Skip to Content

Department of Linguistics

SPEECH PHYSIOLOGY

Respiratory Physiology Assignment (2008)

Aim

To familiarise students with:-

  • scientific method
  • research design
  • experimental methodology
  • data analysis
  • research paper writing

as they pertain to research in speech physiology.

Method

Examine the following two papers(1) on respiratory physiology in speech production and answer the questions outlined below:-

  1. Winkworth, A.L., Davis, P.J., Ellis, E., and Adams, R.D., "Variability and consistency in speech breathing during reading: Lung volumes, speech intensity, and linguistic factors", J. Speech & Hearing Res., 37, 535-556, 1994

  2. Winkworth, A.L., Davis, P.J., Adams, R.D., and Ellis, E., "Breathing patterns during spontaneous speech", J. Speech & Hearing Res., 38, 124-144, 1995

Please note that you should have received (by the afternoon of the first lecture) an email welcoming you to the unit and also explaining how you can obtain a copy of each of these two papers.

Topics and Questions

Interspersed amongst the following 5 topics are 9 questions that students must answer as part of this assignment.

1. Formulating a research question

When a scientist is planning an experiment or a research project, he/she must first become familiar with research carried out by other scientists by examining the available literature on the topic. A good project is one which aims to answer one or more questions that have been left unanswered or have only been partially answered by previous studies. In a typical research paper, such issues are usually discussed in the first part of the paper which reviews the relevant literature. Research questions are usually more general than the hypotheses (see below) of a study and are phrased in less specific language than are hypotheses.

Question 1: What new questions are being asked by paper 1? That is, what questions were left unanswered in preceding work by these and other authors?

Question 2: What additional questions are being asked by paper 2 that were not answered by paper 1?

2. Hypotheses

Once the scientist determines which questions remain unanswered or partially answered and has chosen one or more questions to examine in the proposed research, he/she decides upon one or more hypotheses. A hypothesis is effectively an educated guess (based on knowledge of prior research) as to what the answers to each of the research questions will be. The goal or aim of any research project which follows the scientific method is to determine whether one or more hypotheses are true. In other words, the goal of an experiment is to test a hypothesis. An experiment might prove a hypothesis or disprove a hypothesis, but more often an experiment will either "support the hypothesis" or "not support the hypothesis". Most research is incremental. Experiments rarely "prove" or "disprove" a hypothesis. More often they form part of a larger body of work that either supports or doesn't support a hypothesis. Proof of a hypothesis is often built up from the results of the research of numerous scientists. Note that the questions asked by the researchers are not precisely equivalent to the hypotheses. A hypothesis is usually a positive statement that needs to be verified or rejected.

In paper 1 the hypotheses are implied in the first section of the paper (the part preceeding the method). The hypotheses in this paper are not stated explicitly but could be summarised as:-

  1. Intra-speaker variation in respiratory manoeuvres relating to speech is not as great as inter-speaker variation.
  2. Just as the position of "breath-taking" in speech is strongly related to linguistic factors such as paragraph, sentence and clause boundaries(2), more detailed aspects of pulmonary behaviour such as the lung volume level at which speech is initiated or terminated are also related to the linguistic structure of the passage being read.
  3. During reading, speakers plan their respiratory manoeuvres to accommodate their knowledge of the likely duration of the next sentence or clause.
  4. There is a relationship between the lung volume used for each "breath group"(3) and the intensity of the speech during that breath group.

Question 3: What are the hypotheses (both explicit and implicit) in paper 2?

3. Methodology

Building upon the papers reviewed in the first part of the paper, the section on methodology describes in detail how each experiment was carried out. The details of the methodology are often supported by reference to the methodology of other scientists in previous studies. Established methodology often requires less explanation than does novel methodology as prior studies can be referred to for details.

In the present studies a device known as a Respiratory Inductance Plethysmograph (the "Respitrace") was used to measure lung volume (see figures 1 and 2). The Respitrace indirectly measures the changes in the volume of the lungs by directly measuring changes in the cross-sectional areas of the rib-cage and the abdomen (see the white bands on the chest and abdomen in figure 1). If you look at figure 1, you will see that the subject has a mask over her mouth and nose which measures actual airflow into and out of the respiratory system. This mask is not in place during the actual experiment as the mask interferes with the subject's ability to speak naturally and also interferes with the recording of the speech.

The mask is used to "calibrate" the Respitrace output when setting up the experiment. In many physiological experiments it is necessary to calibrate the equipment being used.

Calibration enables the experimenter to be sure of the reliability of the measurements. In the case of these experiments it was necessary to determine the exact relationship between the measured movements of the rib cage and the abdomen with actual changes of lung volume. The calibration procedures used in these experiments were quite complex (see paper 1, pp 536-538 and paper 2, pp 125-126, the sections entitled "Respiratory" and "Respiratory Measurement" respectively).


Figure 1: "Respitrace" Demonstration of the use of a respiratory inductance plethysmograph manufactured by Non-Invasive Monitoring Systems, Inc., USA. (reproduced with permission: Iwarsson, Thomasson and Sundberg, of the Music Acoustics Group in the Department of Speech, Music and Hearing, Royal Institute of Technology (KTH), Stockholm, Sweden.)


Figure 2: The respiratory inductive plethysmography system including the rib cage (1) and abdominal (2) sensor bands, the oscillator (3), and the signal demodulator (4). (Thomas Strömberg, Department of Biomedical Engineering, Linköping University, Sweden)

Question 4: Explain, in your own words, why you think that calibration was necessary for these experiments.

Question 5: Why was it necessary to perform the calibration for each subject? Why was this calibration also repeated for each experimental session?

In speech research it is normal to describe in detail certain characteristics of the subjects who participated in the experiment. Very often various aspects of the subjects' background and identity can influence the results obtained in the experiment. Certain subject characteristics can affect the outcome of the experiment. Such characteristics are known as "independent variables" if they are independent from each other in their effect and are also independent from the variables being tested. Good experimental design controls independent variables by taking them into account, preferably by eliminating them from the experiment in some way. Uncontrolled additional independent variables make the interpretation of results either very difficult or sometimes even impossible.

In the present experiments the subjects are described in the following way:-

"Six female subjects with no history of speech, hearing or respiratory disorders were recruited. All were native speakers of English, nonsmokers, between 19 and 22 years old." (Winkworth, 1994, p 37)

The following speaker characteristics were considered important to mention and control:-

  1. female
  2. no speech disorders
  3. no hearing disorders
  4. no respiratory disorders
  5. native speakers of English
  6. nonsmokers
  7. between 19 and 22 years old

All of these speaker characteristics are important to the design of these experiments. If any of these restrictions upon subject characteristics were to be relaxed the experimental results would become more complex and difficult to interpret.

Question 6: Why do you think that each of these speaker characteristics was controlled? Deal separately with each characteristic.

(The answer to this question cannot be found in the papers. You will need to deduce the reasons for yourself.)

One of the main differences between the studies outlined in paper 1 and paper 2 is the use of speech resulting from the reading of written passages versus the use of "spontaneous" speech.

Question 7: What is spontaneous speech?

Question 8: What are the relevant differences between read and spontaneous that might affect pulmonary behaviour?

4. Results

In most scientific papers a considerable part of the paper is devoted to the detailed description of the results. In many papers such a description of the results includes tables and graphs which describe the data as well as details of any statistical analyses of the data. The results section rarely includes in-depth discussion of whether the results support or fail to support a hypothesis. Such discussion is usually left to a special discussion or conclusion section.

In the results section of the two papers there are a number of references to such statistical terms as "significant" and "correlate". Further, these papers state that "a significance level of .01 was adopted".

Correlation describes the relationship between two variables. A positive correlation occurs when an increase in one parameter is accompanied by an increase in a second parameter. A negative correlation occurs when an increase in one parameter is accompanied by a decrease in a second parameter. A perfect correlation occurs when all measurements occur on a straight diagonal line. A perfect correlation has a value of 1.0.

Statistical significance means that a statistical test has determined that the relationship that is described between two parameters is unlikely to have occurred by chance. In other words, a statistically significant result for a sample of data is a reliable representation of the true relationship between two parameters in the world at large. A significance level of 0.01 states that the experimenter has decided that a relationship between two parameters will only be accepted as significant if there is 1% or less chance of the relationship occurring in the experiment because of chance or random factors.

Statistical significance can also be applied to a comparison between two parameters. For example, if one says that parameter A is significantly different (at the .01 significance level) from parameter B then this means that there is a 99% probability (or better) that the two parameters really are different. If, on the other hand, one says that parameter A is not significantly different (at the .01 significance level) from parameter B, this means that the probability that this relationship occurred by chance is too high (> 1%) and so we are not able to say with confidence that the two parameters are different.

5. Discussion

In the present two papers the discussion and the conclusions are presented in a single "discussion" section.

Question 9: Describe the main conclusions of the studies outlined in papers 1 and 2.

Format of this Report

This is not a research report on an experiment that you have carried out. This report should be treated as a series of short answer questions rather than a research report. In other words, it is sufficient to type out the question and answer for each question.

  • Do not include extensive background material (i.e. NO literature review)
  • Do not include diagrams or other illustrative material
  • Just answer the 9 questions
  • Keep it short! A typical size is between 2 and 3 pages (A4, 12 point, standard margins)

Submission of Assignment

Email it directly to me (robert.mannell@mq.edu.au). Don't forget to attach the Department of Linguistics coversheet. Please do not email it to the Department Undergraduate Office.

Some Abbreviations

The following are some of the abbreviations used in these papers:-
  RIP Respiratory inductance plethysmography
  VC Vital capacity
  REL Resting expiratory level
  ILV initiation lung volume (lung volume level at which speech was initiated)
  TLV termination lung volume (lung volume level at which speech was terminated)
  SLV speech lung volume (ILV to TLV)

Footnotes

  1. I will provide details, by email, on how to obtain these two papers. These papers will be referred to as paper 1 and paper 2 in this handout.
  2. This has already been established by other studies and does not form part of the present hypothesis.
  3. A "breath group" occurs between two successive inspirations.