Focus-sensitive operator or focus inducer: always and only

Yong-cheol Lee, Satoshi NambuInterspeech 2010

TL;DR

This paper investigates whether focus particles like "always" and "only" induce focus in a sentence or act as operators that are sensitive to existing focus. By analyzing prosodic features (pitch, duration, intensity) in Korean, they found that "always" and "only" are focus-sensitive operators, as the focused word itself isn't prosodically altered by their presence. Interestingly, "always" exhibits more distinct prosodic characteristics than "only," suggesting it carries its own focus feature.

What Problem Does This Paper Address?

There's a long-standing debate in linguistics about how focus particles (FPs) function in language. One view, the "focus inducer" hypothesis, suggests FPs directly assign focus to a following part of the sentence. The alternative, the "focus-sensitive operator" hypothesis, posits that FPs are sensitive to focus that is already present, often determined by discourse context. While both perspectives have arguments, it hasn't been definitively established which is correct, especially concerning the prosodic realization of these particles. This paper aims to resolve this by examining the prosodic differences between two common Korean FPs, "always" (hangsang) and "only" (ocik), to see if they behave as focus inducers or operators and if they have distinct prosodic profiles.

Methods

The researchers designed an experiment using four conditions to elicit focus: context-free, prompt question, discourse context, and a control condition without any FP. They recorded Korean speakers reading sentences with these conditions. The core of their methodology involved prosodic analysis, specifically extracting and measuring Fundamental Frequency (F0), duration, and intensity of the speech signal. They used a Praat script to automatically extract these acoustic parameters after manually marking syllable boundaries. Statistical analysis, including one-way and two-way ANOVAs, was employed to determine the significance of differences in these prosodic measures across the different FP conditions and contexts.

Datasets

The study used a custom dataset of 216 Korean sentences, with 126 serving as target sentences. These sentences were constructed to include the focus particles "always" (hangsang) and "only" (ocik) in various contexts (context-free, prompt question, discourse context) and a control condition without an FP. The stimuli were presented to participants on paper in a randomized order. No pre-existing, publicly available datasets were used; the data was specifically created for this experiment.

Key Insights and Ideas

  • FPs as Operators, Not Inducers: The most significant finding is that FPs like "always" and "only" appear to be focus-sensitive operators. The prosody of the focused word itself remained largely unaffected by the presence or absence of these FPs, suggesting the focus was already established by the context or question, and the FP merely operated on it.
  • Prosodic Asymmetry Between "Always" and "Only": While both FPs function similarly as operators, they exhibit distinct prosodic patterns. "Always" showed significantly higher mean and maximal F0, longer duration, and greater intensity compared to "only."
  • "Always" Carries Its Own Focus: The pronounced prosodic features of "always" suggest it might carry its own focus feature, independent of the sentence's main focus. This implies that a single sentence might convey multiple communicative functions simultaneously.
  • PENTA Model Support: The findings align with the PENTA (Parallel Encoding and Target Approximation) model, which proposes that speech melody can convey multiple intonational functions concurrently, based on function rather than just form.

Key Results

  • No Prosodic Difference in Focused Word: Statistical analysis (one-way ANOVA) showed no significant differences in duration, mean intensity, mean F0, or Max F0 for the focused word across the three FP conditions (No FP, Always, Only). This supports the operator hypothesis.
  • Significant Differences Between "Always" and "Only": Two-way ANOVAs revealed significant differences in duration, mean intensity, mean F0, and Max F0 between the FPs "always" and "only." Specifically, "always" consistently showed higher values for all these prosodic measures compared to "only." For instance, duration for "always" was 357.9 ms vs. 292.2 ms for "only."
  • Context Independence for FP Differences: Crucially, there were no significant interactions between the FP type ("always" vs. "only") and the discourse context. This means the prosodic differences between "always" and "only" were consistent across different contexts.

Limitations and Future Work

  • Limited FP Set: The study focused on only two Korean FPs ("always" and "only"). It would be beneficial to expand this to include a wider range of FPs to see if these findings generalize.
  • Cross-Linguistic Comparison: The authors suggest future work should investigate cross-linguistic typological patterns to understand the nature of FPs more broadly. This includes examining whether FPs universally function as operators/inducers and whether phonetic differences between FPs are common across languages.
  • Prosodic Functionality: While "always" shows distinct prosody, its precise communicative function beyond carrying its own focus needs further exploration. The paper hints at the possibility of simultaneous intonational functions, which is a rich area for future research.

Summary updated

BibTeX
@inproceedings{lee2010focus,
  author = {Yong-cheol Lee and Satoshi Nambu},
  title = {{Focus-sensitive operator or focus inducer: always and only}},
  booktitle = {Interspeech},
  year = {2010},
}
Figure