Perceptual learning of /f/-/s/ by older listeners

Odette Scharenborg, Esther Janse, Andrea WeberInterspeech 2012

TL;DR

Older adults can learn to distinguish between similar speech sounds (/f/ and /s/) even when the acoustic cues are primarily in higher frequencies, which are typically affected by age-related hearing loss. However, the extent and stability of this learning are negatively impacted by the degree of hearing loss.

What Problem Does This Paper Address?

As we age, our hearing, particularly in the higher frequency ranges, often deteriorates. This raises a crucial question: can older adults still adapt to subtle variations in speech sounds, similar to how younger listeners do? Specifically, this paper investigates whether older listeners can learn to distinguish between the fricative sounds /f/ and /s/, which rely heavily on high-frequency acoustic information. Previous research has shown that young listeners can quickly adjust their perception of ambiguous speech sounds based on the context of words they hear. This study aims to determine if this "perceptual learning" ability is preserved in older adults, and if age-related hearing loss plays a significant role in this process.

Methods

The researchers employed a lexically-guided perceptual learning paradigm. This involved two main phases:

  1. Exposure Phase (Lexical Decision Task): Participants listened to Dutch words. Some words ended in a natural /f/ or /s/, while others contained an ambiguous sound that was a blend of /f/ and /s/. Participants were divided into two groups: one group heard the ambiguous sound primarily in words that would naturally end in /f/, and the other group heard it in words that would naturally end in /s/. Their task was to decide if each sound was a real word or a non-word. This phase aimed to "bias" their perception of the ambiguous sound.
  2. Test Phase (Phonetic Categorization Task): Participants were presented with a continuum of sounds ranging from clear /f/ to clear /s/, embedded in words. They had to categorize each sound as either /f/ or /s/. This phase measured how their perception had shifted after the exposure phase.

The researchers also assessed participants' hearing sensitivity using standard audiometry. They then analyzed how hearing loss correlated with the degree and stability of perceptual learning.

Datasets

Not discussed in the paper. The study used custom-recorded Dutch words and synthesized ambiguous speech sounds.

Key Insights and Ideas

  • Age is not a complete barrier to perceptual learning: Older adults can indeed learn to distinguish between /f/ and /s/, even though these sounds rely on high-frequency cues that are often compromised by aging. This suggests that other acoustic cues, like formant transitions, might be utilized by older listeners.
  • Hearing loss is a significant modulator: The degree of high-frequency hearing loss directly impacts the effectiveness and persistence of perceptual learning. This is a critical finding, highlighting the functional consequences of age-related auditory decline.
  • Learning is not always permanent: The perceptual learning effect, while present, was not stable over time. It diminished in later test blocks, suggesting that the adaptation might be temporary, especially for those with poorer hearing.
  • "Unlearning" is more pronounced with hearing loss: Listeners with more significant hearing loss showed a greater tendency to revert to their original perceptual categories, indicating a less robust and less stable learning effect.

Key Results

  • Older listeners successfully demonstrated lexically-guided perceptual learning for the /f/-/s/ contrast. Listeners exposed to the ambiguous sound in /f/-final words were more likely to perceive it as /f/, and vice-versa for /s/-final words.
  • The perceptual learning effect was still observable in the third test block but had disappeared by the fourth block, indicating a limited duration of the learned adaptation.
  • Hearing sensitivity significantly predicted perceptual learning. Listeners with poorer hearing exhibited fewer "learning-consistent" responses and showed greater "unlearning" over time.
  • While hearing loss did not affect the accuracy of lexical decisions, it did slow down response times, suggesting a general cognitive load associated with impaired hearing.

Limitations and Future Work

  • Limited duration of learning: The study found that the perceptual learning effect was not sustained throughout all test blocks. Future research could explore methods to enhance the stability of this learning.
  • Impact of hearing loss: While the study shows hearing loss negatively impacts learning, further investigation into how hearing loss interferes with the use of different acoustic cues (e.g., spectral information vs. formant transitions) would be valuable.
  • Generalizability: The study focused on the /f/-/s/ contrast. Investigating other speech sound contrasts, particularly those with different acoustic properties, would provide a more comprehensive understanding of perceptual learning in older adults.
  • Individual differences: While hearing loss was a key factor, other individual differences (e.g., cognitive abilities, prior experience) might also influence perceptual learning and warrant further exploration.

Summary updated

BibTeX
@inproceedings{scharenborg2012perceptual,
  author = {Odette Scharenborg and Esther Janse and Andrea Weber},
  title = {{Perceptual learning of /f/-/s/ by older listeners}},
  booktitle = {Interspeech},
  year = {2012},
}
Figure