Ch. 10 | Exercise 3

Chapter 10 | Exercise 3

Consider the frequencies of selected collexemes of GO + ADJ construction in COCA, given in Table 10.1.

Table 10.1. Some adjectives that fill in the slot in the GO + ADJ construction. The data are retrieved automatically as the bigram go (lemma) + any adjective found in COCA

Adjective	Frequency in go + ADJ Cx	Total frequency in the corpus
haywire	226	297
hog-wild	12	19
batty	17	215
crazy	1821	24804
sick	5	24764
wrong	884	77845
stir-crazy	14	29
unpunished	182	259
blank	326	8478
undetected	201	698

The total number of occurrences of the construction is 28636. Find out which of the adjectives have high Attraction towards the construction and which ones have high Reliance. Make a plot with adjectives as text labels, with the horizontal axis showing the Attraction scores and the vertical axis displaying the Reliance scores.

Key

First, create two vectors: a, with the frequencies of the adjectives in the GO + ADJ construction, and total, with the total frequencies of the adjectives in the corpus:

> a <- c(226, 12, 17, 1821, 5, 884, 14, 182, 326, 201)
> total <- c(297, 19, 215, 24804, 24764, 77845, 29, 259, 8478, 698)

Next, compute the attraction and reliance scores and combine them as columns in a data frame:

> attr <- 100*a/28636
> rel <- 100*a/total
> go <- cbind(attr, rel)

Create a character vector with the adjectives and use them as the row names of the data frame:

> adj <- c('haywire', 'hog-wild', 'batty', 'crazy', 'sick', 'wrong', 'stir-crazy', 'unpunished', 'blank', 'undetected')
> rownames(go) <- adj
> go
[output omitted]

The highest Attraction scores belong to crazy (6.36), wrong (3.09) and blank (1.14). The highest Reliance scores are observed for haywire (76.09), unpunished (70.27) and hog-wild (63.16).

To make a plot based on the Attraction and Reliance scores, you can use the following code:

> plot(attr, rel, type = "n")
> text(attr, rel, adj)