Tatjana Balažic Bulc UDK 811.163.6:811.163.42'243:801.61
University of Ljubljana*
Vesna Požgaj Hadži
University of Ljubljana**
Damir Horga
University of Zagreb***
SPEECH FLUEn Cy : A r ESULT OF Or AL
LAn g UAg E Pr OFICIEn Cy ?
1 In Tr ODUCTIOn
When communicating with speakers who speak a foreign language well, we often
describe their speech to be fluent, which means it is very similar to the speech of na-
tive speakers. However, what does it actually mean to speak fluently? According to
Lennon (1990), speech fluency has two meanings. In the broader sense, fluency seems
to mean global oral proficiency, and in the narrower sense, it can be considered as a
component of oral proficiency. In other words, when we talk about fluency in the
broader sense we refer to language proficiency, that is, to its general realisation. On
the other hand, when we talk about fluency in the narrower sense, we refer to one of
the aspects of language proficiency, that is, to the actual temporal aspect which has
been studied more intensely since the 1990s within the theory of applied linguistics
and foreign language acquisition. It is in the latter sense that we look at the issue of
fluency in this paper in which we explore the relation between the level of language
proficiency and speech fluency. 
2 FLUEn Cy In FOr EIg n LAn g UAg E ACq UISITIOn 
The issue of fluency has been the object of study for decades. Kormos and Dénes
(2004: 147) describe the development of the field and indicate that one of the first re-
searches of the issue was by Fillmore in 1979 in which the author defined four differ-
ent abilities which reflect speech fluency in the first language (L1): a. the ability to
talk at length with few pauses; b. the ability to talk in coherent, reasoned and seman-
87
*
Author’s address: Filozofska fakulteta, Oddelek za slavistiko, Aškerčeva 2, Ljubljana, Slovenia,
Email: tatjana.balazic-bulc@guest.arnes.si
**
Author’s address: Filozofska fakulteta, Oddelek za slavistiko, Aškerčeva 2, Ljubljana, Slovenia,
Email: vesna.hadzi@guest.arnes.si
***
Author’s address: Filozofski fakultet, Odsjek za fonetiku, Ivana Lučića 3, Zagreb, Croatia, Email:
dhorga@ffzg.hr

tically dense sentences; c. the ability to have appropriate things to say in a wide range
of contexts; and d. the ability to be creative and imaginative in language use (Fillmore
1979: 93). It should be noted that most of those abilities are abstract and are thus
difficult to be researched or confirmed scientifically.
In the 1980s first researches of fluency in the foreign language (LF) have been car-
ried out, motivated by an increased interest in communicative language teaching. By
studying LF performance and proficiency, the researchers have, in most cases, agreed
that the two categories are “multi-componential in nature, and that their principal di-
mensions can be adequately, and comprehensively, captured by the notions of com-
plexity, accuracy and fluency” (Housen/Kuiken 2009: 461). The authors continue that
these three components “have been used both as performance descriptors for the oral
and written assessment of language learners as well as indicators of learners’ profi-
ciency underlying their performance; they have also been used for measuring progress
in language learning.” Based on that, we can conclude that the three components, to
which Skehan (2009) later adds the fourth, not less significant component – lexis, de-
termine the level of language proficiency. In other words, “it is assumed that a profi-
cient speaker will be able to perform tasks fluently and accurately, using complex
language” (Ellis 2009: 475). However, the abovementioned components do not appear
at the same level. As Housen and Kuiken (2009: 462) state, complexity and accuracy
are primarily linked to the current state of the learners’ interlanguage knowledge,
whereas fluency is primarily related to learners’ control over their linguistic LF knowl-
edge, as reflected in the speed and ease with which they access relevant LF informa-
tion to communicate meanings in real time. 
Speech fluency in foreign language acquisition is, therefore, one of the descriptors
of LF oral production. Ellis (2009: 475; based on Skehan and Foster 1999) defines flu-
ency as “the capacity to use language in real time, to emphasize meanings, possibly
drawing on more lexicalized systems.” Housen and Kuiken (2009: 463) highlight its
multicomponentiality which is made up of different subdimensions, such as speed flu-
ency, breakdown fluency and repair fluency. Skehan (2009: 513) is of similar opinion,
stating that regarding fluency “a range of measures are available, broadly examining: a.
breakdown (dys)fluency, indexed by pausing; b. repair (dys)fluency, indexed by measures
such as reformulation, repetition, false starts, and replacements; and c. speed, with
measures such as syllables per minute.” In this research fluent speech is defined as
speech at a natural rate without hesitations, pauses, repetitions, reformulations, filler
words and filled or unfilled pauses in processing, appropriate to the informative and
communicative load of expression, all of which presuppose efficient and coordinated
functioning of all levels of oral production. Although speakers strive for ideal speech,
disfluency cannot be avoided. There are various reasons for this, among which the fol-
lowing are in our opinion certainly the most relevant: informative load of the expression
and the related complexity of information processing, regulators of communicative re-
lations between the speaker and the listener, and in the LF, above all, the imperfect or
incomplete mechanism of oral production (see Požgaj Hadži et al. 2012).
88

3 r ESEAr CH
Although fluency is one of qualitative descriptors in the assessment of oral profi-
ciency in Common European Framework of Reference for Languages: Learning, Teaching,
Assessment (CEFr L; among which are also range, accuracy, interaction and coher-
ence), in teaching and learning Croatian (and Slovenian) as LF, speech fluency does
not receive sufficient attention. This is confirmed by recent textbooks for Croatian as
LF which include only a limited number of exercises in spontaneous speech, while
most exercises are based on the reproduction of content from the text found in the
unit, or are designed for practicing grammar structures, that is, accuracy of speech.
The fact that this field has not been sufficiently researched in both Croatian and
Slovenian and that the studies so far (Horga/Požgaj Hadži 2004, Požgaj Hadži and
others 2012) have left many questions unanswered has motivated us to continue with
the researches of speech fluency in LF.
3.1 r esearch aim
The aim of this research is to find an answer to the question whether speech flu-
ency is influenced by language proficiency or by some other abilities of the speaker.
Based on Slovenian as L1 and Croatian as LF, we explore the differences in speed flu-
ency, i.e. speech rate, articulation rate and mean length of runs, breakdown fluency,
i.e. duration, frequency and position of filled and unfilled pauses, and repair fluency,
i.e. disfluencies such as repetitions, restarts and repairs. Special attention is given to
filler words which do not carry semantic meaning themselves, but are most often a
sign of speech processing. 
3.2 Corpus
There were 11 participants involved in the research, all of whom were students at
the first year of Croatian, Serbian and Macedonian Language and Literature at the
University of Ljubljana Faculty of Arts Department of Slavic Studies (academic year
2008/2009) with Slovenian as L1 and Croatian as LF. Their task was to participate in
an informal discussion on a familiar topic: Student life – advantages and disadvantages.
The topic and formal length of the discussion (15 minutes per group) were defined in
advance. The participants had an opportunity to discuss the topic a week before the
recording in langue practice classes so it can be said that the recorded speech was pre-
pared. There were three participants in each discussion, and the discussion was led by
a moderator (also a student) who took care that each participant spoke for the same
length of time and who motivated the speakers by asking additional questions. The
distance between the speakers was appropriate for a conversation and could be de-
scribed as personal space or the zone of proxemic distance (45 to 120 cm). The chairs
were set up in a circle as a motivation for democratic and cooperative communication.
The recording took place in an office at the Faculty, a familiar surrounding for the stu-
dents, and speech was recorded with a VHS camcorder. The recording was carried out
89

two times in the interval of two weeks. During the first recording the students discussed
the issue in Croatian (LF) and during the second recording in Slovenian (L1). 
As was mentioned before, the participants’ task was to talk in an informal discus-
sion, which is one of the ways to develop interactive competence, and interaction along
with reception, production and mediation, that is, oral interpreting and translating, is
one of language activities within CEFr L which helps develop communicative compe-
tence. Interaction is described in CEFr L as an activity which includes at least two per-
sons whose communication is based on production and reproduction which alternate
and sometimes even overlap. However, the listener usually predicts what the speaker
will say once he or she begins to talk and the listener prepares his or her response,
which means interaction is far more than learning how to receive and produce an ex-
pression – that is why interactive competence is key for successful communication.
According to CEFr L, the participants’ language proficiency level of Croatian (LF)
is B2+, which means, in terms of interaction, that the speaker can use the language
fluently, accurately and efficiently to talk about a wide range of general, academic and
business topics or topics relating to leisure, and can clearly indicate relations among
ideas. The speaker can also communicate naturally, with a good level of grammatical
accuracy, without leaving an impression that he or she must limit him- or herself in
what he or she wants to say, using the level of formality appropriate to the situation. 
3.3 Methodology
The research consisted of two parts. In the first part, the recorded speech was tran-
scribed and then, using the programme Praat, speech analysis of the recordings was
carried out to get speech rate, articulation rate, mean length of runs and the length
and number of breakdowns (see figure 1).
Figure 1. Duration measurment of disfluency (correction) in Praat.
90

Different classifications of pauses can be found in literature. In this research we
classify pauses according to Housen and Kuiken (2009) into two kinds (see Table 1).
On the one hand, there are breakdowns which include a. unfilled pauses and filled
pauses, which are recognised as vocalised hesitation (eee, em, hm, etc.) or as vocalised
lengthening (e.g. connector i being pronounced as i:). On the other hand, there is a
special kind of pause – repair, which includes repeats (whether of a part of a word, of
the entire word, or the entire phrase), corrections, reformulations and false starts. As
a special kind of non-fluent speech we indicate filler words which do not carry seman-
tic meaning themselves, but rather indicate the thought process going on in the brain
during speech, thus they can be defined as time-creating devices. By using the words
such as dobro ‘well’, zapravo ‘actually’, znaš ‘you know’, mislim ‘I think’, samo da pomis-
lim ‘let me think’, koliko znam ‘as far as I know’, etc. the speaker does not interrupt
speech and thus leaves an impression of fluency. However, overuse of such words is a
sign of disfluency.
Table 1: Classification of pauses. 
In the second part of the research, corpus method was used to analyse different
types of disfluency in greater detail. For that purpose two specialised electronic cor-
pora have been built: leraner’ corpus FLU-Cr (3.083 tokens) which includes the
speech of participants in Croatian (LF) and the control corpus FLU-SI (4.577 tokens)
which includes the speech of participants in Slovenian (L1). The marking of corpus
is problem-oriented, which means that only examples of disfluency are marked in the
corpus. Marking was done manually, and speech was partitioned into smaller units,
91
Pauses
Breakdowns Unfilled pause pač način … študija in to
Filled pause Vocal hesitation delam eee potem hodim na faks
Vocal lengthening
hm paa mesečno za autobus je treba
plačat
r epairs r epeat več ne vem več časa je na splošno
Correction
k mam faks urnik tko razdrobljen čez
ceu dan
r eformulation
tko da se lahko zdej hm da bi se
morala naučit samostojnega dela
False start živim u hiši e privat hiši in mi je ušeč
k si lahk hm če mi kdaj ne paše grem
na bon jest
+ Filler words sej je fajn mislm tko dost pocen je

i.e. utterances. Although the analysed sample is relatively small, we believe it is suffi-
ciently representative to indicate tendencies in speech fluency between L1 and LF. 
4 r ESULTS An D DISCUSSIOn
As it was mentioned before, the aim of the research was to find differences in oral
production between L1 (Slovenian) and LF (Croatian). In the following part of the
paper research results are presented. 
4.1 Speed fluency
Speed fluency depends on the automatisation of many levels of oral production
and speaking skills and it is expressed in terms of speech rate (Sr ), articualtion rate
(Ar ) and mean length of runs (MLr ). 
4.1.1 Speech rate (SR)
As can be seen in Chart 1, Sr , i.e. the number of syllables spoken per second, in-
cluding pauses, is somewhat faster in LF than in L1. The speakers utter 2.82 syllables
per second in Slovenian and 3.09 syllables per second in Croatian. 
Chart 1: Speech rate in Slovenian (L1) and Croatian (LF)
Similar results in speech rate are found in the research by Požgaj Hadži et al. (2012)
in which Slovenian participants read a text somewhat faster in LF (4.7 syllables per sec-
ond) than in L1 (4.6 syllables per second). The differences were also noted in the re-
search by Horga and Požgaj Hadži (2004) in which Sr in LF is 3.8 syllables per second. 
The cause of those differences in research results can undoubtedly be found in
different types of language activity. Participants have a faster Sr in oral production
when reading (4.7 sps) and somewhat slower Sr when describing (3.8 sps), whereas
Sr is the slowest in oral interaction (3.9 sps). This confirms the fact that interactive
92

discourse (informal discussion) alongside oral production also requires other cognitive
abilities, which makes the speech slower. 
It is interesting to note that Sr in L1 deviates significantly from the usual speech
of native speakers. According to research by Horga and Mukić (2000), Sr in sponta-
neous speech by native speakers of Croatian on HTV (Croatian national television
network) is 4.8 syllables per second. Such difference is most likely the result of the
participants’ different speaking experience. Although the participants in both studies
talk about a known topic, experts are far more experienced speakers when speaking
in public is concerned, thus their speech is faster and more complex. On the other
hand, students have not as yet acquired such speaking experience and, unfortunately,
the education system does not pay sufficient attention to the issue and hence does
not provide enough opportunities for practicing various oral interactions. 
4.1.2 Articulation rate (AR)
Similarly, articulation rate (Ar ), i.e. the number of syllables spoken per second,
excluding unfilled pauses, is somewhat slower in Slovenian (3.33 sps) and faster in
Croatian (3.76 sps) (Chart 2). 
Chart 2: Articulation rate in Slovenian (L1) and Croatian (LF)
This means that in the research Slovenian participants articulate more slowly when
speaking L1 and faster when speaking LF, which is surprising. Such results might be
influenced, on the one hand, by jitters because it was the participants’ first recording
of that kind, and on the other hand, by the lack of motivation in the second recording
(L1) because they were discussing the topic for the second or the third time. Answers
to these questions can only be found by conducting further research. 
93

4.1.3 Mean length of runs (MLR)
However, if we look at mean length of runs, i.e. an average number of syllables
spoken between two pauses regardless of semantic and syntactic criteria of expression
setup, in L1 and in LF, the case is reverse. MLr in Slovenian is significantly longer
(2.46 s) than in Croatian (1.8 s) (see Chart 3).
Chart 3: Mean length of runs
Similar results appeared in research by Požgaj Hadži et al. (2012) in which par-
ticipants reading a text in L1 achieve an MLr of 2.0 seconds and in LF an MLr of
1.7 seconds.
It is interesting to note that MLr in L1 matches MLr found in the research by
Horga and Mukić (2000) which showed that there is a difference between male and
female speakers. Male speakers have an average MLr of 1.64 s, whereas female speak-
ers have an MLr of 2.87 s.
4.2 Breakdown fluency
Breakdown in speaking is described as an unfilled pause expressed as silence or a
filled pause expressed as a vocalised hesitation (eee, em, hm, etc.) or vocalised length-
ening where in most cases the last sound in the word is lengthened, most often the
vowel (pa:, i:, etc.). A question arises as to what can be considered a breakdown. Differ-
ent studies have different values of breakdowns; our research identifies a pause as a
breakdown lasting at least 0.2 to 0.25 seconds.
4.2.1 Breakdown duration
As can be seen in Chart 4, average pause duration (filled and unfilled) is 0.63 s in
Slovenian and 0.68 s in Croatian. It can be concluded that the results indicate a ten-
dency for somewhat longer pauses in LF than in L1, which was not confirmed by the
research of Požgaj Hadži et al. (2012). 
94

Chart 4: Breakdown duration
It is interesting to note that pauses are significantly shorter when the text is read
– in Slovenian participants the duration of the pauses is 0.45 s when reading a text in
L1 and 0.43 s when reading a text in LF (Požgaj Hadži et al. 2012). 
4.2.2 Number of breakdowns
Whereas the number of unfilled pauses in both languages is approximately the
same, i.e. 16.35 per minute in L1 and 16.61 per minute in LF, there are a lot more
filled pauses in LF than in L1. In Croatian, there are 11.93 filled pauses per minute
and in Slovenian there are 8.38. The results can be seen in Chart 5.
Chart 5: n umber of pauses (filled and unfilled)
95

4.3 r epair fluency
According to n ooteboom (1980), oral production is largely subject to refinement
during production, although speakers repair only 50% of errors they make. However, re-
pair does not necessarily mean there is an error – it can be one of the forms of disfluency
when the speaker hesitates because he or she needs more time for further planning of
speech and speaking. r epair is carried out by various means. In our research the most
frequent type of repair is repetition, then reformulation, false start and correction.
4.3.1 Number of repairs
The number of repairs is also somewhat higher in LF than in L1. There are 3.96
such pauses per minute in Slovenian and 4.72 such pauses per minute in Croatian
(Chart 6).
Chart 6: n umber of repairs per minute
Higher number of repairs in LF was confirmed in previous studies. When reading
texts in L1 there are on average 23 repairs per minute and in LF 26 repairs per minute
(Požgaj Hadži et al. 2012), whereas in spoken texts of Slovenian participants there are
32.9 repairs per minute in LF, or almost every 2 seconds (in our research 33.26 in LF
and 28.96 in L1), which is significantly more than 7.6 repairs per minute in sponta-
neous speech on the radio (Horga/Požgaj Hadži 2004).
A more detailed frequency of filled pauses was calculated in corpus analysis
(Table 2).
96

Table 2: Average frequency of filled pauses per minute
As can be seen in Table 2, when speaking in LF the participants use somewhat
less repeats and somewhat more other forms of repairs. 
4.4 Filler words
The biggest difference between L1 and LF speech was noted in using filler words.
When speaking Slovenian (L1) the speakers use 8.53 filler words per minute and when
speaking Croatian (LF) they use only 1.31 filler words (Table 3). 
Table 3: Filler words
It is interesting to note that when describing a picture in LF Slovenian participants
used 8.4 filler words per minute (Požgaj Hadži et al. 2012), which is almost the same
result that we got in this research. However, it is important to note that the participants
from previous research were at the second and third year of study, and were at a higher
level of language proficiency.
5 COn CLUSIOn
The aim of our research was to find an answer to the question to what extent
speech fluency depends on language proficiency. To find the answer, we compared
the speech of the participants in L1 (Slovenian) and in LF (Croatian). Chambers (1197:
541) points out the following differences in speech fluency between L1 and LF: a. the
frequency of pauses rather than the length; b. the length of run; c. the place of pauses
in an utterance; d. the transfer (or not) of pausing pattern from L1 to LF. In this re-
search we primarily explored the first two, that is, the length and frequency of pauses,
mean length of runs and speech and articulation rate. It was shown that there are no
97
FLU-SI FLU-Cr
Type of pause n o./min
n o. of
particip.
n o./min
n o. of
particip.
Difference
SI:Cr
r epeats 2.3 10 2.25 9 - 0.05
Correction 0.36 4 0.68 6 + 0.32
r eformulation 0.63 6 0.95 6 + 0.32
False start 0.67 6 0.84 7 + 0.17
total 3.96 4.72
FLU-SI FLU-Cr
Type of pause n o./min
n o. of par-
ticip.
n o./min
n o. of par-
ticip.
Difference
SI:Cr
Filler word 8,53 11 1,31 10 - 7,22

significant differences in the rate of speech and articulation between L1 and LF. How-
ever, speech rate in LF is different in different discourse types. The more complex the
discourse, the shorter the speech rate (reading 4.7 sps, description 3.8 sps, discussion
3.09 sps). Values of mean length of runs confirm that speech in L1 is more fluent be-
cause it is 1.36 seconds longer than in LF. Therefore, we can conclude that speech
fluency is primarily influenced by the speaker’s language proficiency, but cognitive
abilities are also important. 
As for pauses, the difference in the length of breakdowns and the frequency of un-
filled pauses and repairs between L1 and LF is almost non-existent, whereas unfilled
pauses are more frequent in LF (+ 3.55 pauses per minute). Most differences are found
in the use of filler words that appear in L1 7.22 more times in a minute than in LF.
Based on the results from the research we can agree with the assumption that “becoming
fluent therefore is not about speaking faster (articulation rate), but about pausing less
often and pausing at the appropriate junctures in an utterance” (Chambers 1997: 540).
Difference in speech fluency between participants in the research and professional
speakers point to the problem of education where insufficient attention is paid to the
issues of fluency and acquisition of more demanding language activities, such as in-
teractive discourse. Such language activities require more than just knowing the lan-
guage, thus they should be incorporated into curricula of first and foreign languages.
As Kormos and Dénes (2004: 160) point out, “if the aim of language teaching is to
help students to become fluent speakers, both exercises that prompt students to ex-
press their communicative intent within the limited time-constraints of real-life in-
teractions and those that promote accuracy should be applied.” Those are primarily
exercises which encourage speakers to engage in independent oral production, such
as discussions and debates, games, simulations, role-plays, etc., in which speech flu-
ency comes before accuracy.
The research has left many questions unanswered. First of all, it is important to
look into where the pauses appear especially because the research confirmed the fact
that “what distinguishes native and non-native pausing is more likely to be where
pauses occur rather than that they occur” (Skehan 2009: 513, according to Davies
2003). Furthermore, it would be interesting to find out what is considered fluent
speech, that is, how speech fluency is evaluated in first and foreign language teaching.
If we look at the descriptors for speech fluency in CEFr L, we can see that they are
too general. Successful evaluation in teaching would certainly require some time pa-
rameters, such as speech rate, mean length of runs, etc.
r eferences
CEFr L: Common European Framework of r eference for Languages: Learning, Teach-
ing, Assessment. http://www.coe.int/t/dg4/linguistic/Source/Framework_En .pdf 
CHAMBEr S, Francine (1997) What do we mean by fluency? System 25/4, 535–544.
http://www.sciencedirect.com/science/journal/0346251X/25
98

ELLIS, r od (2009) The differential effects of three types of task planning on the
fluency, complexity, and accuracy in L2 oral production. Applied Linguistics 30/4,
474–509. http://applij.oxfordjournals.org/content/30/4.toc
FILLMOr E, Charles J. (1979) On fluency. In: C. Fillmore/D. Kempler/W. S.-y. Wang
(eds), Individual Differences in Language Ability and Language Behavior. n ew york:
Academic. 85–101.
HOr g A, Damir (2008) r epetitions in interrupted speech production. Beszed Kutatas
1, 157–171. 
HOr g A, Damir/Vesna POž g AJ HADž I (2004) g ovorna fluentnost u stranom jeziku: hrvat-
ski i slovenski. In: D. Stolac/n . Ivanetić/B. Pritchard (eds), Suvremena kretanja u
nastavi stranih jezika. r ijeka: Hrvatsko društvo za primijenjenu lingvistiku. 175–185.
HOr g A, Damir/Igor MUKIć (2000) n eki vremenski parametri govora u dnevnicima
HTV-a. Govor 27/2, 105–121.
HOr g A, Damir/Vesna POž g AJ HADž I/Marko LIKEr (2012): Vremenski parametri go-
vora. In: V. Požgaj Hadži and others Izazovi kontrastivne lingvistike. Ljubljana: Znan-
stvena založba Filozofske fakultete, 115–125.
HOUSEn , Alex/Folkert KUIKEn (2009) Complexity, Accuracy, and Fluency in second
language acquisition. Applied Linguistics 30/4, 461–473. http://applij.oxfordjour-
nals.org/content/30/4.toc
KOr MOS, Judit/Mariann Dé n ES (2004) Exploring measures and perceptions of fluency
in the speech of second language learners. System 32, 145–164. http://www.scien-
cedirect.com/science/journal/0346251X/32/2
LEn n On , Paul (1990) Investigating fluency in EFL: A quantitative approach. Lan-
guage Learning 40/3, 387–412.
http://onlinelibrary.wiley.com/doi/10.1111/lang.1990.40.issue-3/issuetoc
n OOTEBOOM, Sieb g . (1980) Speaking and unspeaking: detection and correction of
phonological and lexical errors in spontaneous speech. In: V.A. Fromkin (ed), Er-
rors in Linguistic Performance: Slips of the Tongue, Ear, Pen, and Hand. n ew york:
Academic Press. 87–96. 
POž g AJ HADž I, Vesna et al. (2012) Izazovi kontrastivne lingvistike. Ljubljana: Znanstvena
založba Filozofske fakultete.
SKEHAn , Peter (2009) Modelling second language performance: integrating comple-
xity, accuracy, fluency, and lexis. Applied Linguistics 30/4, 510–532.
ZEr OJ: Zajednički europski referentni okvir za jezike: učenje, poučavanje, vrednovanje. Za-
greb: Školska knjiga. 2005. 
99

Abstract
SPEECH FLUEn Cy : A r ESULT OF Or AL LAn g UAg E Pr OFICIEn Cy ?
The aim of this paper is to answer the question of the influence of language proficiency on
speech fluency in relation to speakers’ other cognitive abilities by comparing the speech of re-
search participants who speak Slovenian as L1 and Croatian as LF. By using the method of
acoustic and corpus analysis, the values of speech rate, articulation rate, mean length of runs
and the length and frequency of certain pauses are presented. 
Keywords: oral language proficiency, speech fluency, speech rate, pauses, filler words.
Povzetek
g OVOr n A FLUEn Tn OST: r EZULTAT g OVOr n E JEZIKOVn E ZMOž n OSTI?
n a primeru govorcev slovenščine kot prvega in hrvaščine kot tujega jezika poskušamo v
prispevku odgovoriti na vprašanje, koliko na govorno fluentnost v tujem jeziku vpliva stopnja
znanja tujega jezika, koliko pa nanjo vplivajo druge kognitivne sposobnosti govorca. Z meto-
dologijo akustične in korpusne analize so bile v raziskavi izračunane vrednosti časovnih para-
metrov hitrosti govora, hitrosti izgovora, trajanja govornega bloka in trajanja disfluentnosti ter
frekventnosti pojavitev posameznih vrst disfluentosti.
Ključne besede: ustno znanje jezika, govorna fluentnost, hitrost govora, premori, mašila.
100