# Chapter 6 | Exercise 2

What do you think about the relationship between the imagery and concreteness scores from the experimental data in `pym_high`

and `pym_low`

(see Chapter 4)? Would you expect a positive or negative correlation? Compute the Kendall *tau* and perform the correlation test for the concreteness scores and imagery scores of the high-frequency nouns in `pym_high`

and repeat the procedure for the low-frequency nouns in `pym_low`

. Are your expectations met? Are the relationships between the scores the same for the high- and low-frequency nouns?

One can expect a positive correlation between the scores. Therefore, you can use a one-tailed test.

```
> library(Rling)
> data(pym_high)
> data(pym_low)
> cor.test(pym_high$imag, pym_high$conc, method = "kendall", alternative = "greater")
[output omitted]
> cor.test(pym_low$imag, pym_low$conc, method = "kendall", alternative = "greater")
[output omitted]
```

There are significant strong positive correlations (*p* < 0.001). The coefficients are very similar, although for the high-frequency nouns the coefficient is slightly higher (*τ* = 0.716) than that for the low-frequency nouns (*τ* = 0.704).