Systematic Acoustic Variation in Spontaneous Filled Pauses

18 concepts7 papers

Filled pauses like "uh," "um," "ano," or "eto" are often dismissed as mere "disfluencies"—accidental byproducts of a struggling brain. However, a growing body of research suggests these vocalizations are far from random. Instead, they possess systematic acoustic structures that vary based on speaker identity, social context, and communicative intent.

The Voice Quality of Hesitation

If filled pauses were simply "stalling for time," we might expect them to be acoustically identical to the vowels found in ordinary words. However, Maekawa et al. demonstrated that this is not the case. By analyzing Japanese spontaneous speech, the authors found that the vowels in filled pauses exhibit distinct "voice qualities"—subtle variations in intensity, fundamental frequency (F0), jitter, and spectral tilt—that differ significantly from vowels in lexical items.

These differences suggest that filled pauses are not acoustically underspecified. Rather, speakers may leverage specific vocal variations to signal pragmatic information, such as the length of the intended delay or the speaker's desire to hold the floor.

Sociolinguistic and Contextual Shaping

The way a filler sounds is not just a matter of individual biology; it is shaped by a complex interplay of sociolinguistic factors. Vasilescu et al. found that while the "timbre" (spectral characteristics) of a filler is primarily determined by the speaker's native language, other features like duration are heavily influenced by speaking style and gender. Interestingly, the authors even suggested that a speaker’s language proficiency could be estimated from the timbre of their hesitations.

Contextual factors also dictate the choice of filler. In Japanese presentations, Watanabe et al. observed that factors such as speech formality, speaker age, and gender significantly influence whether a speaker chooses "ano" versus "eto." This indicates that fillers serve a functional role in managing the social level of an interaction, not just the cognitive load of the speaker.

Prosody and Position

The "melody" of a filled pause—its pitch, loudness, and duration—is also highly dependent on its location within a conversation. Nakamura et al. found that the prosodic features of Japanese fillers are not uniform; they change based on where the filler occurs in a dialogue. A filler at the start of a turn may sound quite different from one used mid-sentence.

This positional variation is crucial for naturalness. When these nuances are ignored, as is often the case in current AI systems, human-computer interaction can feel robotic. The systematicity of these markers is so ingrained in human dialogue that even in specialized contexts like simultaneous interpretation, the rhythm and stability of pauses (including voiced fillers) significantly impact how listeners perceive the quality of the translation, as noted by Tohyama et al..

Speaker-Oriented vs. Hearer-Oriented Functions

A central debate in this field is whether filled pauses are "speaker-oriented" (a result of internal cognitive processing) or "hearer-oriented" (a tool used to signal meaning to others).

Hutin et al. investigated this by looking for "conversational convergence"—the tendency for speakers to mimic each other's speech patterns. They found that filled pauses are highly idiosyncratic and do not converge between partners. This suggests that while fillers have systematic acoustic properties, they may be primarily driven by the speaker's own production processes rather than an intentional effort to align with a listener.

However, the presence of distinct prosodic markers in other forms of address—such as the specific prosodic signatures Niebuhr et al. found in spoken prayer—reminds us that humans consistently use subtle acoustic cues to frame their relationship with their interlocutor, whether that interlocutor is a person, a machine, or a divine entity. Filled pauses, through their systematic variation in voice quality and prosody, are a vital part of this communicative toolkit.

Key Foundational Work

Voice-Quality Difference Between the Vowels in Filled Pauses and Ordinary Lexical Items (2016) — contributed 6 concept elements to this direction.
Language, gender, speaking style and language proficiency as factors influencing the autonomous vocalic filler production in spontaneous speech (2006) — contributed 4 concept elements to this direction.
Factors affecting speakers² choice of fillers in Japanese presentations (2006) — contributed 4 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_vocalic_voice_quality_differences_539a9fbe,
  title = {vocalic voice quality differences},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/vocalic-voice-quality-differences/rdr_539a9fbe582d499db6f2c3e4e81fcb3f},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_vocalic_voice_quality_differences_539a9fbe_v1,
  title = {vocalic voice quality differences},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/vocalic-voice-quality-differences/rdr_539a9fbe582d499db6f2c3e4e81fcb3f/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}

See something wrong? Report a mistake