‘medical treatment’
"Ruta ‘scarf’ ‘FEM name Ruta, Ruth’
"mula ‘mule’ ‘Muslim priest’
"lisa ‘cow’ ‘patch’
"slaVa ‘fame’ ‘FEM name Slava’
"wRana ‘FEM crow’ ‘ACC SG MASC crow’
"Zar@k ‘stale’ ‘ray’
"tuRna ‘GEN SG tower’ ‘ADJ FEM SG ‘tour’
"ROtSka ‘handle’ ‘cup handle’
"matSka ‘ACC SG MASC cat’ ‘FEM cat’
"ValtS@k ‘waltz’ ‘DEMwave’
pot ‘path’ ‘sweath’
"stikati ‘search for’ ‘talk’
"Salitsa ‘DEM joke’ ‘cup’
"Sibitsa ‘match’ ‘DEM whip’
Table 1: Tonal minimal pairs used in the experiment.
From the list one can see that all words have word-initial stress. Based on the previous
findings (see previous section), pre-stressed syllables were not analyzed.
The words were then put in frame sentences, using actual contexts of the selected
words, based on the text corpus Nova beseda (Jakopin, 2000−2006). The sentences were
randomized and put into a Power Point presentation, to be later used in the experiment
(read by the subjects).
Next, the subjects were selected. Eight speakers from central Slovenia, residents of
Ljubljana, some of them originally form different parts of Lower (2), Upper Carniola
(1) and Carinthia (1). All were educated past high school, four of them female and four
male. Median age at the time of the experiment was 38 years. However, it was later
established that minimal pairs of two male speakers do not differ significantly at all, so
they must not be speakers of the tonal variety Slovene. Both speakers were excluded
from the analysis completely, and so only the tones of six speakers (four females and two
males) were acknowledged.9
8 From here on, Class I refers to traditionally called the acute and Class II to the circumflex.
9 This causes a typical frequency bias, such that average F0 would be higher than otherwise. It is not the
absolute values that are at question here, but relative pitch contours.
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 199
The subjects were instructed first to read the sentences that appeared on the computer
screen. These were cues for contextualizing relevant words (minimal pairs in tone). After
reading the sentence, a word changed the color on the screen and then disappeared. The
speakers were instructed to read those words in a carrier sentence (2), which was only
presented to them in a spoken form by the investigator. They were instructed to repeat
each sentence twice.
(2) Carrier sentence
Reci [kila] navadno, ne posebno.
‘Say [kila] regularly, not in a special way.’
The recordings took place in several sound-proof rooms (usually radio studios, different
locations). Recording was digital, with a standard sampling frequency of 44.1 kHz, at a
16 bit rate. Recordings were stored on digital storage devices and later transferred to a
computer for acoustic analysis. For acoustic analysis Praat software program (Boersma
& Weenink, 1992−2006), was used. Relevant words were segmented manually, and
then analyzed using modified scripts. Normally the first utterance of the carrier sentence
was analyzed. Duration was measured for all stressed and post-stressed vowels, as well
as time-normalized F0 and intensity contours. For each vowel, 11 points of intensity
and F0 were measured.10 The window of each point of measurement for fundamental
frequency was determined by the duration of the vowel, such that it was 2/11 of the
total duration of the vowel. (Elsewhere Praat’s defaults were used.) In this manner, 192
words were obtained for analysis, i.e. 96 minimal pairs, or 408 vowels. However, 9
words (4.7%) were excluded from the analysis because of various reasons (e.g. incorrect
pronunciation, irregular or non-detectable F0).
3 Results
3.1 Duration
Durations of individual vowels (not whole syllables nor voiced parts thereof) in the ob-
served words were measured, ratios calculated and averaged. The results are presented
in Table 2.
The results indicate, that the stressed vowel is slightly longer in Class I words, such
that difference is independent of any segmental, (sentence) intonational and even speaker
dependent influences. Moreover, it is consistent across all word types. Also, the differ-
ence is never statistically significant (0.1 < p < 0.6). This leads to the conclusion, that
no statistically significant differences between both tonal types were detected.
3.2 Intensity
Cross-linguistically, intensity contours are known to complement tonal contours. In
Slove-ne, however, no statistically significant differences (p > 0.3) in intonation con-
tours were detected, see Figure 1.11
As intensity was not studied systematically in connection to Slovene so far at all, it is
and interesting finding, that in words with antepenultimate stress, a secondary intensity
10 Intensity and F0 themselves were not normalized, as they did not differ significantly from the absolute
values.
11 Here, only CVCV and CVCVCV words are presented, as the results for all words were similar.
200 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
Word Class I Class II Ratio
type Dur %v1 %v2 %v3 Dur %v1 %v2 %v3 II/I
cv.cv 235 51.3 48.7 220 52.9 47.1 .937
cvc.cvc 218 61.0 39.0 201 61.8 38.2 .921
cv.cv.cv 243 35.4 30.0 34.6 219 38.6 31.3 30.1 .902
cvc 122 109 .893
Table 2: Duration of vowels with respect to tone. Abbreviations: Dur – averaged sum of
individual vowel durations in a word, %v1 – percentage of the first vowel with respect
to the sum of durations of all vowels in a word, Ratio II/I – Ratio of average total vowel
durations of Class II with respect to Class I words.
peak in the final syllable/vowel was found. This may be indicative of secondary stress,
also connected to tonal prominence of final vowels (see Section 3.3). Increase in intensity
is displayed in Figure 2.
Figure 1: CVCV intensity contours (for each vowel).
3.3 Fundamental frequency
Previous studies have already established that Slovene tones differ primarily in F0. How-
ever, they disagree how this difference is realized and how it should be interpreted phono-
logically.
An average "CVCV word was realized as presented in Figure 3. As evident from
the graph, stressed syllables do not differ in F0 (at least not significantly), whereas post-
stressed syllables have higher F0 in Class I words, and both contours are falling. Differ-
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 201
Figure 2: CVCVCV intensity contours (for each vowel).
ence in the poststressed syllable is statistically highly significant (p < 0.005). However,
in the closed syllables, situation is somewhat different. Class I tone is realized with a
lower F0 than Class II. This might be to the fact, that not the whole F0 contour is visible
(or was analyzed). Yet, both tones are distinctive in this case, see Figure 4.
Figure 3: CVCV F0 contours (for each vowel).
Tones in trisyllables also differ highly significantly (p < 0.001). The distinction is
limited to the stressed (antepenultimate) and the final syllable, the former having lower
F0 in Class I tone, and the latter being considerably higher, see Frigure 5. This is a
202 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
Figure 4: CVC(C)V(C) F0 contours (for each vowel).
surprising finding, since acoustic phoneticians who studied Slovene tones so far only
considered the stressed and the immediately post-stressed syllable, but not the rest of the
Prosodic Word.12
Figure 5: CVCVCV F0 contours (for each vowel).
12 Vodušek (1961) mentions the possibility of tones being realized phonetically on all post-tonic vowels, but
does not offer a tangible analysis.
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 203
Monosyllables differ only marginally significantly (p = 0.042), although a clear ten-
dency is visible in Figure 6: Class I has a rising contour, while Class II has a falling
one. This seems to be in accordance with the findings in Vodušek (1961) and Srebot Re-
jec (1988, 2000): in the speech of Central Slovenia, tonal differences are neutralized
in monosyllables. The question, whether this is phonological proper (l.c.), effectively
neutralizing all tonal contrast in final-stressed words, positionally phonological or just
phonetic, remains beyond the scope of this article. The results should be interpreted with
caution: only six tokens were analyzed for each tone, and only one monosyllabic word.13
Figure 6: CVC F0 contours (for the vowel).
4 Discussion and conclusion
The results of the experiments are summed up in (3).
(3) The conclusions of the present acoustic analysis
a. There is no significant difference in duration between both tones
b. There is also no significant difference in intensity
c. Fundamental frequency of both lexical classes differs considerably
(i) The differences in mean F0 integrated over the entire final vowel were
greatest in trisyllables, comparing the two tone classes (> 40 Hz)
(ii) Monosyllables also differ in tone (in the range of 10 Hz)
d. Contemporary Standard Slovene as spoken by educated speakers in Ljubl-
jana is tonal
All these findings were connected to the questions addressed in the course of acoustic
investigation of Slovene tones in the 20th century. For example, vowels two or more
13 There is a small number of other monosyllables that are minimal pairs in tone (e.g. wRat ‘GEN PLdoor vs.
neck’), but all were deemed inappropriate because of segmental variance in realization or the distribution of
the tone among Slovene dialects and/or standard speech.
204 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
syllables after the stressed vowel were never properly analyzed. The present findings
suggest that it is not the post-stressed syllables but rather final syllables that are actually
phonetically (and phonologically) relevant.14 Now it is also clear, that Slovene tones do
not simply translate into High or Low tone (as suggested by Toporišič) on the stressed
vowel, and that all other tonal information is predictable, viz. phonetic. Actual realiza-
tions are much more complex (4). The acoustic data from the antepenultimates cannot be
explained by any of the theories so far. Instead, High vs. Low is distributed on the head
of the Prosodic Word (i.e. the stressed vowel) and the right edge of the Prosodic Word. I
shall constrain myself from further conclusions until a larger acoustic study is conducted
and more material is available for analysis.
(4) Prosodic notation, most consistent with the findings
Final stress Penultimate stress Antepenult stress
Class I pǒt "k̀ılâ "Sàĺıtsâ
Class II pôt "ḱılâ "Sáĺıtsà
Further investigation is needed. First, the status of pre-stressed syllables should be es-
tablished. At this point we cannot conclude that it is either High, Low or extra-tonal (the
latter is stipulated traditionally). The role of sentence intonation should also be investi-
gated. From a broader perspective, genetically related (e.g. Pletikos (2003) for Standard
Croatian), typologically similar and European tone languages in general are vastly under-
described, and much work is yet to be done (van der Hulst, 1999; Gussenhoven, 2004;
Yip, 2002). The same is true for the majority of Slovene dialects. Finally, acoustic data is
in need of a proper phonological account that would capture the facts about contemporary
Slovene as actually spoken.
References
Becker, Michael & Peter Jurgec (in preparation). Tone and Stress in Slovenian. Unpub-
lished manuscript.
Bezlaj, France (1939). Oris slovenskega knjižnega jezika. No. 17 in Razprave znanstve-
nega društva v Ljubljani, Ljubljana: Znanstveno društvo.
Boersma, Paul & David Weenink (1992−2006). Praat. http://www.praat.org.
Gussenhoven, Carlos (2004). The Phonology of Tone and Intonation. Research Surveys
in Linguistics, Cambridge: Cambridge University Press.
Jakopin, Primož et. al. (2000−2006). Nova beseda. http://bos.zrc-sazu.si/s beseda.html.
Jurgec, Peter (2005a). Fonetični opis govora Ovčje vasi / Descrizione fonetica della par-
lata di Valbruna. In Vera Smole & Nataša Komac (eds.) Ovčja vas in njena slovenska
govorica / Valbruna e la sua parlata slovena, SKS Planika, Založba ZRC, ZRC SAZU.
60–84.
Jurgec, Peter (2005b). Formant Frequencies of Standard Slovenian Vowels. Govor 22.
127–144.
Jurgec, Peter (2005c). Začetne opombe k laringalizaciji v slovenščini. Slovenski jezik /
Slovene Linguistic Studies 5. 135–168.
14 Pre-antepenultimate stress with respect to tone should be further investigated. Native Slovene roots do not
permit pre-antepentultimate stress, but it can surface in the derivation or inflection.
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 205
Jurgec, Peter (2006). Formant Frequencies of Vowels in Tonal and Non-Tonal Standard
Slovenian. Slavistična revija 54. 467–478.
Jurgec, Peter (in preparation). Perception of Tones in Slovenian. Unpublished
manuscript.
Lundberg, Grant H. (2001). Typology of Tone Loss in Haloze, Slovenia: An Acoustic
and Autosegmental Analysis. Slovenski jezik / Slovene Linguistic Studies 3. 169–189.
Neweklowsky, Gerhard (1973). Slowenische Akzentstudien: Akustische und linguistische
Untersuchungen am Material slowenischer Mundartenaus Kärnten. Wien: Verlag der
Österrechischen Akademie der Wissenschaften.
Petek, Bojan, Rastislav Šuštaršič & Smiljana Komar (1996). An Acoustic Analysis of
Contemporary Vowels of the Standard Slovenian Language. In Proceedings ICSLP
96: Fourth International Conference on Spoken Language Processing, October 3–6,
1996, Philadelphia, PA, USA. University of Delaware, 133–136.
Pletikos, Elenmari (2003). Akustički opis hrvatskih standardnih naglasaka. Govor 20.
321–345.
Srebot Rejec, Tatjana (1988). Word Accent and Vowel Duration in Standard Slovene:
An Acoustic and Linguistic Investigation. No. 226 in Slavistische Beiträge, München:
Otto Sagner.
Srebot Rejec, Tatjana (2000). Ali je današnja knjižna slovenšcina še tonematična. In
Razprave II. razreda, SAZU, vol. 17. 51–66.
Toporišič, Jože (1967). Pojmovanje tonemičnosti slovenskega jezika. Slavistična revija
15. 64–108.
Toporišič, Jože (1968). Liki slovenskih tonemov. Slavistična revija 16. 315–393.
Toporišič, Jože (2000). Slovenska slovnica. Maribor: Obzorja.
van der Hulst, Harry (ed.) (1999). Word Prosodic Systems in the Languages of Europe.
Berlin: Mouton de Gruyter.
Vodušek, Božo (1961). Grudsätzliche Betrachtungen über den melodischen Verlauf der
Wortakzente in den zentralen Slowenischen Mundarten. Linguistica 4. 20–38.
Šuštaršič, Rastislav & Hotimir Tivadar (2005). Perception of Tonemicity in Standard
Slovene. Govor 22. 23–35.
Woznicki, Katherine (2006). An Acoustic Analysis of Word Prosody in Lju-
bljana Slovene. Unpublished. https://kb.osu.edu/dspace/bitstream/1811/6628/1/KW-
BA−Thesis.pdf.
Yip, Moira (2002). Tone. Cambridge: Cambridge University Press.
Prispelo decembra 2006, sprejeto marca 2007
Received December 2006, accepted March 2007
Akustična analiza tona v sodobni standardni slovenščini:
Predhodne ugotovitve
Avtor predstavlja predhodne rezultate, ki so del obsežnejše akustične študije o prozodiji
slovenščine. Leksikalni ton v slovenščini je obravnavalo več slovenskih jezikoslovcev
prejšnjega stoletja. Bezlaj (1939) je prvi obravnaval ton akustično. Vodušek (1961)
je ugotovil, da sta za slovenski ton pomembna relativna višina osnovne frekvence in/ali
206 Slovenski jezik – Slovene Linguistic Studies 6 (2007)
njegova smer. Toporišič (1967, 1968, 2000) je prvi interpretiral ton tudi opisno fonološko
na podlagi akustični podatkov. Študija T. Srebot Rejec (1988) je prinesla natančnejše
podatke o tem, kako je ton realiziran fonetično, tj. da je odločilen položaj vrha F0, ki je
v drugi polovici naglašenega samoglasnika pri cirkumfleksu in na koncu ponaglasnega
zloga pri akutu. Namen te raziskave je pridobiti natančnejše akustične podatke z uporabo
sodobnih orodij za akustično analizo govora.
Za analizo je bilo izbranih 16 tonskih minimalnih. Besede so bile prebrane kot del
nosilnega stavka, predtem pa so govorci prebrali kontekstne stavke, iz katerih je bil raz-
viden pomen besede. Izbrane besede so bile vedno naglašene na prvem zlogu in eno-
do trizložne. Osem govorcev iz Ljubljane smo posneli digitalno v studiu. Dva govorca
sta bila kasneje izločena, tako da je bilo na koncu upoštevanih 181 besed, kakor so jih
izgovorili štiri ženske in dva moška govorca. Besede so bile analizirane v programu
Praat (Boersma & Weenink, 1992−2006). Izmerili smo trajanje samoglasnikov, jakost
in osnovno frekvenco. Pri zadnjih dveh smo izbrali 11 enakomerno porazdeljenih točk.
Rezultati so naslednji: Trajanje in jakost nista bila statistično značilna. Nasprotno pa
so bile razlike v osnovni frekvenci visoko statistično značilne. Največja razlika je bila
v zadnjem zlogu trizložnic (>40 Hz), sledil je prvi zlog. V dvozložnicah so bile razlike
statistično značilne v obeh zlogih (oz. vsaj v enem, odvisno od segmentnega konteksta).
Enozložnice se razlikujejo mejno. V celoti torej rezultati kažejo, da so vsi ponaglasni
zlogi potencialno pomembni za realizacijo tona. V nadaljevanju je treba odgovoriti
na vprašanje, kakšen je ton prednaglasnih zlogov in pa kakšne so interakcije med lek-
sikalnim tonom na eni strani ter frazno in stavčno intonacijo na drugi.
Acoustic Analysis of Tones in Contemporary Standard Slovene:
Preliminary Findings
The author presents preliminary findings of a larger acoustic study into prosodic structure
of Slovene. Lexical tones in Slovene have been investigated by several linguists in the
past century. Bezlaj (1939) was the first to measure the tones by means of acoustic pho-
netics. Vodušek (1961) established that the height and/or direction of F0 is acoustically
relevant for distinguishing the tones in Slovene. Toporišič (1967, 1968, 2000) posited
the first phonological interpretation of tones in Slovene on the basis of actual phonetic
data. Srebot Rejec (1988) provided more detailed account how vowels are encoded pho-
netically, viz. with F0 peak in the second half of the stressed vowel in circumflexes and
at the and of the post-tonic syllable in the acutes. The aim of the present analysis is to
provide more detailed acoustic description using contemporary speech analysis tools.
Speech corpus consisted of 16 minimal pairs in lexical tone. The words were read in
a frame sentence and introduced by a context sentence. These words all had initial stress,
and were mono-, di- and trisyllabic. Eight speakers, residents of Ljubljana, were recorded
digitally. Two speakers were later excluded from the analysis, and the data of the total
of 181 words, as pronounced by four female and two male speakers, were analyzed
using Praat software program (Boersma & Weenink, 1992−2006). Three variables were
measured: duration, intensity and fundamental frequency. For intensity and fundamental
frequency 11 points of measurements were made.
The results are as follows: Duration and intensity differences between both tones
were found not statistically significant. On the other hand, fundamental frequency was
found highly significant. The difference was the highest in trisyllables, in the final vowel
(> 40 Hz), followed by the initial vowel. In disyllables, differences in both syllables
were found significant (depending on the segmental structure). Monosyllables differ
P. Jurgec, Acoustic Analysis of Tones in Contemporary Standard Slovene ... 207
marginally. All in all, results indicate that all post-tonic syllables are potentially the
domain of the tone. Further investigation should address the pre-tonic syllables as well
as interaction with phrase and sentence intonation.