Effects of semantic context and type and amount of acoustic distortion on lexical decision in younger and older adults
Huiwen Goy, Marianne Pelletier, Marco Coletta, M. Kathleen Pichora-Fuller
When we hold a conversation, our ability to perceive speech is frequently affected by environmental noise and distortions of the speech signal. One way of filling in missing information is to rely on context, but even so, speech comprehension may be effortful in the presence of communication barriers. The goal of this study was to measure listening effort in younger and older adults when they listened to speech that was distorted in ways that resembled common problems such as noise and hearing loss. In this study, listeners heard a variety of sentences and were asked to decide as quickly as possible whether the final word in the sentence was a real word or a non-word. Some of the sentences contained contextual information that helped listeners in making lexical decisions, and other sentences did not have helpful contexts. The sentences were distorted in different ways: in one condition, sentences were presented with babble noise (mimicking listening to one speaker in a large crowd); in a second condition, sentences were low-pass filtered (mimicking high-frequency hearing loss), and in a third condition, sentences were time-compressed (leading to a fast rate of speech). The amount of time it took for people to make lexical decisions at the end of the sentence was used as an index of the effort expended in comprehending the sentence. Older adults’ reaction times improved more than younger adults’ reaction times when they heard helpful sentence contexts, in line with other findings that older adults make more use out of context than younger adults. Younger adults were affected by low-pass filtering and time compression in similar ways, but older adults were more affected by time compression than by low-pass filtering, which may reflect age-related declines in auditory temporal processing or a generalized slowing in cognitive processing.
Back to Speech Perception and Identification
Scene Analysis in Aging: The Effect of Noise Onset and Aging on Speech Comprehension
Professor Bruce Schneider and Dr. Boaz Ben-David,
Centre for Research on Biological Communication Systems, University of Toronto Mississauga.
Research, the Toronto Rehabilitation Institute.
How can you differentiate between the words spoken to you and the noise around you? How long does it take you to get acquainted with the characteristics of the noise around you? Will it change in aging? This study is designed, in part, to answer these questions.
For example, imagine Mrs. Levin, a 67 year old grandma entering a very noisy restaurant. She is overwhelmed by the noise of babble (many people talking) around her, and the host is asking her a question. How loud should he speak for Mrs. Levin to understand him? Should he wait a second, let her get used to the noise around her before he asks a question? Would your answers change if Mrs. Levine's 18 year old grand-daughter, Hannah, enters the same restaurant? In this study we try to answer these specific questions.
In the experiment you will be asked to listen to words spoken in noise and repeat what you have heard. Words will be presented in two types of noise: a speech spectrum noise, or in babble. These types of noise are artificially manufactured to resemble a situation where many people are talking simultaneously. In each block, we will present the noise differently. In some blocks the noise will start before the word, in other blocks, it will start simultaneously with the word, and in others it will start after the beginning of the word. We will test whether this will affect your chances to correctly identify the word.
Back to Speech Perception and Identification
The effects of phonological and semantic similarity on speech perception in noisy scenarios.
Cristina Rabaglia, Bruce Schneider, and Meredyth Daneman; University of Toronto Mississauga
Abstract: Older adults experience many language-related changes with increasing age. Important among these is difficulty with understanding spoken language. Older adults find comprehending speech particularly challenging in noisy environments where competing sounds – other conversations occurring in the background, a radio playing, or street noise, for example – are also present. However, exactly why background noise is so detrimental to spoken language comprehension remains unclear.
When listening to spoken language in a noisy environment, previous research has demonstrated that the type of background noise seems to matter. Sounds with acoustic properties similar to speech are more harmful than sounds very different from human speech, and a background conversation occurring in a familiar language is more detrimental than one occurring in an unfamiliar language. However, if the target speech (the attended conversation or radio program, for example) and distracting speech (occurring in the background) are both in the same language, these two speech streams have many potential areas of overlap. Acoustic, phonological, or semantic similarities between same-language target and background speech all might potentially contribute to confusability between the two streams. It is therefore still unknown which specific aspects of background noise are the most detrimental to target speech comprehension. Understanding which of these potential areas of target-distractor overlap are the most problematic for speech comprehension is especially important in the context of age-related difficulties, since older adults typically experience age-related changes at both sensory and cognitive levels. As a result, older adults could experience speech comprehension difficulty stemming from any of these factors, and that difficulty could potentially be different in nature than that experienced by younger adults.
This study will investigate the degree to which acoustic, phonological, and semantic characteristics of competing speech influence target speech comprehension. Participants in this study will repeat “target” nonsense sentences played in a sound attenuated booth. Each target sentence will be played concurrently with an associated “masker” in one of four conditions:
- Speech spectrum noise (no linguistic content)
- A phonologically related sentence (target-overlapping phonological content)
- A semantically related sentence (target-overlapping semantic content)
- A semantically and phonologically unrelated sentence (same-language content that does not directly semantically or phonologically overlap with the masker)
By comparing performance across two age groups (younger and older adults), we can also test how these different levels of overlap between target and competing speech may differentially influence target speech comprehension in aging. The results of this study will thus not only provide clues as to why background noise is so detrimental to speech comprehension in general, but also assist in understanding how best to alleviate spoken language difficulties in aging populations.
Back to Speech Perception and Identification
Age related changes in how the brain processes two auditory sources
Many of our conversations in day-to-day life occur in environments with other noise in the background– traffic noise on a city street, a radio playing in the background during a car ride, or other conversations in a crowded restaurant. To successfully understand and engage in such conversations, we must somehow isolate the voice of the person we are speaking with from this noise generated elsewhere in the environment. This aspect of communication appears to be a particularly difficult one, and apparently grows even more so as we age; we know from previous research that even people with hearing deficits are able to attend to speech relatively well in quiet environments, but we are still trying to determine what factors are responsible for making speech so difficult to understand in noise.
The current project is based on recent evidence suggesting that the amount of time required to segregate two auditory sources (like an attended conversation and background noise) changes with age. To investigate why this might be the case, this study looks at the length of exposure to noise prior to onset of the ‘target’ talker (the person the listener is attempting to understand). Participants listen to lists of words in different types of noise, while electrical activity in their brain is recorded using a non-invasive cap that measures event-related potentials (ERP). Using ERP to explore the processing of spoken words in background noise can help reveal what is happening in the brain as we attempt to separate two sources of auditory information. Results of this study can thus point us to the neural markers (brain regions and electrical signatures) that characterize successful word recognition under these challenging conditions, and help us understand what may be going wrong at a neural level when background noise levels cause us to fail at understanding spoken language.
Back to Speech Perception and Identification
Can the diffuseness of a sound source effect how well we hear in noise?
Avivi-Reich, M., Schneider, B.A.
We investigated how the perceived compactness of sound sources affects speech recognition (Compact sound images are perceived to have a precise location in space. Diffuse sound images are perceived as filling auditory space). This investigation was prompted by the observation that modern amplification systems, such as surround sound, routinely alter the degree to which a sound source appears to be diffuse or compact. Younger and older native-English listeners and young nonnative-English listeners were asked to repeat short syntactically-correct-but-semantically-anomalous sentences played in different types of background noise, when there was either a contrast (as there often is in natural settings) between the diffuseness of the target and masker sound sources (compact target versus diffused masker), or not (compact target and masker). Results showed a significant improvement in performance when there was a contrast of diffuseness in each of the three groups. The release from masking was significantly greater when the masker was speech, compared with when it was static noise.
Back to Speech, Perception and Identification
Could age related declines in sensory processing alter the nature of multisensory integration in the presence of energetic and informational masking?
Avivi-Reich, M., Puka, K. & Schneider, B.A.
Age-related deterioration in one’s ability to comprehend speech in noise plays a primary role in the difficulties many older adults experience when communicating. When the talker is visible, older adults may greatly benefit from visual cues accompanying the acoustic input. The additional visual information provided by the face of the talker requires efficient integration of audio-visual information if this visual information is to enhance speech understanding. Age related declines in sensory processing could alter the nature of multisensory integration (either reducing or enhancing it). In addition, the benefit provided by visual cues may depend on the degree of masking created by competing sound sources, which, in turn, may be affected by aging. In order to address the possible effects of age and type of masker on auditory visual speech perception, 24 younger and 24 older adults with normal hearing and vision were asked to repeat semantically anomalous sentences (e.g., “A rose could paint a fish”) masked by either static noise, 12-talker babble or two other talkers in two conditions: when the target sentences were accompanied by videos presenting the talker; and when the target sentences were played over loudspeakers only. Each masker type and condition combination was played different intensity levels. The results demonstrate a substantial benefit from visual cues in both younger and older adults. Visual cues provide more benefit when the masker consists other voices. In the presence of visual cues, older adults’ speech perception is very similar to that of the younger adults.
Back to Speech, Perception and Identification