Beszédkutatás

A zöngeképzés változásai egy érzelemindukciós számítógépes játék során

Márton Bartók — 2019-06-01

Phonatory changes during emotion-inducing game events

Phonatory changes during emotion-inducing game events: the effect of discrepancy from expectations and goal conduciveness.

This paper aims to describe how phonation changes during emotion-inducing stimuli. 34 Hungarian speakers (17 female, 17 male) were asked to compete in a simple computerised guessing game, using voice commands to proceed after faced with the result of their guess. We expected that acoustic measures taken on these voice commands differ based on two affective components: goal conduciveness (successful or unsuccessful result) and discrepancy from expectations (unexpected or expected result based on the uncertainty of the guess).

According to the results, only female subjects show phonatory variation as a result of varying emotional states: their fundamental frequency is higher at unexpected game events and their phonation is less breathy (lower H1-H2) at unexpected, unsuccessful events. Both of these changes can be caused by higher muscle tension in unexpected, unsuccessful situations and lower tension when experiencing goal conducive, expected events.

Megnyilatkozáskezdő magánhangzók glottális jelöltsége a szintaktikai pozíció és a magánhangzó-minőség függvényében

Alexandra Markó — 2019-06-01

Glottal marking of utterance-initial vowels as a function of the syntactic position and the vowel quality

In the present study, the glottal marking of utterance-initially appearing vowels was analysed with respect to the syntactic position, the vowel quality, the formal characteristics of the glottal marking (glottal stop, glottalization, and their combination), and the inter-speaker variability.
Four members of the Hungarian vowel-inventory were chosen for the analysis: front and high /i/, back and high /u/, front and low /ɛ/ and back and low /ɒ/ (in these examples the feature backness co-varies with lip spreading). From these vowels, V 1 pV 1 structured words (/ipi/, /upu/, /ɛpɛ/, /ɒpɒ/) were constructed, in which we analysed the word-initial vowel. The (pseudo-)words were embedded into meaningful sentences. We analysed the target words in two conditions: they were positioned in pre-focal topic and in focus positions, both occurring sentence-initially. Acoustic recordings were made with 20 female native Hungarian speakers (aged from 19 to 28 years) with an omnidirectional condenser microphone at 44.1 kHz sampling rate. With each participant, 40 target utterances (5 repetitions per each vowel in each condition) and 80 filler utterances (with the same dialogue and sentence construction) were recorded.
The results showed that the frequency of glottal marking did not differ between the analysed utterance-initial syntactic positions (focus and topic), in both cases more than 70% of the vowels were glottally marked. The vowel quality, however, had an effect on the frequency of glottal marking. With respect to vowel openness, we found (in accordance with earlier studies) that open vowels showed glottal marking in a higher frequency than their closed counterparts did. The backness of the vowels, however, did not show the expected tendency, since the frequency of glottal marking was higher in the case of the front vowels. The ratio of the analysed formal variants (glottal stop, glottalization, and their combination) did not differ in terms of syntactic position, although they did with respect to the dimensions of vowel quality. The inter-speaker variability was highly observed in all of the analysed parameters.

Gemináták artikulációs szerveződése a magyarban

Andrea Deme — 2019-06-01

Articulatory organization of geminates in Hungarian

It is traditionally assumed that geminates undergo degemination when being flanked by another consonant in Hungarian. As in Hungarian duration is considered to be the main acoustic cue to the singleton-geminate opposition, it appears valid to study the phonetic implementation of this process in the acoustic domain. However, previous acoustic analyses lead to inconclusive results on the status of the “degeminated” consonant, while articulatory data on Japanese singletons and geminates imply that it is revealing to study degemination on the level of gestural timing. The present study compared gestural organization of geminates, degeminated and singleton consonants in
heterorganic C-clusters, and in intervocalic positions. We obtained EMA data from 10 female speakers of Hungarian (aged 27.7 ys). Consonant durations, plateau durations and tongue rise data showed that degemination does not yield realizations equivalent to intervocalic singletons, and geminates and singletons in clusters showed equally slower tongue rise than that observed in intervocalic singletons.

Explozívák és az őket megelőző magánhangzók időtartama modális fonációval létrehozott beszédben és suttogásban

Bettina Száraz — 2019-06-01

The duration of stops and preconsonantal vowels in modal voicing and
whispering

Whispering is a common, natural way to produce speech, but the way speech sounds are produced in whisper is fundamentally different from that of “normal” speech, i.e., modal voice. In modal voice, the vibration of the vocal folds creates the source of the speech signal, and its presence or absence makes the main difference between voiced and voiceless obstruent consonants in Hungarian. In whisper, however, the vocal folds are abducted so that they do not vibrate, and the air passes between the arytenoid cartilages creating audible turbulent noise, irrespective of the (phonological) voicing of the speech sound produced. Therefore, the question arises, if we can distinguish voiced and voiceless obstruent phonemes in whisper. And if we are able to distinguish voiced and voiceless obstruents, what types of (secondary) coustic cues of voicing may aid our perception to do so? In modal voice, duration of the consonants and duration of the preconsonantal vowels may serve as secondary acoustic cues to voicing, as voiced obstruents are shorter than their voiceless counterpart, and vowels preceding voiced obstruents are longer than those preceding the obstruents’ voiceless counterpart. Previous studies suggest that these acoustic cues may also be sustained in whisper.
The present study investigated the duration of voiced and voiceless stops and the duration of vowels preceding these stops in modal speech and whispering in Hungarian. We hypothesized 1. that voiced stops are shorter, than their voiceless counterpart both in modal speech and whispering, and 2. that the vowels before voiced stops are longer than those before the voiceless counterpart of the stops at hand, again both in modal speech and whispering.
In our experiment, participants were asked to read C 1 VC 1 nonsense words in isolation (with an intonation contour characteristic of one-word sentences), where the two consonants were either of /b p d t ɡ k/, and the vowel was either /ɔ/ or /ɛ/ resulting in 12 target sequences. We analysed the second consonant in these sequences. Each nonsense word was repeated by each speaker five times, both in modal voice, and whisper, and the stimuli were presented randomly to the speakers.
The results showed that duration of voiced stops was significantly shorter than that of voiceless stops, both in modal speech and whisper. Furthermore, vowels before voiced stops were significantly longer than those before voiceless stops, again both in modal speech and whisper. We thus concluded, that the duration of stops and preconsonantal vowels may sustain the voicing opposition of stops in whisper, when the primary acoustic correlate of voicing (the vibration of the vocal folds) is not present. On the basis of our results the question arises,
if these durational correlates of voicing may also be used in as cues to voicing in perception.

Frázisok időzítési sajátosságai az életkor függvényében

Valéria Krepsz — 2019-06-01

Timing characteristics of spontaneous speech across lifespan

The physiological changes that occur as people age can also be traced in temporal patterns of language. The goal of this research is to analyse the timing characteristics of spontaneous speech in four different age groups: children, teenagers, adults and elderly speakers. The main question of the research is whether and how the temporal characteristics of spontaneous speech change from childhood to old age. The speech materials of the study were selected from three speech databases: children’s material (3–10 years) from GABI (Bóna et al. 2014), teenagers (16–17 years) from TiniBEA (Gyarmathy–Neuberger 2015), and adults (20-90 years) from BEA database (Gósy et al. 2012). We analysed the articulation and speech rate of the speakers, the duration of the speech units, as well as the frequency and duration of pauses. The timing values were compared by age and gender of the speakers. The timing of the phrases appearing in the speech production of
teenagers showed similarities with the statements of adults, while in old age the rate of speech, the articulation and the number of words per speech units also decreased. The results of the research contribute to the understanding of the temporal peculiarities of speech, as well as the systematic, objective description of the changes taking place over the course of life.

Narratívák temporális mintázata tinédzserkortól időskorig

Mária Gósy — 2019-06-01

Temporal patterns of narratives across the lifespan

Narrative is an important interactional speech communication form that each generation uses. Narrative production involves organizing and expressing a series of events and facts that were experienced by the speaker. Narrative as specific verbal behavior shares similarities with various styles of spontaneous speech; however, there are marked differences in several important ways like memory processes. There is a specific type of narratives when the speaker is asked to summarize a heard text immediately. The input is based on speech comprehension followed by the usual speech planning processes. Our research question is whether the temporal parameters of narratives show changes from teenagers to old speakers. 56 speakers (ages between 16 and 80 years) forming six age groups were randomly selected from two databases (BEA and TiniBEA) according to their ages. The length of the narratives, the number of words the narratives consisted of, the subjects’ speech tempi, relative frequency and durations of phrases, silent and filled pauses, as well as speaking fluency and individual differences were analyzed (using Praat software). Results showed large overlaps between data of the neighboring age groups. Significant differences could be proved between the three young and the three old groups in a number of temporal parameters, like durations of
phrases and pauses. Some of the measured data of the narratives analyzed reflect the dominant usage of either short-term or long-term memory that depends on age.

Társalgás, beszélőváltás és diskurzusszerveződés új megközelítésben – fonetikai jellemzők és pragmatikai tényezők összefüggései magyar társalgásokban (pilot study)

Ágnes Hámori — 2019-06-01

Conversation, turn-taking and discourse organization: new insights on the interface of phonetics and pragmatics in a Hungarian conversation. A pilot study

One of the most important way of language use is conversation (verbal interaction or spoken discourse). Phonetic details play crucial role in it, however, there are only a few research studies on phonetic aspects of Hungarian conversations and their connections with the pragmatic dimension. The aim of this paper is to analyse the organisation of a Hungarian conversation from a complex pragmatic and phonetic view, bringing together phonetic analytic techniques and methods of Conversation Analysis in the framework of a functional pragmatic approach. The main question was what the characteristics of the turns and turn-taking system were and how phonetic features interact in them with pragmatic aspects. Furthermore, we investigate two less-discussed, but important topics as well: the presence and possible factors of extensive prediction in turn-taking system and the connections of phonetic patterns with global discourse structure units. One recording from Hungarian Spontaneous Speech Database (BEA) was used for the study. Three female speakers participated in the conversation: their age and level of education were the same, and they have known each other for years. The material was manually annotated in Praat, the annotation contained the utterance and the word level of each speaker, the overlapping speech, the turns and turn-takings. The frequency, the duration and main pragmatic characteristics of local discourse units were analysed on the one hand. On the other hand, the global structure and timing of the discourse were analysed in terms of the narrative and interactive discourse periods. The results showed that both short pauses and overlapping speech have been common during the turn-takings. The result also shed lights on that half of the turn- takings were carried out in less than 400 ms; these data refer to the extensive prediction of turn-taking based on syntactic, prosodic and pragmatic features of turns.

A néma szünetek és a hallható levegővétel viszonya a spontán beszédben

Dorottya Gyarmathy — 2019-06-01

The relation of silent pauses and audible breathing in spontaneous speech

Speech is occasionally interrupted by pauses of various length. They are essential parts of human speech. Until the first half of the 20 th century researchers examined the pauses mainly from rhetorical aspects, they analysed the relationship between punctuation of written texts and their spoken realisations. Until now it is not clarified what can be actually considered as pause, how differently pauses can appear in speech, what are their acceptable minimum and maximum durations, and what functions they may have in speech. Empirical researches showed that silent pause is the most common phenomenon in the spontaneous speech, and it has many different functions in spontaneous speech (e.g. breathing, marking syntactic boundaries as well as speech planning difficulties, time for self-repair). There are still many controversies about the relationship between silent pause and breath-taking.

The aim of the present study is to investigate whether breath-taking is subordinated to thinking and is not performed by biological functions in spontaneous speech. Our study focuses on the analysis of the temporal structure of silent pauses and breaths in Hungarian spontaneous speech. We hypothesized that there would be proportional and/or durational differences among pause categories depending on breath-taking.

Results provide detailed information about the relationship between silent pauses and breath-taking. In addition, results can be used in various areas of speech technology, or speech therapy.

A diskurzusjelölő-használat életkori sajátosságai a nyelvi interjú szövegtípusában

Anita Schirm — 2019-06-01

Effects of age on the use of discourse markers in sociolinguistic
interviews

This study analyzes 8 interviews (1,030 minutes) from the Szeged Sociolinguistic Interview project, focusing on the six most frequently used
discourse markers of Hungarian (hát “well”, ugye “[tag]”, persze “of course”, szóval “so”, egyébként “by the way”, and úgyhogy “thus”) in the four age groups of the corpus. These discourse markers were used most by the 60-70-year-olds, followed by the under-18 group, the over-70 group, and finally the young adults/middle-aged group. In every age group the most frequently used discourse marker was hát, its proportion of use compared to the 6 discourse markers’ average was between 46.06% and 82.25%. The results of the various age groups differ not only in the proportions of discourse markers used, but also in their functions, which is indicative of differences in strategies of forward planning of speech. Younger respondents used discourse markers more for self-correction, word search, and mitigation, and much less for signaling attitudes. In contrast, middle-aged and older respondents used discourse markers for conversation management, self-
reflection, and expression of emotions as well. Respondents in the 60-70 age group produced longer narratives and were more deeply involved in stories, their discourse markers signaling more emotional involvement and various speaker attitudes. Narratives produced by the over-70 group were longest, but these respondents produced more off-topic digressions. Still, they used discourse markers also for commenting on stories they produced.

Inszubordinált (függetlenedett) mellékmondatok a magyar beszélt és írott beszélt nyelvben

Csilla Ilona Dér — 2019-06-01

Insubordination in spoken and in written spoken Hungarian

This study is the first attempt to examine insubordinate hogy (’that’)-clauses in spoken and written spoken Hungarian. Insubordination is the phenomenon when a clause has the formal features of subordinate clause but is used as independent, conventionalized main clause (for example: Hogy te milyen kényes vagy! ’How spoilt you are’). We found differences between the spoken and written spoken corpora in terms of insubordination, but we couldn’t find significant gender or age disparities. Hogy-insubordinations preferred to form collocations with discourse markers (tehát ’so’, hát ’well’, zóval ’so’) only in spoken genres, especially in conversations. The most frequent types of hogy-insubordinations had clear pragmatic, mostly metapragmatic functions: they expressed quotation, reformulation, added rhetoricity or emphasized speech acts like saying, questioning.

A magánhangzós hosszúsági fonológiai kontraszt a dajkanyelvben a csecsemő életkorának függvényében

Andrea Deme — 2019-06-01

Vowel length contrast in motherese as a function of age

According to Lindblom’s H&H theory, speech production is speaker-oriented, i.e., in adverse listening conditions, speakers enhance intelligibility by the exaggeration of contrasts to produce „hyperspeech”. As infants are at an early stage of language acquisition, it is often hypothesized that infant directed speech (IDS) is also a type of hyperspeech (as opposed to adult directed speech, ADS, that is more on the hypospeech end of the continuum). Therefore, in IDS exaggeration of (linguistic) contrasts, e.g., phonemic contrasts, is suggested. With respect to the phonological vowel length contrast, there is only a limited amount of data available for IDS. Moreover, these data are also quite inconclusive (showing the expected contrast exaggeration in some cases, but revealing the lack of it in others), leaving us to the possible conclusion that contrast exaggeration is not necessary and/or economical for IDS speech to employ. In the present study we aim to contribute to this field of knowledge through the analysis of Hungarian IDS. Hungarian is quite unique (as opposed to e.g., English or Japanese), as it uses both spectral and durational cues combined to express phonological length contrast: while low vowels are distinguished both by spectral and durational cues, high vowels are claimed to be differentiated mostly by durational cues. We addressed the questions, i) if we find contrast exaggeration in IDS vs. ADS in Hungarian, ii) if durational or spectral cues exaggerate the contrast, and iii) if the patterns of long-short vowel contrasts show an interrelation with the infant’s age between the age of 0 and 8 months.

We analysed /u/–/u:/ and /A/–/a:/ in 22 first-time mothers’ speech. Speakers read target vowels in one-syllable nonsense CVC words introduced as pixie names in a children’s story book to the experimenter (ADS), and then to their child (IDS), in pre-written sentences that contained the names in pitch-accented positions. We recorded IDS and ADS 3 times with each participant, i) the day after the baby was born (0m), ii) at the age of 4 months (4m), and iii) at the age of 8 months (8m) of the baby. We measured V durations, F1, and F2 values, and analysed absolute V durations, durational differences, and durational ratios of V pairs, and spatial differences of the Vs in the F1×F2 plane.

Results showed that although in general mothers did lengthen V durations especially in long vowels, durational contrast was only exaggerated in IDS, if operationalized by durational differences (and not by duration ratios). Further, we found no enhancement in the spectral domain in any of the analysed pairs; and lastly, we found no effect of the baby’s age. We concluded that these results (especially the duration ratio data) provided further evidence to the claim that phonological vowel length contrast is not exaggerated in IDS, and pointed to the questions, if durational cues of length are sufficient in IDS to support distributional learning, or word segmentation skills developing around the age of 1 are needed for the babies to acquire phonological vowel length contrast.

A dajkanyelv temporális jellemzői 4 és 8 hónapos csecsemőkhöz szóló beszédben

Anna Kohári — 2019-06-01

The temporal characteristics of motherese directed to 4 and 8 month-old children

Infant-directed speech (IDS) is typically characterized by slower speech rate compared to adult-directed speech (ADS). Whether final lengthening is disproportionally exaggerated in IDS vs. ADS relative to the global speech rate still remains a controversial issue. The present study aims to investigate the temporal characteristics of Hungarian infant-directed read speech. Results show that mothers spoke slower in IDS vs. ADS. The duration of the last three (the ultimate, the penultimate, and the antepenultimate) syllables in IDS vs. ADS differed significantly by being substantially longer. However, the duration ratios of the last, the last two or last three syllables compared to the rest of the utterance were independent of directedness (IDS vs. ADS). Findings suggest that IDS is characterized by an overall lower articulation rate compared to ADS but, contrarily to some earlier findings, the final syllables did not extend disproportionally in IDS. Thus the boundary marking role of utterance-final syllables was not exaggerated by mothers addressing infants.

A hangos olvasás időviszonyainak vizsgálata különböző életkorú férfi beszélőknél

Ákos Gocsál — 2019-06-01

The temporal structure of read speech of male speakers of different ages

The purpose of the present study was to demonstrate if differences exist between Hungarian male speakers of different ages in (1) speech rate and (2) frequency and duration of pauses when their read speech is analysed. As expected, a strong correlation was found between age and the tempo values, however, the difference between the two tempo values, i.e. speech rate and articulation rate did not correlate with age. The number of pauses did not correlate with age and the durations of pauses between sentences did not vary by age either. It is concluded that the older speakers’ lower speech rate is a result of slower articulation and is not due to differences in the frequency or duration of pauses, however, it was also found that several speech parameters covered a wider range in older speakers, suggesting the existence of ’typical’ and ’atypical’ aged voices.

A spontán beszéd és a felolvasás temporális jellemzői kisiskolás korban

Judit Bóna — 2019-06-01

Temporal characteristics of spontaneous speech and reading in primary and middle school age

The aim of this study is to examine the temporal differences between spontaneous speech and reading of schoolchildren of different ages. Third, fifth and seventh-grade children participated in the study. The results show that speech tempo does not change linearly with the increase of age. In the comparison of the age groups, the difference between speech production processes (planning of spontaneous speech and becoming practiced in reading) has been observed. On the other hand, the comparison of the temporal characteristics of the two speech tasks showed how the two types of speech planning processes are related in naive readers and in experienced readers.

Nyelvtanulók megakadásjelenségei tolmácsolt szövegekben

László Bence Csárdás — 2019-06-01

Disfluencies produced by language learners during interpreting

Disfluencies are omnipresent in human speech and can be ascribed to the temporary malfunctioning of speech planning and execution. Interpreted speech is different from spontaneous speech in many respects, the most relevant of which is the dominance of the input over the macroplanning. The aim of this study was to examine the disfluencies produced by B2 language learners during interpreting. 4 subjects participated in the research. Their task was to interpret a short Hungarian text into English and another short section from English into Hungarian. Aspects of analysis were (i) the disfluencies that occurred in the speech materials, (ii) the distribution of disfluencies according to type and source and target language, (iii) the distribution of speech disfluencies according to parts of speech and (iv) the complexity of the output produced by the language learners. Based on the results we can state that (i) we use approximately the same strategies in speech planning and execution irrespective of the source and target language, (ii) both language competence and transfer competence play a vital role in interpreting and (iii) the distribution of parts of speech differs according to the source and target language.

Alaphangjellemzők vizsgálata enyhe és középsúlyos értelmi fogyatékkal élő felnőttek beszédében

Julianna Jankovics — 2019-06-01

Fundamental frequency characteristics of mild and moderate intellectual disabled adults’ speech

The purpose of this study was to analyse the fundamental frequency characteristics of mild and moderate intellectual disabled adults’ speech. Altogether 10 women (5 participants in the intellectual disabled group and 5 control adults, aged between 20 and 62 years) participated in the research. The speech samples consisted of spontaneous and read speech. The mean fundamental frequency and f 0 maximum/f 0 minimum ratio were analysed in the declarative utterances. The results were compared between the group of the intellectual disabled adults and the age- and gender-matched control group of typical speakers. The findings of this experiment were the following: 1. The mean fundamental frequency was higher in read speech than in spontaneous speech in both the intellectual disabled adults’ and the control adults’ speech. 2. The mean fundamental frequency was higher in the intellectual disabled adults’ than in the control adults’ speech, in accordance with earlier studies of the international literature.

A nyelvkontúr automatikus követése ultrahangos felvételeken

Lu Zhao — 2019-06-01

Automatic tongue contour tracking in ultrasound recordings

In recent decades, several types of devices have been constructed to track the invisible movement of the tongue. One of these is ultrasound imaging. The advantage of the measurement method is that it is suitable to track rapid movements, as well, image and sound can be synchronized, and it only causes the speaker, who is not affected by harmful radiation, minimal inconvenience. Its disadvantage is that it does not provide a full three-dimensional image but only shows a longitudinal or transverse section and the tip of the tongue cannot be seen. The longitudinal (midsagittal) section was used to track tongue movements. The automatic assessment of ultrasound pictures is a key issue as the number of frames amounts to several undred thousand so it is impossible to process them manually. The difficulty of tracking tongue contour lies in the fact that the images have a noisy background and the contour line is discontinuous. An algorithm based on dynamic programming was elaborated to track the movement of the back of the tongue. With an extreme size edge enhancing kernel and averaging construction, the process simultaneously handles the problems of break of continuity and noise. In the image obtained after smoothing, the lightest line is searched for from the left to the right edge of the image. The points of the curve thus obtained follow the uneven line of the tongue contour. To smooth the curve, screening based on discrete cosine transformation was applied.

Márkanevek és fonetikai hasonlóságon alapuló asszociációk szóasszociációs adatbázisokban

László Kovács — 2019-06-01

Brands in the mental lexicon: brand names evoked by phonetic similarities in word association databases

Brands came into the focus of research in the last three decades. Recently the center of attention shifted from brands as identifying signs to brands as psychological constructs implying that brand names, as essential parts of brands, can be considered to be entities residing in the mental lexicon. Present paper assumes that brand names, as part of the lexicon appear in
word association databases and that phonetic similarities play a role in evoking brand names as associations. The research analyzes word association databases from the 1980s and 2000s collected in Hungarian in Hungary and in Zakarpattia Oblast and investigates brand names appearing in those databases. In the research, brand names evoked by means of phonetic similarities are collected and categorized. Results show that the underlying processes evoking brand names by means of phonetic similarities are similar to those observed by Gósy–Kovács
(2001) and Lengyel (2012). Some features of associations like association chains or wrong associations are, however, easier to discover when they are connected to brand names. The final section of the paper presents managerial implications by showing how the results can be used to improve a brand’s marketing communication.

Beszédtudomány a mindennapokban ‒ előadássorozat a Magyar Tudomány Ünnepe alkalmából a székházban, 2018. november 8-án (ismertetés)

Janka Szappanos — 2019-06-01

On the workshop "Beszédtudomány a mindennapokban" ('Speech Research in Every Days') held November 8, 2018.