1

SPPA 6150 Project

You have two choices for your SPPA 6150 project: (1) carry out a modest-sized study, or (2) write a conventional literature review. Both options involve writing a paper, although the nature of that paper will differ depending on which direction you choose. You will also give a presentation on your project at the end of the term. The two types of projects are discussed in detail below. Even if you have already decided which of these options you plan to pursue, you will need to read all of the material below since some of this information applies to both.

Study

The study option involves designing, conducting, and describing the results of a modest-scale experiment. The study need not be very extensive and, depending on the exact nature of the project, may involve a fairly small group of subjects. You are encouraged to apply any knowledge that you may have of descriptive or inferential statistics, but formal statistics may not be needed for some projects. (Major exceptions: studies for which t-tests or correlations are appropriate.[1]) The main purpose of the project is to identify an interesting question (although not necessarily an original idea), to lay out a good way to test the question, and to provide a clear description of the study. The results of your research will be reported in the form of a written report (~12 pages, double-spaced, not including figures, tables, references, etc.), and in the form of a 13-minute oral presentation.

To give you a general idea of the sort of project that I have in mind, brief descriptions of few sample studies are provided below, in no special order.

Example 1: Relationships between Voice Pitch and the Physical Characteristics of Speakers

Is there any relationship between the average fundamental frequency of a speaker and physical characteristics such as height, weight, head circumference, neck circumference, etc. The Visi-Pitch could be used to measure fundamental frequency.

Example 2: Speaker Identification Based on Audio Recordings

To what extent can listeners identify known speakers based on listening to audio recordings? Variables that could be studied might include the length of the recording, the type of speech sample (e.g., sustained vowels vs. reading of a standard passage, vs. conversational speech), the length of time that elapses between two sets of recordings that are to be matched, and the effects of vocal disguise. You might also consider an experiment involving speaker identification based on an examination of spectrograms (I could give you a hand setting this up if you need help. This is not at all difficult to set up or run.).

A related project would involve determining whether listeners can identify the gender of pre-adolescent boys and girls (e.g., about 11-12 years or younger) from voice recordings. Research indicates that the acoustic characteristics of boys' and girls' voices are very similar, so one might think that identification of gender would be difficult or impossible. However, some studies indicate that listeners perform above chance on gender identification tasks involving the pre-adolescent boys and girls. Other studies have shown negative results. This would be a fairly simple study to run. Some thought would have to go into the choice of speech samples (e.g., care would have to be taken to ensure that gender clues were not available in the content of the test utterances).

Example 3: Reliability Studies

As we will soon see, a very important indicator of the precision of a test is its consistency or repeatability. A lot can be learned by conducting fairly simple studies to determine, for example: (1) the extent to which two listeners agree on the severity of a voice disorder, (2) the extent to which an individual listener is internally consistent in his/her judgment of the severity of a voice disorder (see below), and (3) the repeatability of standardized tests of speech reading or some other commonly assessed ability, (4) the extent to which apparently simple measures such as speech intelligibility can be made in a way that is repeatable within and/or across observers.

Special Note:For anyone interested in voice quality, I have a nice database of recordings of disordered voices that can be used for a wide variety of experiments on the perception of vocal quality. There are hundreds of voices in the database showing a wide range of severities, etiologies, and voice qualities. Dr. Tasko also has a very interesting database of voices exhibiting functional voice disorders that were recorded both before and after treatment. I also have software that makes these kinds of experiments fairly easy to run. There are quite a few useful, interesting, and simple experiments that can be done in this area. For some ideas and literature, see Kreiman et al., Perceptual evaluation of voice quality: Review, tutorial, and a framework for future research. J Speech Lang Hear Res., 1993, 36(1):21-40.

Example 4: Replication of Studies in the Literature

Replicating existing published studies is just fine. One nice example from a few semesters back was a replication of the original experiment that produced the sone scale of loudness estimation. This was a nice idea since the experiment could be run with an ordinary audiometer, and meaningful results could be obtained from a fairly small number of subjects. Many other examples can be found by scanning the journals. You are looking for experiments that do not require elaborate equipment or methodologies.

Example 5: Measurement of Speech Intelligibility

There are many interesting and fairly simple studies that can be carried out that involve the measurement of speech intelligibility in dysarthric speakers or other speaker groups that tend to show intelligibility deficits. For example, how well do subjective estimates of speech intelligibility (e.g., 50% intelligible vs. 80% intelligible, etc.) correlate with intelligibility estimates that are measured objectively using standardized sentences or word lists?

Example 6: Tongue twisters

Do tongue twisters really create problemsfor a speaker by twisting the tongue? Probably not, or at least not entirely. There is evidence that utterances like “wrist watch” and “toy boat” that are difficult to say aloud also slow a “speaker” down when they are uttered subvocally (i.e., inside the head). Since the articulators are not moving during subvocal speech, it cannot be that motor execution is the only problem. It must be that tongue twisters create problems at the motor planning stage. The finding also indicates that subvocal speech must involve a motor-planning component rather than just the generation of an auditory image. This would be a very simple experiment to run, requiring only a volunteer and a watch with a second hand. (See Smith Hillenbrand, “Durational characteristics of vocal and subvocal speech: Implications concerning phonological organization and articulatory difficulty,”Journal of Phonetics, 1986, 14, 265-281.)

Example 7: Effects of experience on auditory abilities

Are musicians better at discriminating pitch differences than non-musicians? Are SLPs better at identifying or discriminating speech sounds than untrained listeners? Are experienced SLPs better at making these kinds of judgments than graduate students in their last semester of study? There are several simple-to-run experiments that could be set up to test questions such as these. I would be happy to help you get a project like this off the ground. (For just one example, see Hillenbrand et al. “Perception of intraphonemic differences by phoneticians, musicians, and inexperienced listeners,”Journal of the Acoustical Society of America, 1990, 88, 655-662.)

Example 8: Bite-block speech

I have not had time to write a description. I’ll explain it in class. It’s a pretty simple idea, it would be easy to carry out, and it’s a really interesting area of research.

Example 9: Effects on intonation contour on speech intelligibility

[fill this in]

Example 10: The role of stress in parsing

[fill this in: light housekeeper vs. lighthouse keeper, black board-eraser vs. blackboard eraser – measurement of f0, intensity, and duration]

Example 11: Berko-Gleason wugs replication

[fill this in]

Example 12: Effects of speaking rate on judgments of naturalness in dysarthric speech

[fill this in – see dysrate]

Example 13: Auditory Stroop effect (stereo/male-female)

[I’ll explain this in class. See Alvin sample experiments; search google scholar for Auditory Stroop effect – see especially Jerger, S., Martin, R.C., and Pirozzol, F.J. (1988). A developmental study of the auditory Stroop effect.Brain and Language, 35, 86–104.

Example 14: Mental rotation

Example 15: Effects of intonation on sentence intelligibility in quiet and noise

Example 16: Effects of listening experience on three-choice phonetic identification

Other Details about the Project

A. IDEAS

You are encouraged to come up with an idea of your own, but there is nothing wrong with carrying out one of the sample projects described above. Modest-size replications of studies already in the literature are appropriate.There are many studies in the literature that would be fairly simple to replicate on a small scale. For example, audiology students might be interested in replicating any one of several classic psychophysical studies; e.g., the mel scale, the phon scale, the effects of duration on loudness, etc. See me if you would like suggestions.

You will need to choose a project that can be completed in the relatively limited time that is available to you. Projects that require fancy instrumentation are not good choices, unless you are quite familiar with the instruments or can get help from someone who is. If you will require special assistance from someone, arrange for it immediately. Think carefully about where you will get subjects, and make arrangements right away.

B. HOW MANY SUBJECTS DO I NEED?

Nobody really knows exactly how to answer this question. However, when real studies are conducted in real laboratories, the researcher tries to think through how many subjects are needed to answer the questions that are being posed in the study. This is pretty obvious, but in the case of this seminar, we do not have the luxury of approaching the problem in this way. We will have to think in terms of the number of subjects that can reasonably be run in the relatively short time that you have to conduct your projects. These practical considerations are likely to vary quite a bit depending on the nature of your experimental tasks. For example, some of you may end up running studies involving tasks that can be completed in 15 or 20 minutes, and in those cases, it might be quite feasible to run 20 subjects or more. For studies with longer tasks or methods that require long and involved data analysis it may be feasible to collect data from much smaller groups. Try to think through this on your own, then see me for advice.

C. WHO CAN BE USED AS SUBJECTS

The safest bet is to find a project that can be carried out using other students as subjects. Any project involving children needs to be very simple. The tasks that the children are going to be asked to perform need to be either routine clinical procedures (articulation tests, language samples, diadokokinetic rate, receptive vocabulary tests, etc.) or some other similarly simple, non-stressful, and safe procedure. If you are interested in studying a clinical population, see me. Some clinical populations would work out fine (e.g., children with articulation disorders or relatively mild language disorders), but others involving cognitive impairments (e.g., autistic children) are not a good choice.

D. LITERATURE SEARCHES

There are all sorts of methods available for finding literature on a topic. Do not limit yourself to just one method. In particular, topic searches using search facilities such as Medline are worth doing, but in my experience they often miss quite a bit. The secret is to find the one or two key papers that will have references to many other related papers which, in turn, will lead you to still more literature, and so on. Sometimes Medline will find these key papers and sometimes not. Don’t overlook textbooks, and do not be hesitant about asking faculty members who have expertise in your topic. Google Scholar, of course, is an excellent search tool ( although there are many others.

E. LITERATURE REVIEW PAPERS

In place of conducting a research project, you may choose to write a literature review. The key is to find a topic that will allow you to apply some of the research methods concepts that we will be discussing in this course. There should be some central question that you are trying to address, and it needs to be the kind of question that researchers have tried to answer with experimental evidence. Some examples might give you an idea of the kinds of topics that would be appropriate:

  1. Do cochlear implants result in significant improvements in: (a) speech production, (b) speech perception, or (c) language development? This has become a very extensive literature. You’ll have to pick and choose, focusing on what seems to be the most reliable evidence.
  1. Does chronic otitis media result in speech and/or language delays? You would think that this question would have been settled long ago, and most SLPs believe that it has been. It hasn’t. This is a great topic.
  1. Are perceptual judgments of voice quality reliable and/or valid? A lot has been written on this topic. Although it is getting a little long in the tooth, there is a very good review paper that provides a good framework for understanding the key questions in this area (Kreiman et al., Perceptual evaluation of voice quality: Review, tutorial, and a framework for future research. Journal of Speech and Hearing Research, 1993, 36, 21-40).
  1. What are the mechanisms by which the sound pattern of a native language interferes with the acquisition of a new language?
  1. Is swallowing therapy effective? This has become a pretty extensive literature by now, and it is almost certain that the topic will have to be narrowed down in some way. It is a manageable topic, though.
  1. Are non-speech oral-motor exercises effective in treating articulation disorders? There is an excellent review paper that will give you an excellent start on this topic (Forrest, K. Are Oral-Motor Exercises Useful in the Treatment of Phonological/Articulatory Disorders? Seminars in Speech and Language, 2002, 23, 15-26.) This paper is getting a little old, but it is well written and a good place to start.
  1. What is the evidence for and against a genetic basis for stuttering (or language disorders or speech-sound disorders or … – you name it)?
  1. Why do some children spontaneously recover from fluency problems while others go on to become chronic stutterers?
  1. Does speech or non-speech auditory perception play some role in the etiology of articulation (or language) disorders?
  1. What are the effects of aphasia on multilingual speakers? For example, are the two (three, etc.) languages affected equally, does it matter which language was learned first, the degree of pre-morbid fluency in each language, etc.?
  1. Is aphasia therapy effective? This is a huge topic with an extensive literature, and you would have to find some way to narrow it down. It is not an easy topic, but it can be done, and it’s a rich, interesting topic.
  1. What is the relationship between performance on nonword repetition tasks and developmental language disorders?In a nonword repetition task, as the name suggests, subjects are asked to repeat a made-up word, such as “urzuhla.” If there is a relationship between performance on nonword repetition tasks and developmental language disorders. It turns out that there is a very strong relationship.Why is nonword repetition such a strong predictor of SLI?
  1. What role does “theory of mind” play in autism. From Wiki: “Theory of mind is the ability to attribute mental states—beliefs, intents, desires, pretending, knowledge, etc.—to oneself and others and to understand that others have beliefs, desires and intentions that are different from one's own.” The central idea is way simpler than that it may seem from that egghead description. Here’s a concrete example of a theory of mind experiment. We have three people:Sally, Ann, and Milo. Milo is the subject in the experiment. Ann watches as Sally puts a doll in a box. While Ann leaves the room for a moment, Sally, being a jerk, moves the doll from the box to a basket. After Ann returns, the simple question for Milo (our subject) is this: Where will Ann look for the doll? The answer, of course, is that Ann will look in the box, the last place she saw it. Autistic children tend to perform poorly on this kind of task, and there are some investigators who have argued that a poorly developed theory of mind(the ability to reason based on an internal model of other people’s thoughts) is the central deficit in autism. There is a beautifully written, down-to-earth article in Scientific American that does a terrific job of explaining the idea and reviewing some of the experimental evidence (Frith, U.,“Autism,”Scientific American, 1993, 268, 108-114). This is a wonderfully interesting topic.
  1. What role does hearing play in the regulation of the speech of postlingually deaf speakers? Much of the literature in this area comes from a series of studies that were done with postlingually deaf cochlear implant users. The method was about a simple as it gets: analyze the speech of the CI users with the device on, then turn the implant off and do it again. Will the speech differ? If, in what way? It turns out that some aspects of speech are largely unaffected while other characteristics are almost immediately affected. For the most part, the studies are not technically complicated or hard to understand. It’s a good topic. For a start on this literature, see papers by various combinations of H. Lane, J. Perkell, M Svirsky, and M. Matthies.

*** There is nothing at all wrong with picking one of the topics listed above. ***