Sounds Processing, Phonetic Drift and New Research

Sound Processing, Phonetic Drift and Other New Research |
11/19/2009
Advance for Speech-Language Pathologists & Audiologists - Online

International experts in acoustics reported on new research at the 158th meeting of the Acoustical Society of America (ASA) Oct. 26-30 in San Antonio, TX. They represented scientific disciplines as diverse as medicine, music, psychology, engineering, speech communication and marine biology.

How does a woman's voice differ from a man's? Do babies who briefly hear a foreign language start to babble in that language? These are just a few of the questions that were discussed during more than 650 talks and poster sessions. Presentations also addressed such topics as teacher voice use in the classroom, how to make a high school gymnasium less noisy, listening levels for portable digital music devices, and cochlear dead regions in people with hearing impairment.

Ingo Titze, PhD, and Eric Hunter, PhD, executive director and deputy executive director, respectively, of the National Center for Voice and Speech (NCVS) at the University of Utah, presented "Variations in Intensity, Fundamental Frequency, and Voicing for Teachers in Occupational Versus Non-Occupational Settings." Their study created a more concise picture of the vocal demands placed on teachers by comparing occupational and non-occupational voice use.

Teachers tend to spend more time speaking than most professionals, putting them at a greater risk for hurting their voices. They are 32 times more likely to experience voice problems, research has shown. Unlike singers and actors, they can't take a day off when their voice hurts. The NCVS study revealed how teachers use their voices at work and home and the differences between male and female teachers.

The researchers equipped teachers with the NCVS voice dosimeter, a device that captures voicing characteristics such as pitch and loudness rather than actual speech. The dosimeter sampled voices at 33 times per second. The researchers analyzed 20 million of these samplings that were collected over a 14-day period.

Female teachers used their voice about 10 percent more than males when teaching and 7 percent more when not teaching, the study showed. The data also indicated that female teachers speak louder than male teachers at work. "These results may indicate an underlying reason for female teachers' increased voice problems," stated Dr. Hunter.

All of the teachers spoke about 50 percent more when at work, at both a higher pitch and volume. Instead of resting their overworked voices at home, the teachers spent significant amounts of time speaking outside of work.

A team of student researchers led by Stephanie Hoeman from the University of Kansas presented on "Acoustics in a High School Gymnasium." The architectural students studied a gym constructed entirely of concave shapes, which resulted in severe sound focusing. The space was extremely reverberant and exhibited many distinct, distracting reflections.

The gym constructed with unintentionally interesting acoustics was scheduled for a new sound reinforcement system. However, the system would be of little consequence until the architectural acoustics were addressed. The student researchers made impulse responses in the space and computer models in an effort to reduce unwanted reflections and shorten reverberation time. Their solution had to be economical and uncomplicated to install because the project was for a public school with limited funds.

The researchers compared an acoustical analysis and auralizations of the space in its current condition and the projected performance of recommendations to find an effective, cost-efficient solution that the school board could implement to improve the architectural acoustics before the sound reinforcement system was added.

The recent development and popular use of personally worn digital music devices (PDMDs) have led to concerns about the intensity levels they produce. One concern is that users will set their device to levels known to be hazardous to hearing. This concern stems from the private nature of a PDMD, in that insert earphones are worn that restrict the output to the user's ears, thus prohibiting monitoring by parents, teachers and employers.

"In-Situ Measures of User's Preferred Listening Levels with a Portable Digital Music Device" was presented by Edward Goshorn, PhD, CCC-A, graduate student Kathryn White, and Brett Kemker, PhD, CCC-A, of the Department of Speech and Hearing Sciences at the University of Southern Mississippi. They measured the in-situ dB SPL produced by PDMDs worn by 31 young adults, ages 18-23, who had been wearing the devices for three years or less.

Subjects set the device to their preferred listening level (PLL) for a popular tune. The researchers then obtained in-situ SPL measures of the tune at a reference point and then for white noise using a probe microphone located near the eardrum. They also obtained hearing thresholds. There was a range of 52 to 107 dB, with a mean PLL of 83 dB, for music and a range of 53-99 dB, with a mean PLL of 79 dB, for white noise. None of the subjects had significant hearing loss.

Graduate student Riya Singh, doctoral student Woojae Han, MS, and Jont Allen, PhD, of the Department of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign, spoke on "Comodulation Masking Release to Detect Cochlear Dead Regions in Hearing Impaired Ears." They investigated whether comodulation masking release experiments with individuals who are hearing impaired can be used as a reliable method to detect any dead regions in the cochlea.

The experiment involved detecting a pure tone in a narrow band of noise in the presence of four other narrow bands of noise that fall outside the auditory bandwidth at the center frequency of the on-signal band. These flanking bands either were co-modulated to have the same amplitude envelope as the on-signal band or modulated randomly and presented monaurally to an impaired ear. Each band was level-adjusted to be audible to the hearing-impaired ear. The researchers are collecting data for both cases from listeners with normal hearing and from those who are hearing impaired and have no dead regions.

Every person with hearing loss also undergoes psychophysical tuning curve and threshold-equalizing noise tests, which are used to detect cochlear dead regions. Speech perception tests with CV syllables also are administered under various noise conditions. The consonant loss profiles from the perception experiments help make a reasonable estimate of the possibility of a dead region. The research was supported by the National Institutes of Health.

The question of how hormones affect a woman's voice is relevant to professional singers because hormonal fluctuations may place them at risk of injury. Knowing when the risks are greatest would help singers avoid performing at those times. One of the most dramatic hormonal fluctuations occurs during pregnancy, and many professional singers have experienced difficulty singing while pregnant. However, scientists do not know if this effect is due to hormones or to some other cause, such as decreased lung capacity as the baby grows.

In order to assess the effect of hormones on a pregnant singer's voice, Filipa Lã, of Aveiro University in Portugal, followed a professionally-trained Portuguese singer through 12 weeks of pregnancy and for 12 weeks after birth. She reported her findings in "Observations of the Singing Voice During Pregnancy: A Case Study," the first longitudinal study of the effect of hormones on a singer's voice during pregnancy.

Once a week, including just two days after the baby was born, Lã recorded the singer reading and singing into a device that measures the pressure exerted to make each sound. She collaborated with Johan Sundberg, of KTH in Stockholm, Sweden, to reconcile the data with measurements of the singer's hormone levels. The increased levels of hormones correlated with changes to the singer's vocal folds, they found. Though temporary, the changes forced the singer to exert more pressure from her lungs to make the same notes.

"It's harder work during pregnancy to sing," stated Lã. However, she acknowledged that her preliminary research was based on a single case study. Larger studies are needed before doctors can give solid advice to professional singers.

For the past year James Dembowski, PhD, CCC-SLP, has been working with a middle-aged transgendered woman, teaching her to use her male vocal anatomy to speak in a womanly way. He detailed the process in "Acoustic Changes Associated with Transgender Speech Therapy: A Case Study."

"She did not want a high-pitched voice," said Dr. Dembowski, of the Department of Speech, Language and Hearing Sciences at Texas Tech University Health Science Center. "As a successful academic in a local university, she posed an interesting challenge. She wanted to sound feminine but not stereotypical."

When boys reach puberty, rising levels of testosterone change the anatomy of their vocal cords, causing the voice to crack and drop. Older female-to-male transgendered individuals taking hormone supplements experience similar changes. But when a man transgenders to a woman, the vocal anatomy stays fixed, and old speech habits have to be unlearned.

Men and women tend to speak at different pitches, around 130 Hz for men and 200 Hz for woman. However, many women also have lower voices, said Dr. Dembowski, who focused on other characteristics drawn from the scientific literature. For example, women tend to speak slower, and men often have more rasp and croak in their voices. Men tend to emphasize words by speaking louder, while women vary their pitch.

By recording his client as the therapy progressed and analyzing the speech quantitatively, Dr. Dembowski learned that some features are easier to change than others. "It takes a lot of work," he said.

Researchers from the University of California, Los Angeles (UCLA) and the University of Washington (UW) reported on the "Consequences of Short-Term Language Exposure in Infancy on Babbling." Babies begin to develop the ability to understand their parents' speech at 6 months, long before they can actually speak. As they acquire the ability to perceive their native language, they become especially attuned to its subtle nuances in sound, research has shown. At the same time they lose the ability to perceive differences between foreign sounds. Even short-term exposure to a foreign language can change this. Limited experience with a foreign language can affect a child's ability to process the sounds of that language.

To find out if exposing infants to foreign languages influences how they babble, Nancy Ward and Megha Sundara, PhD, of UCLA, and Patricia Kuhl, PhD, and Barbara Conboy, PhD, of the UW Institute for Learning and Brain Sciences, studied a group of 13 1-year-old children from English-speaking households. During five one-hour sessions over several weeks, the babies interacted with a native Spanish speaker. They then were recorded while interacting with a parent in English or with a Spanish speaker, and the recordings were played to adult listeners who attempted to identify which babbles were from English-speaking babies.

Linguists have known for some time that when someone who is fluent in one language becomes fluent in another, the pendulum can swing too far in the opposite direction. The way they pronounce certain words in their first language changes, making a native speaker sound foreign. This "phonetic drift" often happens to people who move to other countries and become assimilated in the culture. They lose some of their native language because they don't use it much.

Charles Chang, MA, a doctoral student in the Department of Linguistics at the University of California, Berkeley, who specializes in phonetics and second language acquisition, showed that this effect can happen faster than was previously thought in his presentation on "Native Language Phonetic Drift in Beginning Second Language Acquisition."

He studied a group of 20 Americans who took a six-week Korean immersion program in Chuncheon, South Korea. The way they pronounced certain sounds in English changed in as little as one week. This small study raises fundamental questions about the neuroscience of language. How does the brain keep sounds from different languages distinct, and how does learning a new language affect the memory of another one?

Food that is swallowed and air that is breathed both aim down the throat, but they must go down the right pipe to the stomach or lungs. While healthy people usually can sort this out, food occasionally goes down the wrong way, resulting in coughing. Sometimes food or drink will get lodged on the vocal folds of the larynx.

Shanmugam Murugappan, PhD, and colleagues at the University of Cincinnati studied how the presence of a foreign substance changes the quality of speech. In his presentation on "Characterization of Acoustic Behavior When Prandial Material Is Present in the Larynx," he expressed his belief that this knowledge will help physicians in diagnosing patients with swallowing or certain breathing problems.

"Low-Back Vowel Merger in Minnesotan English" was the topic of Christina Esposito, PhD, a visiting professor of linguistics at Macalester College in St. Paul, MN. The two low-back vowel sounds in "caught" and "cot" sound as different as their spellings to native New Yorkers, but they are pronounced identically by Californians. The two sounds are now merging region by region. Dr. Esposito tracked the change, moving east across the Midwest into Minnesota.

Working with graduate students Hannah Kinney and Kaitlyn Arctander, she asked Minnesotans to read a list of 100 words that contain the vowels, recorded their speech, and analyzed patterns within the recordings. Spectrograms provide a visual representation of speech. "Every single vowel has its own unique frequencies, like a fingerprint," said Dr. Esposito, who recorded and dissected the speech quantitatively. Her results suggest that 30 percent of Minnesotans have lost the distinction between the two vowel sounds.