1887

n Tydskrif vir Geesteswetenskappe - Oor die status van die "oe" in Afrikaans : 'n akoestiese analise s

Volume 50, Issue 1
  • ISSN : 0041-4751

Abstract

Die hoofvraag wat in hierdie eksperimentele ondersoek aan die orde is, is of die Afrikaanse /u/- vokaal (geskryf "oe") inderdaad 'n hoë agtervokaal (d.i. kardinale vokaal Nommer 8 op die artikulatoriese vokaalkaart) is soos wat dit tradisioneel, en ook in 'n enkele onlangse akoestiese beskrywing van Van der Merwe, Groenewald, Van Aardt, Tesner en Grimbeek (1993) beskryf word. Eie waarneming van 'n sterk ontronding daarvan word hier opgevolg deur uitgebreide akoestiese ondersoeke, aangevul deur persepsietoetse. Die spraak van 19 Afrikaanssprekende jongmense van verskillende dele van die land is ontleed en die resultate is statisties verwerk. Dit blyk dat die tweede vokaalformant, F2, wat primêr 'n aanduider is van die mate van ronding van agtervokale soos /u/, oorweldigend wys op 'n sterk skuif weg van die buiterand van die artikulatoriese vokaalruimte na binne. Die feit dat dit egter nie ook na onder skuif nie (soos blyk uit die stabiele eersteformantwaardes in byna alle metings), dus steeds 'n hoë vokaal bly, en nie /@/-agtig word nie, kan geneem word as argument teen die moontlikheid dat dit hier om gewone akoestiese reduksie gaan.


Die genoemde akoestiese resultate word gevind in byna alle omstandighede wat ondersoek is, dit wil sê eksterne en interne kontekste. Dit geld nie net geslag van die deelnemers aan die eksperiment nie (vrouens ontrond beduidend meer as mans), maar ook die mate waarin die vokaal benadruk word (sekondêr benadrukte woorde is meer sensitief vir ontronding), die mate van frekwensie van gebruik van die individuele woorde (ontronding word meer aangetref in woorde wat baie frekwent gebruik word), die tipe stimuli (vokale ontrond meer in sinne), sillabestruktuur (oop sillabes is meer gevoelig vir ontronding). Leestempo is meer ingewikkeld, en toon nie op 'n eenvoudige wyse sodanige verskille nie. Betreffende plek van vorming van die konsonant volgende op die vokaal is die situasie nog meer ingewikkeld. Dit word wel gevind dat hierdie faktor 'n invloed kan hê, byvoorbeeld dat slot-/l/ soms 'n teenwerkende invloed in sommige woorde het op ontronding. Die hoofklasse konsonante in artikulatoriese terme, labiaal, alveolêr en velêr, verskil ook slegs in bepaalde gevalle van mekaar wat betref die uitwerking wat hulle het op die voorafgaande vokaal.
Verskeie persepsietoetse toon aan dat Afrikaansluisteraars dikwels groot moeite het met die interpretering van woorde waarvan die vokaal bedoel wat as /u/; dit word in sulke gevalle in gedwonge luistertoetse geïnterpreteer as /i/, wat sterk ondersteunend is vir die resultate van die akoestiese metinge in hierdie ondersoek as sou /u/ nie (meer) 'n (hoë) agtervokaal wees nie. Hoewel moeilik om presies te karakteriseer, lyk dit wel asof dit kan of moet deurgaan as 'n hoë, ontronde agtervokaal. Tyd sal leer of Afrikaanse /u/ permanent van kwaliteit tot die hoë, ongeronde vokaal /U/ sal verander.
Daar is vlugtig aandag gegee aan die implikasies wat hierdie studie het vir spraaktegnologiese toepassings, asook in algemeen-taalwetenskaplike opsig vir die aansprake van die dispersieteorie.


The main aim of this experimental investigation was to establish the precise nature of the Afrikaans vowel /u/. Traditionally it was characterised as a short, high back vowel, situated in the uppermost right corner of the articulatory vowel chart, known as cardinal vowel number 8. Van der Merwe, Groenewald, Van Aardt, Tesner and Grimbeek (1993), the only detailed study of this kind up to date, reported acoustical measurements to be much in line with the following: F1 = 266 Hz; F2 = 961 Hz. For some versions of American English similar measurements are known, for example in the monumental study of Peterson and Barney (1952), but more recent studies show a tendency of fronting (Hagiwara 1997). Bekker (2009) describes the same for the speech of some young, South African female speakers. In all the above-mentioned studies, F2 values are indicative of such fronting process, while F1 seems to be relatively unaffected.
Careful listening to the pronunciation of this vowel by a wide variety of Afrikaans speaking persons has led to this experiment. Such speakers include those of both genders, students from a variety of geographical origin, participants to radio and television broadcasts and the like. Fronting of /u/ is seemingly present in many cases, justifying this study.
The present acoustical production experiment comprised the collection and analysis of the speech of nineteen Afrikaans persons (six of them male). They read two sets of stimuli (a word list and a number of sentences) two times each, once at a normal tempo, once at a fast rate. The word list was made up of 25 words, while there were ten natural sentences in the other set. All of these contained words including the /u/-vowel. Vowels were segmented and annotated in the Praat program, and consequently all relevant acoustic parameters, of which the most important being the first two vowel formants, were extracted and imported into the statistics program Statistica. Following a process of normalisation of the measurements, analyses of variance (ANOVAs) were performed with respect to a number of linguistic and extra-linguistic factors, for example emphasis (stress and sentence accent), syllable structure, place of articulation of consonants following the focus vowel, stimulus material, reading tempo, and frequency of use of words.
Results clearly show a salient degree of /u/ fronting in the production of all participants, especially those of female speakers. Support in favour of the characterisation of this vowel as a fronted high vowel comes from a variety of specific acoustic measurements, but also from perception evidence. As to the former aspect, it is evident from the dominant F2 movement away from the outer borders of the acoustic (and, consequently, articulation) vowel space in the direction of the centre thereof. It should, however, be noted that this inner movement is not accompanied by a simultaneous downward movement (indicated by F1 readings) of similar extent. Such movement would indicate a process of normal neutralisation instead. With respect to perception, participants in the listening tests frequently interpreted /u/ as /i/ (e.g. sien instead of soen), but never as schwa /@/ (as in sin).
As to the independent variables, the results show a robust fronting of the /u/ vowel in almost all of those involved in this study. Generally, vowels in words with a high frequency tend to front significantly more than those in words with a low frequency, as does secondary emphasised vowels in contrast to primary emphasised ones; vowels in sentences rather than in the list of words displayed the same, as did vowels in open syllables. Only reading tempo did not show this tendency. No difference was found as to degree of fronting between slow versus fast reading - both did, nonetheless, exhibit fronting on a significant scale. Interestingly, place of articulation turned out to be quite complicated. The main classes of consonants following the focus vowel /u/ demonstrated a diverse pattern as to the degree of fronting influence. As to /l/, fronting did not occur in some instances, but in others fronting surpassed that in words with following non-/l/ consonants. This stands in contrast to some South African English (reported by Bekker 2009). Per implication he claims fronting /l/-effect on preceding /u/ to be a steady feature.
On the grounds of the results of the experiments reported here, it seems as if Afrikaans is on its way to a change in vowel quality of the high, back front vowel /u/ to its unrounded fronted counterpart /U/. Initial observations suggest a similar trend of fronting of the Cardinal Vowel 7 /o/ (as in groot), which is the other (mid-) high, back, rounded Afrikaans vowel.
Implications are touched upon for speech technological applications as well as for the dispersion theory of vowel distribution in vowel space.

Loading full text...

Full text loading...

Loading

Article metrics loading...

/content/akgees/50/1/EJC20181
2010-03-01
2019-10-20

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error