123
Damjan Huber*   UDK 811.163.6'342:316.774
University of Ljubljana     DOI: 10.4312/linguistica.57.1.123-136
SENTENCE STRESS IN SLOVENE MEDIA SPEECH
1 INTRODUCTION
Prosody or sentence phonetics, which studies sentence stress, the partition of texts with 
pauses, intonation, register, speech tempo and tone, has been relatively little researched 
within Slovene linguistics. This is particularly true of spontaneous speech production, 
as the rare studies of Slovene speech are mainly based on researchersʼ subjective evalu-
ation and an extremely limited corpus of non-spontaneous text formation, such as (lit-
erary) texts being read out (e.g. Toporišič 1972; Podbevšek 2006). 
With the development of new computer tools for the transcription and analysis of 
speech production (e.g. Praat; Boersma/Weenink 2013), speech studies can also be 
based on real, spontaneous media speech that is classified as one of the bearers of for-
mal speech in Slovenia. Programme presenters and their guests produce an ideal media 
speech norm, which has a relatively large influence on listeners and viewers. Since 
national radio and television stations also play the role of connecting people nationally, 
the presenters on RTV Slovenia are now trained and taught the correct speech, which 
means that “media speech is in principle the ‘ideal speech realisation’, the speech reali-
sation most harmonised with the norm” (Vitez 1999: 42). Media speech is thus an ideal 
opportunity for observing how the speech of (trained) speakers deviates from an ideal 
or the linguistic norm presented in reference books.
2 RESEARCH PREMISES 
In this paper, sentence stress is understood as emphasis within the flow of speech, or as 
a prosodically realised emphasis on the surface of the text, which is a reflection of the 
information structure of the text at the deep level (Duběda 2005: 183) and is phoneti-
cally realised through the combination of a number of acoustic factors: a change in the 
length of individual sounds, the tonal level or frequency of the fundamental frequency 
(F0), the intensity and speed of speech, and the appearance of pauses before and/or after 
stresses, etc. Researchers do not agree which of these signals is the most important for 
the perception of sentence stress, while stress lies on the syllables of the words within 
certain segments, which interlocutors or listeners perceive as marked.1 
* damjan.huber@ff.uni-lj.si
1 The decision which word will carry sentence stress depends mainly on the speaker’s choice. As 
stress expresses the speaker’s judgment about whether, within a certain moment and context, 
a particular word is most important to him with regard to the information he/she is conveying, 
those words are stressed in speech that represent the knowledge shared by the speaker and the 
listener (Komar 1996: 39–41).
Linguistica_2017_FINAL.indd   123 12.3.2018   13:08:24
124
The research into sentence stress presented in this paper is part of a wider study where 
sentence stress in speech-trained people (presenters of television programmes) and (prob-
ably) somewhat less trained speakers (politicians) in chosen TV Slovenia programmes 
was analysed with regard to the frequency of appearance, intensity, F0, sound lengthen-
ing, predictability of word types, frequency of appearance in the selected language units 
(interrogatives, negatory elements, affirmatives, indicative pronouns, comparatives and 
superlatives)2 and connections with pauses (Huber 2013). In addition, stresses can be 
researched with regard to the roles they play in speech, intensification and repetition of 
what has been said, non-verbal means, with regard to the connections with other prosodic 
characteristics, such as intonation, speed of speech and so on. During the study, the com-
bined auditory-experimental method of analysing sentence stress was used.
2.1 Hypotheses
On the basis of three methodological approaches, research hypotheses based on the Slo-
vene linguistic reality and observation of media (political) speech were examined. The 
first hypothesis, that the speech of politicians contains more examples of sentence stress 
than that of television presenters, and the second hypothesis, that sentence stress by poli-
ticians is on average more intense (dB) and has a higher F0 (Hz) than that of present-
ers, were formed on the basis of the assumption that in general politicians are less well 
trained than presenters on national television, which has for decades been giving speech 
training to its journalists. Moreover, we assumed that presenters are better prepared for 
the discussion in a programme (spoken interventions are pre-written). Due to these two 
assumptions we can conclude that the speech of politicians is more spontaneous than that 
of presenters and thus also more subject to unexpected and uncontrolled sentence stress.
The third hypothesis assumes that with regard to the auditory perception of sentence 
stress the most important role is played by a change/increase in the intensity of speech, 
while raising F0 or lengthening sounds is slightly less important, and is based on the as-
sumption that in Slovene speech a greater role is played by the dynamic accent and sen-
tence stress, i.e. the changing of speech intensity, while tonemic stress, i.e. the changing 
of F0, appears more rarely in media speech, and generally not at all in trained speakers.
2.2 Material and Speakers
Sentence stress was analysed within a corpus of authentic and more or less spontaneous 
speech of chosen politicians and presenters in TV Slovenia talk shows with political 
content (Intervju, Pogovor s predsednikom vlade, Vroči stol, Pod žarometom, Omizje, 
Studio City). In the programmes shown between 2006 and 2007 during evening slots 
between 20.00 and 00.10, the guests (politicians) speak mostly spontaneously, while 
2 In comparison with English, e.g. Šuštaršič (1995: 164–178) states that in Slovene negative 
phrases, the stress usually lies on the negation, while in English it follows the negation. In open 
interrogatives, in English the unmarked stress position is on the last lexical unit, and in Slovene 
on the interrogative word. Within an analysis of comparatives and superlatives, Šuštaršič found 
that in English stress lies after the comparatives more and most, and in Slovene on bolj (more), 
najbolj (most) and nadvse (exceedingly).
Linguistica_2017_FINAL.indd   124 12.3.2018   13:08:24
125
the presenters speak both spontaneously and non-spontaneously – the latter, for in-
stance, when presenting the guests and the programme theme. The programmes include 
dialogues, multilogues and monologues that turn into dialogues, as well as formal and 
public speech units, recorded in the Ljubljana studios of TV Slovenia without an audi-
ence. The specialised speech corpus Korp-2MeGo of a total duration of over two hours 
(involving 3 male and 3 female politicians and 5 male and 4 female TV presenters) 
was created specifically for the purpose of this study, with emphasis on the analysis of 
spontaneous speech, that is particularly the speech of politicians (see Table 1). All the 
chosen speakers are well known media or political personalities.
Table 1: Speech duration with regard to the sex of the speakers
Speakers Male Female Male + female
Politicians 44 min and 51 s 37 min and 51 s 82 min and 42 s
Presenters 20 min and 35 s 15 min and 23 s 35 min and 58 s
Simultaneous speech / / 3 min and 55 s
Total: 122 min and 35 s
All the speakers, whose first language is Slovene, have a university degree in social 
sciences; they were born between 1941 and 1970, live and work mostly in Ljubljana, 
and spent their childhood and teenage years in various places around Slovenia. The 
speakers have no speech defects and in a neutral, non-emotional speaking situation use 
unmarked sentence stress.
3 RESULTS AND DISCUSSION
This section provides detailed results of the study of sentence stress with regard to the 
frequency of appearance in speech, intensity, F0 and the lengthening of sounds.
3.1 Frequency of Sentence Stress in Speech
The first part of the study is aimed at an analysis of the frequency of sentence stress in 
speech. A similar study was carried out by Fox (1989), when he analysed the speech 
of the presenters and reporters in three national TV news programmes: the speakers at 
the BBC on average use sentence stress 39.16 times per hundred spoken words, at the 
German ARD 22.31 times and at the Croatian TV Zagreb, 18.68 times.
In determining the number of instances of sentence stress, i.e. establishing the aver-
age number of words connected by one example of sentence stress within one or more 
segments, all the words that were perceived as stressed through auditory perception were 
taken into account. At the same time, these words can also be the bearers of the syllable 
with the highest measured intensity and/or the highest measured F0 within the segment 
– the starting point is thus the auditorily perceived examples of sentence stress. All the 
words spoken during simultaneous speech and all the forms of filled pauses (lengthen-
ing, repetition, false starts, etc.) were excluded from the study.3 The main interest in this 
3 On filled pauses, see Huber 2013: 375–382.
Linguistica_2017_FINAL.indd   125 12.3.2018   13:08:24
126
part of the study was the differences in the number of spoken words between two sen-
tence stresses that appear in the speech of politicians and TV presenters. The hypothesis 
was put forward that in the politicians’ speech there are more instances of sentence stress 
than in the speech of TV presenters – the latter were assumed to pronounce on average 
more words between two examples of sentence stress than the former.
Table 2 shows the number of spoken words in the corpus (16,629), the number 
of auditorily perceived examples of sentence stress (4,401) and the average number 
of words connected by one example of sentence stress within one or more segments 
(3.78). The results are shown by individual categories (politicians and presenters) and 
the speakers’ gender.
The results show that in the speech of presenters, a slightly higher number of sen-
tence stresses appear (every 3.71 words, i.e. sentence stress appears in 26.98% of all the 
spoken words) than in the speech of politicians (every 3.81 words, 26.23%). The study 
confirmed the first hypothesis about a slightly higher frequency of sentence stresses in 
politicians, i.e. speakers who are presumably less well trained in speaking and less well 
prepared for the discussion in the programme. In the case of men, it was shown that 
politicians used sentence stress more often (every 3.84 words, 26.07%) than presenters 
(every 4.04, 24.74%), while with female speakers the situation was completely differ-
ent: female presenters use sentence stress more frequently (every 3.33 words, 29.99%) 
than female politicians (every 3.79 words, 26.40%). In general, male speakers use sen-
tence stress less (every 3.90 words, 25.64%) than female speakers (every 3.64 words, 
27.44%).
Table 2: Frequency of appearances of sentence stress in speech4
Speaker4 Number of 
words spoken
Number of auditorily per-
ceived sentence stresses 
% stressed 
words 
Number of words/
sentence stress
Male and female politicians
Pm 5,945 1,550 26.07 3.84
Pf 5,466 1,443 26.40 3.79
Pm+Pf 11,411 2,993 26.23 3.81
Male and female presenters
Jm 2,987 739 24.74 4.04
Jf 2,231 669 29.99 3.33
Jm+Jf 5,218 1,408 26.98 3.71
Male and female politicians and presenters
Pm+Jm 8,932 2,289 25.63 3.90
Pf+Jf 7,697 2,112 27.44 3.64
P+J 16,629 4,401 26.47 3.78
4 Abbreviations: Spkr – male/female speaker. P – male/female politician. J – male/female journal-
ist/presenter. m – male, f – female, individual numbers 1 – 22 are used for individual speakers.
Linguistica_2017_FINAL.indd   126 12.3.2018   13:08:24
127
Among the politicians, in one of the two analysed programmes where he features, 
speaker SpkrPm5+16 used sentence stress the most frequently (every 3.61 words) 
of all the politicians: SpkrPm16 (every 3.25 words, 30.75%), but in the other pro-
gramme, as speaker SpkrPm5, he used it considerably less frequently (every 4.09 
words, 24.42%). The example of the speaker SpkrPm5+16 clearly shows a connection 
between (un)emotional speech and sentence stress, since in one of the programmes the 
speaker replied calmly to questions and thus used sentence stress much less frequently 
(SpkrPm5) than in the other programme, when he responded to questions more emo-
tionally and hence used stress very frequently (SpkrPm16). In the programme where 
he used sentence stress more, he was emotionally responding to current events on 
the border with Croatia and criticising the work of the government’s political op-
tion; emotional speech, by definition, is presumed to contain a more frequent use of 
sentence stress, probably including higher utterance intensity and F0 (which will be 
examined in section 3.2). All the other politicians used sentence stress very constantly 
and comparably, namely, on 22.1 to 27.08% of the spoken words.
Among the presenters with fewest examples of sentence stress, the most notice-
able is journalist SpkrJm2, who uses sentence stress on average every 5.19 words 
(19.28%), followed by the presenter SpkrJm6+13, who in one of the analysed pro-
grammes, as presenter SpkrJm6, used sentence stress every 4.92 words (20.33%). 
In the other analysed programme (SpkrJm13) he used sentence stress much more 
frequently, every 3.03 words (33.06%), which is the second highest percentage in 
the corpus. The most frequent use of sentence stress appeared in the female presenter 
SpkrJf15 (33.33%).
Within one hundred spoken words, the average use of sentence stress was 26.47% 
– among politicians 26.23% and among presenters 26.98%. A comparison of these re-
sults with the study carried out by Fox (1989: 82) shows that the analysed presenters use 
sentence stress much less than the BBC speakers (26.98 per one hundred spoken words 
versus 39.16), slightly more than the ARD speakers (22.31) and considerably more 
than the speakers in the TV Zagreb news programmes (18.68). According to Fox’s rule, 
which says that the span of the auditory memory is up to 15 words maximum and that 
for successful communication sentence stress must be used at least seven times per one 
hundred spoken words, the speakers whose speech was analysed in this study made 
good use of this prosodic characteristic with the aim of successful communication. 
3.2 Intensity and F0 in Stressed Words
In Slovene linguistics, both accent and sentence stress are most often defined primar-
ily in terms of dynamic accent or intensification (pronunciation with a greater quantity 
of exhaled air) of a certain syllable or a longer speech unit, but instrumental analyses 
(particularly for other languages) have shown that expressiveness (accent or stress) 
depends on the combined influence of a number of characteristics, e.g. F0, intensity, 
duration and quality of vowels (Palková 1997: 156–157). Sentence stress is thus pho-
netically realised through a combination of a number of acoustic factors, among which 
the literature most often mentions a change in the duration of individual sounds, F0 and 
Linguistica_2017_FINAL.indd   127 12.3.2018   13:08:24
128
intensity (more details in section 3.3), while with regard to the perception of sentence 
stress, various combinations of the above (acoustic) parameters are decisive.5
Although different researchers agree with the selection of the acoustic signals of 
emphasis, they do not fully agree which of the above signals is the most important for 
the perception of sentence stress. Some, such as Škarić (1991) believe that it is the 
change of intensity that plays the biggest role. Ladefoged (1975) claims that the length 
of the sound is most important, followed by intensity and, lastly, F0. But Bolinger 
(1972) gives F0 as the most important acoustic signal of sentence stress, since he be-
lieves that F0 must almost definitely change if we wish to emphasise a word. He main-
tains that sentence stress is perceptually strongest in the cases when the change in F0 
within a sentence is very large or varies strongly. Information about sentence stress is 
thus encoded in the acoustic signal as the relative intensity of the speech segment, as 
duration, F0 and the precision of articulation. The relative contribution of each of these 
characteristics in accent and stress, according to the data of various researchers, varies 
between zero and one hundred percent, so that the feeling of accentuation or stress of 
a segment can be achieved even with the change of just one of the above dimensions 
of sound, but most often there is a small change in each of the characteristics (Bakran 
1996: 245–248).
Because F0 and speech intensity rank among the most important signals of the per-
ception of sentence stress, they were measured in this part of the study. In the calcula-
tion of the average values of F0 and speech intensity, the highest values of the empha-
sised word (words) within a segment were taken into account,6 whereby words can be 
emphasised auditorily (perceptively), in terms of intensity, and/or pitch. Each segment 
may contain one or more auditorily perceived words and only one peak in terms of in-
tensity and one in terms of frequency, where the two peaks can lie on a syllable of one 
of the auditorily stressed words or on one of the auditorily unstressed words. Below is 
a test of the hypothesis that the sentence stress of politicians is on average stronger in 
intensity and that politicians have a higher F0 than the sentence stress of TV presenters.
Table 3 shows the number of words that constitute peaks in intensity and/or pitch 
within individual segments – words that were perceived as emphasised only on the 
basis of auditory analysis (usually in cases when within one segment more emphasised 
words were auditorily perceived) were not included in this – average intensities and 
average values of F0 in emphasised words in terms of individual category (politicians 
or presenters) and the speaker’s gender.
The average intensity value of all 3,373 highlighted words in the corpus is 81.46 dB, 
while the average F0 value is 205.04 Hz. Generally, emphasised (stressed) words in male 
5 I am aware of the key role of sentence stress within sentence intonation and the inseparability 
of these two prosodic characteristics. In spite of this, I have tried to examine the characteristics 
of sentence stress independently of intonation patterns, i.e. to use intonation peaks in the 
interpretation of the results. The concrete connections between sentence stress and intonation 
patterns will need to be further researched in future (see also Šuštaršič 1995).
6 A segment is understood as part of the speech flow, i.e. a language string, delimited on both sides 
by a pause, spoken by the same speaker, irrespective of its length.
Linguistica_2017_FINAL.indd   128 12.3.2018   13:08:24
129
speakers measured lower values (80.91 dB, 149.33 Hz) than in female speakers (82.06 dB, 
265.98 Hz). The difference in average intensity values of emphasised words is rela-
tively small (1.15 dB), but it is more pronounced in the pitch of the F0 (116.65 Hz). 
In a comparison of politicians and presenters, Table 3 shows that politicians empha-
sised words slightly more strongly, i.e. with greater intensity and higher F0 (81.55 dB, 
207.25 Hz) than presenters (81.27 dB, 200.25 Hz), which only partly confirms the 
second hypothesis as this statement applies only to the female speakers (female politi-
cians 82.33 dB and 271.17 Hz, female presenters 81.46 dB and 254.27 Hz), while the 
speech of male presenters is stronger in intensity (81.12 dB) and higher in terms of tone 
(153.26 Hz) than the speech of male politicians (80.81 dB, 147.46 Hz). Although the 
two values were not measured in words that were not emphasised or stressed, on the 
basis of subjective auditory analysis it can, as expected, be concluded that the speech of 
female politicians and presenters is characterised by higher F0 and higher intensity than 
that of male politicians and presenters. In the future it would be good to build upon the 
research and to measure the highest value of F0 and the intensity of all the unstressed 
words; in this way we could ascertain what (if any) differences there are between the 
stressed and unstressed words in the speech of politicians and presenters. 
Table 3: Average intensity (dB) and average F0 (Hz) in emphasised words
Speaker Number of  
emphasised words
Average intensity  
(in dB)
Average F0  
(in Hz)
Male and female politicians
Pm 1,193 80.81 147.46
Pf 1,116 82.33 271.17
Pm+Pf 2,309 81.55 207.25
Male and female presenters
Jm 569 81.12 153.26
Jf 495 81.46 254.27
Jm+Jf 1,064 81.27 200.25
Male and female politicians and presenters
Pm+Jm 1,762 80.91 149.33
Pf+Jf 1,611 82.06 265.98
P+J 3,373 81.46 205.04
An examination of the intensity and F0 in individual speakers led to the conclusion that 
the two values are mutually connected since when one increases, the other also increases. 
If in one speaker a relatively high average intensity of emphasised words was measured, 
that speaker’s F0 pitch, in comparison with the speech of other analysed speakers in the 
corpus, was relatively high, and if the average intensity was low, the average F0 was usu-
ally also low. For example, in the case of politicians, the greatest intensity was measured 
in the female speaker SpkrPf7 (85.54 dB), while the same speaker’s average F0 value was 
Linguistica_2017_FINAL.indd   129 12.3.2018   13:08:24
130
also measured as the highest (299.35 Hz). In the case of presenters, the lowest average in-
tensity of emphasised words was measured in the speaker SpkrJm13 (79.66 dB), in whom 
the lowest average F0 was also recorded (126.93 Hz).
The measured intensities and F0 pitches are to a certain extent undoubtedly a reflec-
tion of the situation in which the speakers find themselves (whether they are appearing 
in a programme where they answer questions without any special surprises or whether 
they are responding emotionally to statements by the other guests) and, above all, on 
the personal characteristics of individual speakers (the characteristics of their voice and 
manner of speaking), which is particularly apparent in the politician SpkrPm5+16, in 
whom higher values were measured (82.49 dB, 171.74 Hz) than the average values of all 
the politicians (80.81 dB, 147.46 Hz) and presenters (81.12 dB, 153.26 Hz). The meas-
ured intensity is even higher than in the female politicians (82.33 dB) and the female 
presenters (81.46 dB), or in comparison to all the speakers in the corpus (81.46 dB). 
This applies to the values measured in the politician SpkrPm3+12 (79.88 dB, 119.87 Hz), 
which are among the lowest in the whole corpus. The characteristic of the speaker 
SpkrPm5+16 is a higher speech register and a lively manner of speaking, and the charac-
teristic of the speaker SpkrPm3+12 is just the opposite: low register and a monotonous 
manner of speaking, without any great variations in intensity and F0 pitch.
The subjective auditory impression that the sentence stress of TV presenters is, on 
average, less pronounced than that of politicians was also confirmed in the measure-
ments of intensity and F0 pitch, but the differences were relatively small (81.55 dB 
and 207.25 Hz versus 81.27 dB and 200.25 Hz). There are greater differences between 
female and male speakers with regard to the measured values (82.06 dB and 265.98 Hz 
versus 80.91 dB and 149.33 Hz), which is not unexpected since among female speakers 
greater intensity is accompanied by an inherently higher F0.
3.3 Sentence Stress – Higher F0, Greater Intensity or a Lengthened Sound? 
The data on F0 pitch and speech intensity presented in section 3.2 gain meaning only in 
a specific context. This section is thus a continuation of the analysis from the previous 
one, ascertaining which of the mentioned phonetic characteristics has greater influence 
on the auditory perception of sentence stress: the raising of the F0, increasing speech 
intensity or the lengthening of sounds. In a comparable study of Swedish speech it was 
shown that stronger sentence stress also means a higher F0 (Strangert 2003: 613–615), 
which was also confirmed in section 3.2 of this study. The absolute values are less in-
formative than the variations of individual values within the whole context. A greater 
influence of F0 on sentence stress in comparison with the duration or intensity has also 
been reported by Volk (2011). 
This part of the study takes into account only the sentence stress that was perceived 
by the auditory analysis of stresses (Stress-AP), ignoring the words within a specific seg-
ment which had the highest measured F0 and/or the highest intensity, but which were not 
auditorily perceived as stressed. Thus the research parameters included intensity, F0 and 
lengthened sound, as well as the various combinations of these three basic elements (Ta-
ble 4). The analysis tests the hypothesis that in auditory perception of sentence stress the 
Linguistica_2017_FINAL.indd   130 12.3.2018   13:08:24
131
most important role in the analysed material is played by the increase in speech intensity, 
while higher F0 or the lengthening of sounds plays a slightly smaller role.
Table 4: Stress sentence in terms of a chosen acoustic characteristic – symbols and descriptions
Symbol Explanation of the acoustic characteristics of sentence 
stresses
Stress-AP+IHP (1) auditory stress resulting from increased intensity
Stress-AP+BHP (2) auditory stress resulting from higher F0
Stress-AP+BHP+SL (3) auditory stress resulting from the co-influence of higher 
F0 and sound lengthening
Stress-AP+IHP+BHP (4) auditory stress resulting from the co-influence of an 
increase in intensity and higher F0
Stress-AP+SL (5) auditory stress resulting from sound lengthening in the 
AP position and simultaneously possibly also increased 
intensity and/or higher F0
Stress-AP+IHP+BHP+SL (6) stress emphasis resulting from the co-influence of an 
increase in intensity, higher F0 and sound lengthening
Stress-AP1 (7.1) intensity and/or F0 are higher/sound lengthened on 
another auditorily perceived stressed word within the 
segment (phrase, sentence, etc.)
Stress-AP2 (7.2) auditory stress within a segment, while the highest 
intensity and/or highest F0 are measured (or perceived 
lengthened sound) on an auditorily unemphasised word 
or words within the segment; it could also be that no part 
of the segment deviates (considerably) from the average
Stress-AP+IHP+SL (8) auditory stress resulting from the co-influence of inten-
sity and sound lengthening
The results in Table 5 are shown separately by individual types of stress, by individ-
ual speaker category (politicians/presenters) and by the speakers’ gender. In this part 
of the study, 4,401 emphasised words were studied (2,112 uttered by female speakers 
and 2,289 by male speakers, 2,993 by male/female politicians and 1,408 by male/fe-
male presenters). The study confirmed the assumption from section 3.2 that in Slovene 
the most important role in the perception of stress is played by a change in F0 pitch, as 
in 830 auditorily perceived words (18.9% of all the analysed emphasised words) the 
highest F0 value was also measured in an analysed segment (AP+BHP), and in addition 
as many as 1,716 auditorily perceived stresses (39.0%), which had at the same time 
the highest F0 and the highest measured intensity (AP+IHP+BHP). The highest inten-
sity appears as an independent category in 726 auditorily perceived emphasised words 
(AP+IHP; 16.5%), while a considerable number of auditorily perceived emphases were 
observed, which could not be supported either by the highest intensity or highest F0 
within the analysed segment; these are categories AP1 (16.4% of all the auditorily per-
ceived emphasised words in the corpus), AP2 (7.1%) and AP+SL (0.7%).
Linguistica_2017_FINAL.indd   131 12.3.2018   13:08:24
132
Ta
bl
e 
5:
 P
ro
so
di
c/
ac
ou
st
ic
 c
ha
ra
ct
er
is
tic
s o
f s
en
te
nc
e 
st
re
ss
es
Sp
ea
ke
r
A
P+
 
IH
P 
(1
)
A
P+
BH
P 
(2
)
A
P+
BH
P+
 
SL
 (3
)
A
P+
IH
P+
 
BH
P 
(4
)
A
P+
SL
 
(5
)
A
P+
IH
P+
 
BH
P+
SL
 (6
)
A
P1
 
(7
.1
)
A
P2
 
(7
.2
)
A
P+
IH
P+
 
SL
 (8
)
To
ta
l
no
./%
no
./%
no
./%
no
./%
no
./%
no
./%
no
./%
no
./%
no
./%
no
.
M
al
e 
an
d 
fe
m
al
e 
po
lit
ic
ia
ns
Pm
28
9/
18
.6
27
2/
17
.5
4/
0.
3
56
9/
36
.7
10
/0
.6
7/
0.
5
27
3/
17
.6
11
7/
7.
5
9/
0.
6
1,
55
0
Pf
20
0/
13
.9
26
9/
18
.6
6/
0.
4
64
5/
44
.7
6/
0.
4
6/
0.
4
22
1/
15
.3
86
/6
.0
4/
0.
3
1,
44
3
Pm
+P
f
48
9/
16
.3
54
1/
18
.1
10
/0
.3
1,
21
4/
40
.6
16
/0
.5
13
/0
.4
49
4/
16
.5
20
3/
6.
8
13
/0
.4
2,
99
3
M
al
e 
an
d 
fe
m
al
e 
pr
es
en
te
rs
Jm
12
3/
16
.6
14
6/
19
.8
4/
0.
5
26
5/
35
.9
7/
0.
9
8/
1.
1
13
3/
18
.0
50
/6
.8
3/
0.
4
73
9
Jf
11
4/
17
.0
14
3/
21
.4
4/
0.
6
23
7/
35
.4
6/
0.
9
5/
0.
7
93
/1
3.
9
59
/8
.8
8/
1.
2
66
9
Jm
+J
f
23
7/
16
.8
28
9/
20
.5
8/
0.
6
50
2/
35
.7
13
/0
.9
13
/0
.9
22
6/
16
.1
10
9/
7.
7
11
/0
.8
1,
40
8
M
al
e 
an
d 
fe
m
al
e 
po
lit
ic
ia
ns
 a
nd
 p
re
se
nt
er
s
Pm
+J
m
41
2/
18
.0
41
8/
18
.3
8/
0.
3
83
4/
36
.4
17
/0
.7
15
/0
.7
40
6/
17
.7
16
7/
7.
3
12
/0
.5
2,
28
9
Pf
+J
f
31
4/
14
.9
41
2/
19
.5
10
/0
.5
88
2/
41
.8
12
/0
.6
11
/0
.5
31
4/
14
.9
14
5/
6.
9
12
/0
.6
2,
11
2
P+
J
72
6/
16
.5
83
0/
18
.9
18
/0
.4
1,
71
6/
39
.0
29
/0
.7
26
/0
.6
72
0/
16
.4
31
2/
7.
1
24
/0
.5
4,
40
1
Linguistica_2017_FINAL.indd   132 12.3.2018   13:08:24
133
The sequence AP+IHP+BHP – AP+BHP – AP+IHP – AP1 – AP2, in terms of the 
frequency of appearance of different types of emphasis in the corpus, can be confirmed in 
both male and female speakers; smaller differences appear only in the percentages with-
in individual categories, e.g. the category AP+IHP+BHP appears in female speakers in 
41.8% and in male speakers in 36.4%. The category AP+IHP appears in female speakers 
in 14.9%, and in male speakers in 18.0%; the third difference greater than one and a half 
percent appears in the category AP1 (female speakers 14.9% and male speakers 17.7%). 
The sentence stress of male speakers is divided more among the categories AP+IHP and 
AP1 than the sentence stress of female speakers in the category AP+IHP+BHP. The 
highest percentage of sentence stresses in the category AP+IHP+BHP appears in female 
politicians (44.7%), and the lowest percentage in the female presenters (35.4%). It is in 
the female presenters that the highest percentage of sentence stress appears in the cate-
gory AP+BHP (21.4%), while the fewest sentence stresses appear in the male politicians 
(17.5%). Sentence stress of the type AP+IHP appears most often in female presenters 
(17.0%) and least frequently in female politicians (13.9%). The stress category AP1 ap-
pears most frequently in male presenters (18.0%), least frequently in female presenters 
(13.9%), while stresses of the type AP2 are most frequent in female presenters (8.8%) 
and least frequent in female politicians (6.0%). With regard to most stress categories the 
differences between different groups of speakers (male and female politicians and pre-
senters) are very small: in four categories they are less than one per cent (AP+BHP+SL 
0.3%, AP+SL 0.5%, AP+IHP+BHP+SL 0.7%, AP+IHP+SL 0.9%) and in the next four 
less than 5 per cent (AP2 2.8%, AP+BHP 3.9%, AP1 4.1%, AP+IHP 4.7%), while the 
greatest differences arise in the category AP+IHP+BHP (9.3%), where female politi-
cians in particular stand out, with 44.7 per cent use of this type of stress. 
In the overview of the appearance of sentence stress in individual speakers, it is strik-
ing that in 12 out of 15 speakers stress of the type AP+IHP+BHP predominates, while 
stress of the type AP+BHP is usually (nine times) in second place; stress of the type 
AP+IHP is most often found in third or fourth place (six times). With a reduction of the 
percentage in the category AP+IHP+BHP, most was on average gained by the categories 
AP+BHP and AP+IHP, which even further strengthened the influence of F0 on the suc-
cessful perception of stresses. What is surprising is the (greatest) influence of intensity 
on the perception of stress in the speaker SpkrPm3+12, since in that place, due to the 
tonemic base of the accent of the analysed speaker, the greater influence of the funda-
mental frequency was expected (AP+IHP 26.5%, AP+IHP+BHP 26.1%, AP1 20.5%, 
AP+BHP 17.9%). The analysed material thus confirms the slightly greater influence of F0 
on the auditory perception of sentence stress, while the influence of change in intensity 
plays a slightly smaller role, in spite of the fact that in Slovene (particularly in the media 
and education) non-tonemic or dynamic accent is presumed to be more widespread. This 
overthrows the third hypothesis, which was to a certain extent already hinted at in Vitez’s 
(1995: 39) analysis of French and Slovene sentence intonation:
It is undoubtedly true that the pitch of the fundamental frequency [...] is by far the 
most significant prosodic parameter, which realises within a sentence the forms 
Linguistica_2017_FINAL.indd   133 12.3.2018   13:08:24
134
of sentence intonation. This is why the fact that the key places of intonation 
patterns are precisely the syllables, the core of which consists of accented and 
stressed vowels, is that much more interesting. It is known that for their realisa-
tion particularly the higher values of the other two prosodic factors are typical, 
i.e. sound intensity and duration.
Type AP1 sentence stress, which is relatively common in most speakers, is charac-
terised by the fact that the words with this type of stress were perceived as stressed, but 
at the same time other words within the same segment, on which lay the intensity and 
frequency peak of the segment (often, there was a difference of only a few hundredths 
or tens of a dB or Hz), were perceived as emphasised. This type of emphasis is also 
the bearer of relatively high intensity and frequency values, but the words with these 
values, due to the limitation of one intensity or frequency peak within a segment, were 
placed into a special category.
With regard to 312 stressed words, or 7.1% of all the auditorily perceived sen-
tence stresses (of type AP2) within the corpus, it was impossible to ascertain, using 
our methodology, why they are in certain segments perceived as emphasised, as the 
highest F0 and the highest intensities were measured on other, auditorily unstressed 
words. Because in these words not even lengthened sounds were perceived, the various 
internal and/or external factors of stress perception can only be speculated upon (Ver-
schueren 2000). This confirmed the findings of a number of Slovene and foreign speech 
researchers that sentence stress “is realised not only with the most intensive sound, but 
also with a raised tone” (Podbevšek 2006: 112).
4 CONCLUSION
To study sentence stress in terms of frequency of appearance, intensity, F0 and sound 
lengthening, a spoken corpus was created involving 15 politicians and TV presenters 
on 10 programmes with political content on TV Slovenia. The corpus has a total dura-
tion of 122 minutes and 35 seconds. The methodology used was based on approaches 
established in other European countries and adapted to the goal of the study. Three 
theoretical hypotheses were formulated and then tested on the chosen material. On the 
basis of the results of the study, it can be concluded that in the speech of male politi-
cians and female presenters there is on average a slightly higher occurrence of sentence 
stress than in that of female politicians and male presenters. The sentence stress of 
female politicians and male presenters is on average slightly more intense and these 
speakers have a higher F0 than male politicians and female presenters. In all the speaker 
categories (except male politicians) a change or increase F0 plays a more important role 
in the auditory perception of stress than an increase in the intensity of speech. The study 
also showed that both values, i.e. intensity and F0, are closely connected with each 
other since an increase in one often meant an increase in the other. However, caution is 
needed in the interpretation of the results as they may be influenced by various linguis-
tic, extra-linguistic or completely personal traits of an individual speaker, such as the 
speed and manner of speaking, age, experience, emotional state and so on.
Linguistica_2017_FINAL.indd   134 12.3.2018   13:08:24
135
References
BAKRAN, Juraj (1996) Zvučna slika hrvatskoga govora. Zagreb: Ibis grafika.
BOERSMA, Paul/David WEENINK (2013) Praat: doing phonetics by computer 
[Computer program]. Version 5.3.42, retrieved 2 March 2013 from http://www.
praat.org/
BOLINGER, Dwight (1972) “Accent is Predictable (If you are a Mind Reader).” Lan-
guage 48, 633–644.
DUBĚDA, Tomáš (2005) Jazyky a jejich zvuky. Univerzálie a typologie ve fonetice a 
fonologii. Praga: Univerzita Karlova.
FOX, John (1989) “Analiza TV-dnevnika kao poruke.” Govor 6/1, 75–85.
HUBER, Damjan (2013) Poudarek in pavza v standardnem slovenskem govoru. Lju-
bljana: Univerza v Ljubljani.
KOMAR, Smiljana (1996) Funkcija intonacije v diskurzni analizi intervjuja v angleščini 
in slovenščini. Ljubljana: Univerza v Ljubljani.
LADEFOGED, Peter (1975) A Course in Phonetics. New York: Harcourt Brace 
Jovanovich.
PALKOVÁ, Zdena (1997) Fonetika a fonologie češtiny. Prague: Univerzita Karlova.
PODBEVŠEK, Katarina (2006) Govorna interpretacija literarnih besedil v pedagoški 
in umetniški praksi. Ljubljana: Slavistično društvo Slovenije.
STRANGERT, Eva (2003) “Emphasis by Pausing.” Proceedings ICPhS 2003, 29 March 
2016, 612–615. http://www.speech.kth.se/grog/publications/ICPhS_0612.pdf
ŠKARIĆ, Ivo (1991) “Fonetika hrvatskoga književnog jezika.” In: S. Babić et al. (eds), 
Povijesni pregled, glasovi i oblici hrvatskoga književnog jezika: nacrti za grama-
tiku. Zagreb: HAZU, Globus, 61–377.
ŠUŠTARŠIČ, Rastislav (1995) “Naglas in poudarek v angleščini in slovenščini.” 
Slavistična revija 43/2, 157–182.
TOPORIŠIČ, Jože (1972) “Členitev besedila s premori.” Slavistična revija 20/1, 
149–158.
VERSCHUEREN, Jef (2000) Razumeti pragmatiko. Prev. I. Prosenc. Ljubljana: 
Založba *cf.
VITEZ, Primož (1995) Protistavna analiza francoske in slovenske stavčne intonacije. 
Ljubljana: Univerza v Ljubljani.
VITEZ, Primož (1999) “Od idealnih jezikovnih struktur k strategiji realnega govora.” 
Slavistična revija 47/1, 23–48.
VOLK, Jana (2011) Intonacija v spontanem neformalnem govoru slovenskih govorcev 
na dvojezičnem področju Slovenske Istre. Ljubljana: Univerza v Ljubljani.
Linguistica_2017_FINAL.indd   135 12.3.2018   13:08:24
136
Abstract
SENTENCE STRESS IN SLOVENE MEDIA SPEECH
Media speech, at least on national radio and television, is one of the bearers of 
standard speech in Slovenia. Thanks to its relatively wide distribution and influence 
on the speech habitis of a broad circle of viewers and listeners, research into media 
speech is easy to justify. In this contribution, on the basis of three methodological 
approaches and with the help of the programme Praat, an analysis is carried out into 
a hitherto poorly researched prosodic phenomenon in Slovene – sentence stress. The 
study focuses on sentence stress with regard to frequency, intensity, tone and sound 
lengthening. Using a corpus taken from 10 television programmes with political con-
tent, with a total duration of two hours, three hypotheses were examined: 1) that the 
speech of male/female politicians contains sentence stress more frequently than that 
of male/female TV presenters; 2) that sentence stress used by female/male politicians 
is on average stronger in intensity (dB) and has a higher F0 (Hz) than sentence stress 
used by male/female presenters; and 3) that in terms of auditory perception of sentence 
stress, the most important role is played by a change/increase in speech intensity and a 
slightly smaller role by the raising of F0 or the lengthening of sounds.
Keywords: auditory-experimental analysis, media speech, prosody/sentence phonet-
ics, sentence stress
Povzetek
POUDAREK V SLOVENSKEM MEDIJSKEM GOVORU
Medijski govor (vsaj nacionalne radiotelevizije) uvrščamo med nosilce govornega 
standarda na Slovenskem. Zaradi njegove razmeroma velike razširjenosti in vplivnosti 
na govorne navade poslušalcev oz. gledalcev, tj. uporabnikov jezika, je raziskovanje 
jezika v okviru medijskega govora vsekakor smiselno. V prispevku na podlagi treh 
metodoloških pristopov ter s pomočjo programa za analizo govora Praat analiziramo 
eno izmed besedilnofonetičnih značilnosti – poudarek –, ki je v slovenskem jeziko-
slovju razmeroma slabo raziskana. Prispevek je usmerjen v raziskovanje poudarkov z 
vidika pogostosti pojavljanja, jakosti, osnovnega tona in podaljševanja glasov. V okvi-
ru dvournega korpusa desetih TV-oddaj s politično vsebino smo preverili tri teoretične 
hipoteze, 1) da je v govoru politikov/političark prisotnih več poudarkov kot v govoru 
voditeljev/voditeljic TV-oddaj, 2) da so poudarki politikov/političark v povprečju ja-
kostno močnejši (dB) in imajo višji F0 (Hz) kot poudarki voditeljev/voditeljic ter 3) da 
ima pri slušni zaznavi poudarkov najpomembnejšo vlogo sprememba/povečanje jako-
sti govora, nekoliko manjšo pa povišanje F0 ali podaljševanje glasov.
Ključne besede: avditivno-eksperimentalna analiza, medijski govor, besedilna fone-
tika, poudarek
Linguistica_2017_FINAL.indd   136 12.3.2018   13:08:24