Assignment 4: Prototypes, Spatial Categorization and Psycholinguistics Experiments

Assigned Thursday, February 22th
Due Thursday, March 1, in class -- please turn in paper write-up. Electronic submission on bSpace is okay for late submissions. Problem 2 will require that you find a language informant. Please start early.

---

Problem 1: PROTOTYPES

Here are some types of prototypes and prototype effects, and the forms of reasoning with which they are usually associated:

Pick ONE category from this list: condiment, engineering, pet

1. Give a brief description of the internal structure of your chosen category. Does it have a central case (or cases)? If so, what properties hold for the central case?
2. Briefly describe 5 prototypes or prototype effects relevant to your category, and show how each gives rise to the type of reasoning indicated. (If necessary, provide context/reference to indicate where you found the examples.)

---

Problem 2: SPATIAL CATEGORIZATION

The illustration above is from a paper by Melissa Bowerman and Soonja Choi, titled Space under Construction: Language-Specific Spatial Categorization in First Language Acquisition. In it you see several actions, most of which are encoded by the English words "put in" or "put on". In particular, you see depictions of:

1. Put the cup on the table.
2. Put the magnet on the refrigerator door.
3. Put the hat (on the head).
4. Put the ring on the finger.
5. Put the top on the pen.
6. Put the Lego on the Lego stack.
7. Put the button in the buttonhole. (Button the button.)
8. Close the drawer.
9. Put the cassette in the case.
10. Put the book in the case.
11. Put the puzzle piece in the puzzle.
12. Put the apple in the bowl.
13. Put the book in the bag.

Your task will be to find out how these actions are encoded in another language. After finding a language consultant fluent in a non-English language, ask him or her to translate these sentences with the help of the illustration. (The language should be one that you are not fluent in yourself, although it is permissible for you to have some familiarity with it.) Either you or your language consultant must write down the translations, and you must create a gloss for each translation. A gloss is a word-for-word annotation of the sentence. (You may have to select Unicode UTF-8 encoding on your browser to view the below correctly.)

   Bǎ  bēizi  fàng  zài  zhūo   shàng. (Mandarin)
   BA  cup    put   at   table  top
   Put the cup on the table.

Unglossable elements maybe left untranslated. After completing the glosses, boldface the non-English verbs and underline the non-English spatial terms. For example:

   Bǎ  bēizi  fàng  zài  zhūo   shàng.
   BA  cup    put   at   table  top
   Put the cup on the table.

Then answer these questions:

1. What are the various spatial words that your language uses for these actions?
2. What are the various verbs that your language uses for these actions?
3. What distinctions does your language not encode that English does? Does your language make an on/in distinction?
4. What distinctions does your language encode that English does not? If your language uses different words for what English uses the same word for, how does your language decide which word to use?

---

Problem 3: fMRI study

A series of behavioral and imaging experiments that studied category-specific impairments to semantic memory in patients with brain lesions, suggest that different parts of the brain are involved in encoding, for example, animate versus inanimate objects. And indeed, Martin et al. demonstrated that some different regions of the normal brain are associated with animals versus tools, using the imaging technique known as positron emission tomography (PET) (Martin et al., 1996).

Subjects were asked to identify animals and tools while they undergo PET scans. While some areas of the brain were active in both tasks, naming animals selectively activated the left medial occipital lobe, and naming tools selectively activated the left premotor area and the left middle temporal gyrus. (The left medial occipital lobe is traditionally associated with visual processing, whereas the premotor area is traditionally associated with making movements, the temporal gyrus with generating action words.)

Now, we would like to repeat the PET experiment using fMRI.

1. Give three reasons why scientists might want to try an fMRI version of an experiment that had been done using PET.

Here is the design of our experiment:. We would first like to find further evidence for the use of different brain regions for processing animals and tools. Our hypothesis is that certain regions of the brain will be selectively activated for animals and certain others for tools.

The stimuli consist of 18 photographs and can be evenly divided into three categories: nonsense objects, animals, and tools.

• Nonsense objects include: randomly crossing lines and 3-D shapes that do not correspond to real live objects.
• The animals are: a dog, a horse, a wolverine, a capybara, a giraffe, a badger.
• The tools: a wrench, a hammer, a lathe, a screwdriver, an embosser, a chainsaw.

Each subject is presented with the sequence of 18 stimuli in random order, and is asked to name each stimulus silently. Each stimulus is presented for 100ms, followed by a centrally located fixation cross for 500ms. An fMRI image is taken 100ms after the stimulus is presented. Each subject is told that he will be asked questions about the pictures at the end of the fMRI session to help make sure he pays attention during the experiment.

At the end of the experiment, for each subject, we will compare the fMRI images across different stimuli. We will do three kinds of comparisons:

1. Subtracting activation due to nonsense objects from activation due to animals and tools.
2. Subtracting activation due to animals from activation due to tools.
3. Subtracting activation due to tools from activation due to animals.

More questions:

2. Does this experiment effectively test our hypothesis? What kind of results will support our hypothesis? What are possible flaws of our experiment setup? How can you improve it?
3. Could we do better by combining all the subjects' data? What difficulties might we encounter?