Abstract
This study investigates the multi-modal cues—acoustic, structural, and informational—that signal prominence in Russian and Hindi, two languages characterized by flexible word order. Through behavioral experiments involving silent reading and auditory tasks, the research demonstrates that while acoustic cues like pitch and intensity are significant, the "newness" of information and specific syntactic positions are primary drivers of perceived prominence across both languages.
Introduction
In linguistic discourse, "prominence" identifies words that stand out to a listener or reader. While English predominantly utilizes pitch accents for this purpose, "free word order" languages like Russian and Hindi can also leverage syntactic movement. This paper explores whether prominence is driven by acoustic "punch," structural positioning, or the novelty of information, comparing Russian (SVO) and Hindi (SOV) to identify potential cross-linguistic universals.
Methodology
The researchers employed a multi-pronged approach to analyze prominence:
- Acoustic Analysis: Extraction of fundamental frequency (f0), intensity, and duration from recorded narratives.
- Information Structure (IS) Coding: Content words were categorized using the RefLex framework at referential and lexical levels to account for "givenness."
- Prominence Rating Task (PRT): Naïve participants marked words they perceived as prominent during silent reading (Experiment 1) and auditory listening (Experiment 2).
- Statistical Modeling: Generalized linear models predicted prominence scores (p-scores) based on acoustic, positional, and IS factors.
Datasets
The study utilized two primary corpora:
- Russian Corpus: 39 clause-size excerpts from narratives read by a native female speaker.
- Hindi Corpus: Oral narratives from the OGI Multi-language Telephone Speech Corpus, featuring 16 speakers and approximately 592 content words.
Results and Discussion
The Primacy of Information Structure
A consistent finding across both languages and modalities was the "newness" factor. Referents mentioned for the first time (r-new) or those that are discourse-new but generally known (r-unused) were the strongest predictors of prominence. This suggests a cognitive bias toward flagging novel information as discourse-salient.
Structural "Hotspots"
Structural position significantly influenced prominence even in the absence of audio. In silent reading, readers consistently rated words in non-canonical positions (e.g., fronted or post-posed) as more prominent. Specifically:
- In Hindi, the pre-verbal position serves as a major structural cue for prominence ($z=4.89, p<.001$).
- In Russian, post-posing a word relative to its default position significantly boosts its perceived importance.
Acoustic Universals
Despite differences in prosodic systems, increased intensity and f0 maxima reliably signaled prominence in both languages. In Russian, a half-octave expansion in pitch range or a 20 dB increase in intensity correlated with a 7-10% rise in auditory prominence scores. Interestingly, acoustic cues often overshadowed structural ones in auditory tasks, particularly in Hindi, where the effect of "ex-situ" word order was primarily detected in the silent reading task.
Conclusion
The correlates of prominence are remarkably consistent across Russian and Hindi. While acoustic cues provide immediate signals in speech, the underlying information structure—specifically the introduction of new referents—and strategic syntactic positioning remain fundamental to how humans process and assign importance to words in discourse. Future research should explore the interaction between these cues and their functional specificity in signaling different types of focus.
Summary updated