Darinka Verdonik

Dr. Darinka Verdonik


Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 November 2024 in Language Resources and Evaluation
Verified
Journal Article
Open access
PDF file
DVDarinka Verdonik
Darinka Verdonik
ABAndreja Bizjak
Andreja Bizjak
AZAndrej Zgank
Andrej Zgank
MMMirjam Sepesy Maučec
Mirjam Sepesy Maučec
MTMitja Trojar
Mitja Trojar
Simon DobrišekSimon Dobrišek
Simon Dobrišek
The paper details the creation of an open access speech corpus for a less-resourced language, covering the diversity in terms of accents, dialects, speech styles and demographic characteristics that exist in the target population. Three primary challenges are identified that impact the time and cost efficiency of such a speech corpus development significantly. These challenges are (1) Managing copyrights, personality rights and personal data protection; (2) Obtaining precise word-by-word transcriptions with segmentation into meaningful segments and annotation of speaker exchanges; (3) Managing the new collaborators needed for field recording and manual transcribing or the correction of transcriptions, along with a strictly-defined workflow and platform for data collection. Several strategies are proposed to address each of these challenges, and the experiences are described from the creation of the Slovenian ARTUR corpus regarding these challenges. The ARTUR corpus comprises 1,000 h of carefully selected recordings, with 880 h accompanied by precise, manually checked or manually created transcriptions. It is freely available in the CLARIN.SI repository under the CC BY-SA 4.0 licence. Part of its data was used to upgrade the Slovenian reference speech corpus GOS.

1 March 2016 in International Journal of Lexicography
Verified
Journal Article
Closed access
PDF file
DVDarinka Verdonik
Darinka Verdonik
MMMirjam Sepesy Maučec
Mirjam Sepesy Maučec
This paper presents an investigation of what is gained in the process of dictionary creation by using a speech reference corpus of one million words in conjunction with a huge written reference corpus. It also analyses how much additional effort this requires. Collecting spoken data takes a great deal of effort, and existing speech corpora are rather insignificant in size compared to written corpora, which represent the main and often only source of lexicographic information. However, it is clear that the use of spoken and written language differs with regard to lexical patterns and collocations, and that written corpora, irrespective of their size and structure, cannot provide sufficient data for the description of lexical features in spoken language use. The results demonstrate that even a small speech corpus of one million words provides additional information about the most common, the most general and the most typical spoken usages.

4 January 2021
Verified
Book Chapter
Open access
PDF file
IMIzidor Mlakar
Izidor Mlakar
MRMatej Rojc
Matej Rojc
DVDarinka Verdonik
Darinka Verdonik
SMSimona Majhenič
Simona Majhenič
The present research explores non-verbal behavior that accompanies the management of turns in naturally occurring conversations. To analyze turn management, we implemented the ISO 24617-2 multidimensional dialog act annotation scheme. The classification of the communicative intent of non-verbal behavior was performed with the annotation scheme for spontaneous authentic communication called the EVA annotation scheme. Both dialog acts and non-verbal communicative intent were observed according to their underlying nature and information exchange channel. Both concepts were divided into foreground and background expressions. We hypothesize that turn management dialog acts, being a background expression, co-occur with communication regulators, a class of non-verbal communicative intent, which are also of background nature. Our case analysis confirms this hypothesis. Furthermore, it reveals that another group of non-verbal communicative intent, the deictics, also often accompany turn management dialog acts. As deictics can be both foreground and background expressions, the premise that background non-verbal communicative intent is interlinked with background dialog acts is upheld. And when deictics were perceived as part of the foreground they co-occurred with foreground dialog acts. Therefore, dialog acts and non-verbal communicative intent share the same underlying nature, which implies a duality of the two concepts.

18 July 2024
Verified
Book Chapter
Open access
PDF file
DVDarinka Verdonik
Darinka Verdonik
MTMitja Trojar
Mitja Trojar
ABAndreja Bizjak
Andreja Bizjak
Zapisovanje govora v govornih korpusih je nedvomno časovno največji vložek v procesu izdelave govornega korpusa in pomemben razlog, da so govorni korpusi neprimerno manjši od pisnih. Zapis govora je prevod iz izvorno večmodalnega kanala komunikacije, v katerem verbalno izraženi pomen sooblikujejo glas in način govora, govorica telesa in situacija, v kateri poteka komunikacija, v eno, pisno modalnost. Zaradi variabilnosti govora na vseh jezikovnih ravneh se zapisovalec pri tem nenehno sooča z vprašanjem, kako naj to, kar sliši, zapiše. Da bi bil zapis čim bolj verodostojen, hkrati pa časovno vseeno izvedljiv za velik obseg gradiv, se je tako poleg standardiziranega zapisa vzpostavil tudi tako imenovani pogovorni zapis, ki sledi načelu zapiši, kakor je izgovorjeno. Toda dvojni zapis zahteva dodaten trud, zato v tem prispevku kritično prevprašujemo njegovo utemeljenost glede na prakse drugod, zahtevan dodaten trud in njegove prednosti ter kritično analiziramo še druga težavnejša vprašanja zapisovanja.

18 October 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
MMMirjam Sepesy Maučec
Mirjam Sepesy Maučec
DVDarinka Verdonik
Darinka Verdonik
GDGregor Donaj
Gregor Donaj
Sequence-to-sequence models have been applied to many challenging problems, including those in text and speech technologies. Normalization is one of them. It refers to transforming non-standard language forms into their standard counterparts. Non-standard language forms come from different written and spoken sources. This paper deals with one such source, namely speech from the less-resourced highly inflected Slovenian language. The paper explores speech corpora recently collected in public and private environments. We analyze the efficiencies of three sequence-to-sequence models for automatic normalization from literal transcriptions to standard forms. Experiments were performed using words, subwords, and characters as basic units for normalization. In the article, we demonstrate that the superiority of the approach is linked to the choice of the basic modeling unit. Statistical models prefer words, while neural network-based models prefer characters. The experimental results show that the best results are obtained with neural architectures based on characters. Long short-term memory and transformer architectures gave comparable results. We also present a novel analysis tool, which we use for in-depth error analysis of results obtained by character-based models. This analysis showed that systems with similar overall results can differ in the performance for different types of errors. Errors obtained with the transformer architecture are easier to correct in the post-editing process. This is an important insight, as creating speech corpora is a time-consuming and costly process. The analysis tool also incorporates two statistical significance tests: approximate randomization and bootstrap resampling. Both statistical tests confirm the improved results of neural network-based models compared to statistical ones.

1 January 2021
Verified
Book
Open access
DVDarinka Verdonik
Darinka Verdonik
FEFakulteta Za Elektrotehniko
Fakulteta Za Elektrotehniko
Communication Skills in Engineering Profession. The present teaching material is adapted for students of computer science and technical sciences in general. It addresses a range of communication topics and tries to illuminate various aspects of private, business and public communication, either from the perspective of text/speech production or comprehension. The topics are: (1) various forms of communication in the workplace with an emphasis on work meetings; (2) various forms of public speaking; (3) some basic theoretical concepts about communication and comprehension; (4) the miscommunication; (5) the importance of active listening; (6) nonverbal communication; (7) characteristics of professional writing; (8) politeness; (9) conflict management; (10) various aspects of persuasion. Učno gradivo je prilagojeno za študente računalništva in na splošno tehničnih ved. Obravnava vrsto komunikacijskih tem in skuša skoznje osvetliti različne vidike zasebne, poslovne in javne komunikacije, bodisi z vidika tvorca bodisi z vidika naslovnika. Obranava: (1) različne oblike komunikacije na delovnem mestu s poudarkom na delovnih sestankih kot eni osrednjih oblik skupnega načrtovanja in izvajanja dela; (2) različne oblike nastopanja pred publiko; (3) nekatere temeljne teoretske pojme, s pomočjo katerih skušamo nekoliko bolje razumeti, kako razumemo; (4) problematiko pomanjkljivega sporazumevanja; (5) pomen dobrega poslušanja; (6) neverbalno komunikacijo; (7) značilnosti strokovnega pisanja; (8) vljudnost; (9) izogibanje oz. obvladovanje konfliktov; (10) različne vidike prepoznavanja in uporabe prepričevalnih tehnik.