Voiced Sounds of the Romanian Language
 
 SRoL Sitemap


 First page

 Romanian Language Phonetics

 Sounds archive

   Vowels
   Consonants
   Diphthongs
   Sentences
      Standard prosody
      Emotional tones
      Linguistics particularities

 Dialectal voices

   Colaboration - Prof. D. Scheianu
   RoToBi System

 Protocols

 Analysis Instruments

 References

 Other links



 
If you have accidentally arrived at this page without reading and fully agreeing with the Legal Disclaimer and Copyright Notice, please click here.

The Intonational Model

The model we used to analyze the Romanian intonation adds a functional perspective to Ladd's metrical view of the sentence stress [5] and to the Autosegmental-Metrical (AM) classical perspective based on Liberman [6], Bruce [1] and Pierrehumbert's PhD theses [7]. According to this compound perspective, the intonational contour consists of a hierarchy of prosodic units (the utterance tree), each unit being defined by a tonal contrast between a high target pitch segment and a relatively low target pitch segment. These two parts have a sintagmatic relation and delimit a prosodic unit. In certain Romanian intonational contours, there are pitch events related only to the prosodic unit delimitation and others related to focalization. Using a functional perspective in intonation analysis, these two types of pitch events can be identified, helping in the nuclear accent (focus) identification at each prosodic unit level.
Furthermore, the pitch events involved in the prosodic unit delimitation must be subdivided into two subcategories: pitch events that push forward a communicative act (PUSH function) generated at the prosodic unit level, and pitch events that end the "discourse unit" corresponding to that prosodic unit (POP function) [4]. The pitch events having PUSH/POP functions may bear the focus event or not. In the former case, the focus function is associated only with the PUSH event (PUSH+FOCUS), only with the POP event (POP+FOCUS) or spans both events (as in the case of a "broad focus" statement). In the latter case, there is another pitch event related to the focalization function (FOCUS). In other words, the functional perspective of the Romanian intonation model changes Ladd's perspective of the weak (nonfocus)-strong (focus) partitioning of a prosodic unit to functional pitch event sequences: PUSH - POP ("broad focus"), PUSH - FOCUS - POP, PUSH+FOCUS - POP or PUSH - POP+FOCUS, etc. Consequently, an intonation analysis has to distinguish between the prominence of certain PUSH/POP pitch events and the prominence of the focus events.
The Ro-ToBI annotation system keeps the ToBI linear perspective on the pitch events at each prosodic unit level and tries to improve the labelling to identify easier the pitch events involved in the prosodic unit delimitation and focalization. The functions of the pitch events are not explicitly annotated, but they can be deduced by searching the sequence resulting from a prosodic unit level, within a list of typical sequences involved in focus generation.
The pitch events can be either single or compound. A compound event is a pitch event sequence of a lower level phrase. At the same time, the pitch events of the sequence can be viewed as an event having a single function within the pitch contour partitioning at the parent phrase level. These two views of compound pitch events make possible to combine the linear perspective of the ToBI annotation system with the hierarchical structure of an intonational phrase.
At the intonation description level, the Ro-ToBI annotation system labels only the single pitch events. Their grouping into compound pitch events results from the annotation on the break indices tier to keep the ToBI system framework.
Concerning the single pitch events, it can be noticed that they usually have significant pitch movements. However, we have also taken into account pitch events without significant pitch movements, whose patterns vary around a certain tonal level. They can be involved in tonal contrast generation of the PUSH/POP events and/or in focus generation. When they are connected to the highest or to the lowest tonal level of the tonal space, they can generate emphases. We have particularly marked the prosodic word that has the same level for the beginning and ending tones.
By introducing these annotations for the pitch patterns at the prosodic word level, the Ro-ToBI annotation system brings more phonetic information to the intonation descriptions, compared to those generated by the classical ToBI system. Furthermore, the annotation of the metrically invisible pitch events is necessary in generating the hierarchical structure of the sentence stress in Romanian intonation.

The prosodic domain hierarchy

The hierarchy of the prosodic domains of the model used to describe the Romanian intonation consists in the following prosodic unit domains, presented from the top to the bottom in Fig. 1 a: the intonational phrase (IP); the intermediate phrase or the phonological phrase (both belonging to the domain called Accentual Unit Group - AUG); the prosodic word (called accentual unit - AU). The prosodic words are elementary prosodic units containing a single pitch accent. The intermediate and the phonological phrases are regarded as nonelementary units having more than one pitch event. These are realizations of the same domain type, called Accentual Unit Group. In the most general case, this domain type can have a recursive structure, allowing constituents of the same type (AUG), in addition to the AUs (Fig. 1 b-d).

picture 1a
a
picture 1b
b
picture 1c
c
picture 1d
d
Figure 1

For example, the neutral utterance of the text Lui Winston / (i se treziră / nişte/( vagi / amintiri)) can be described by the units corresponding to the text units marked by "/" and parantheses. In this case, the hierarchy of the utterance tree is depicted in Fig. 2. It contains five AU type elementary units (AU1-AU5) and two AUG type nonelementary units (AUG1 and AUG2).

picture 1a
AU1- AU1- Lui Winston,

AU2- i se treziră,

AU3- nişte,

AU4- vagi

AU5- amintiri
Figure 2

   
Această pagină a fost modificată ultima oară de V. Apopei, D. Jitcă şi O. Păduraru la data de
Copyright©2006 - SRoL