Working with human subjects

• Stanley Milgram experiments in 1961-1962 (http://www.new-life.net/milgram.htm)

  • Question of how Nazis could get people to torture other people. Shock experiment with lab coated scientist standing by.

• Universities require institutional review board approval to insure ethicality, avoid lawsuits, maintain their reputation, protect people.

• At any rate, we have approval for this class.

Interviews

Typically, in interviews you don’t manipulate, lead conversation; but rather, ask specific questions to elicit data.

Examples:

• Study on which states have "best English"

• Questions about rural life 50-60 years ago

Types:

• individual / group

• open / closed:

  • closed (What is this called? What words do you use that your wife doesn't?)

  • open: let them talk and then see what you find (personal experiences, scary experience)

Issues:

1. Is it spontaneous, non-reflective speech?

2. Can the people really answer -- i.e. is it possible to retrieve their intuitions and knowledge?

Questionnaires and surveys

Examples:

• Please pronounce the following words: mail, still/steel, pull/pool, full/fool

• Does the u in student sound like the oo in too or the u in use

• After Bill had _____ (bought) the computer, he realized he'd made a mistake

• Which is better: Who am I talking to? / To whom am I talking?

• What do you call the thing you rent from Blockbuster (movie, show, video)

• What do you call the paper container you take things home from the store in? (sack, bag)

• Count to ten

• What are the days of the week

• What do people here say funny?

Some online examples:

• Take this survey

• pronunciation: creek, aunt, caramel, mayonnaise, pajamas, root, route, grocery

• morphology / syntax: you, coming with, where at, might could, anymore

• vocabulary: shoes, yard sale, lightning bug, shoes, diagonal corner, highway, eye guck, frosting, pop

Word stress (THIR-teen, thir-TEEN)

Dark and Light /l/ survey (click on link on bottom to see results)

Advantages of questionnaires and surveys:

• Specific items relate to specific hypothesis -- easy to elicit what you want

• Every subject gets the same questions (compared to trying to hear same pronunciation or construction in all subjects in spontaneous speech)

Issues in selecting subjects

• Representativeness

  • What if you only polled your friends or people in one neighborhood?

  • What if you did a survey on political views at BYU and extrapolated to all college students?

  • What if you did a mail-in survey on sexual mores and put it in Cosmopolitan vs. Reader's Digest?

    • Would only a certain kind of person answer?

• Compatibility (enough similarity between sub-groups, get info on them)

• Issue of anonymity (vs confidentiality)

How to elicit information:

• Written vs. auditory (still/steel, pill/peel)

• Given in person vs mailed out (what can skews results?)

• Online (advantages? disadvantages?)

• Whether or not researcher is present

• Direct (Rolling Stone at BYU. "I'm studying X", "investigate your dialect") vs. indirect

Different types of questions

• Completion: (A person who helps put out fires is called a ____): are all of the answers relevant?

• Picture: but are all of the answers relevant: tennis shoes, Nikes

• Multiple choice: (sneakers, tennis shoes, gym shoes): but have you listed them all?

• Likert-type scales: (This speaker sounds friendly/Jewish/low class): always positive/negative on one side?

• True/false; Yes/No Is the following sentence grammatically correct?

  • We’re going to the dance, do you want to come with?

  • They donated the relief fund a million dollars.

  • Are there ever any not pretty women in a beauty pageant?

• Open-ended questions ("What do you think is different about Utah dialect"): potentially leading / biasing ("What is is about Utah dialect that you find particularly strange?)

• General issues in terms of questions:

  • Usable data (numbers) vs patronizing (treating them as data-producers)

  • Number of questions

  • Ordering of questions

  • Running a pilot study first (for large projects)

General problems with questionnaires and surveys:

• Observer’s paradox-you want naturally occurring speech, but people may not produce that when they know they are being observed. Have microphone on them, person jotting down answers.

• People’s beliefs about what they do often don’t coincide with their behavior. (“I never use 'there is' with plurals and there's a lot of reasons I don't”)

• People may tell you what you want to hear (Hi, I’m Mormon, what do you think about Mormons?)

• People may sometimes lie and researchers are sometimes biased (Margaret Mead in Samoa; Romani word lists)

• Mismatch between researcher and subjects (e.g. African-American children and white, male, middle-aged researcher)

  • Yo dude, listen I’m Duke Radingstone from Rolling Stone magazine. I’m doing a story about how ridiculously conservative and sexually repressed the atmosphere at BYU is. Could I ask you a few questions?

• They can't measure variability (going vs. goin' in same person's speech)