PCA Online

CSF Walk-through A

Aim: to explore broad differences between the three genres.

1. Click on “Start the CSF applet”.

2. From opening “Text Segmentation Method” screen, select “By Play”. Click “Select Plays”.

3. From “Play Selection” screen, choose “Select All”. Click “Select Characters”.

4. From “Character Selection” screen, choose “Select All”. Click “Select Words”.

5. On the “Word Variable Selection” screen, go to the “Auto Populate Word List” area, select “Entire Corpus” and for “Word List Size” 100. Click on “Populate” and once the list is populated click on “Display Results”.

6. On the “Word Frequencies by Play” screen click on “PCA Plot”.

7. On the “Principal Component Plot by Play” screen, toggle symbol labels off by clicking on the graph, to see the genre pattern, and on again to identify individual plays. Click on “Component Scores”.

8. On the Principal Component Scores by Play” screen click on “PCA Word Plot”.

9. Note the words at the extremes of PC1, i.e. our and their (high values, to the right) and you and I (low values, to the left). Click on “Component Scores”.

10. On the “Principal Component Scores by Word” screen, click on the “First Component” to sort the words in order of First Component score and get the full picture of the heavily weighted words on this component. Scroll down to see the words at the low-scoring end. Words with low scores are evidently associated (among other things) with speakers speaking as individuals, those with high scores with speakers speaking for groups.

11. Click on the “Back” button until the “PCA plot” screen reappears and compare the distribution of genres and plays across the First Component with the words appearing at the two extremes of the Component.