Chapter 20 | Exercise 1
Case study
‘Diachronic and register variation of should and ought to’
For these exercises you will need the data set should_ought in Rling, which displays the frequencies of the modal auxiliaries with infinitives in four periods represented by COCA (1990–1994, 1995–1999, 2000–2004 and 2005–2009). The exercises combine motion charts with some methods that have been discussed in this book.
Investigate how the distribution of infinitives with should and ought to has changed over a short period from 1990 to 2009. To do so, create a motion chart of should vs. ought to using the data in should_ought in Rling. Since the number of words in the corpus components is nearly equal, one can compare the actual frequencies without normalizing them. Create a motion chart of the verbs that were used with the modal auxiliaries and interpret the the general direction of language change.
Compare the distribution of should and ought to across five registers of COCA. The frequencies are shown in Table 20.1.
| Spoken | Fiction | Magazine | Newspaper | Academic | |
|---|---|---|---|---|---|
| should | 88452 | 56658 | 68235 | 59858 | 81660 |
| ought to | 10720 | 3913 | 2191 | 2113 | 2297 |
In which registers are the frequencies lower or higher than expected? Use a mosaic plot with shading. Are the differences statistically significant?
In principle, you can represent any entities as moving objects, not only collexemes. Using the data in Table 20.2, visualize how the frequencies of should and ought to were changing from 1990 to 2009 in five registers, using the registers as moving objects. What are your conclusions?
| Register | Year | should | ought to |
|---|---|---|---|
| Spoken | 1990 | 22233 | 3995 |
| Spoken | 1995 | 21448 | 2870 |
| Spoken | 2000 | 18547 | 1825 |
| Spoken | 2005 | 16986 | 1428 |
| Fiction | 1990 | 12345 | 1074 |
| Fiction | 1995 | 11856 | 819 |
| Fiction | 2000 | 12053 | 818 |
| Fiction | 2005 | 13374 | 789 |
| Magazine | 1990 | 16378 | 705 |
| Magazine | 1995 | 16025 | 581 |
| Magazine | 2000 | 15053 | 442 |
| Magazine | 2005 | 14078 | 347 |
| Newspaper | 1990 | 13430 | 599 |
| Newspaper | 1995 | 14637 | 525 |
| Newspaper | 2000 | 13110 | 462 |
| Newspaper | 2005 | 12495 | 361 |
| Academic | 1990 | 18798 | 636 |
| Academic | 1995 | 18916 | 589 |
| Academic | 2000 | 17691 | 518 |
| Academic | 2005 | 17361 | 389 |
> library(Rling);library(googleVis)
> data(should_ought)
> mch <- gvisMotionChart(should_ought, idvar = "Verb", timevar = "Year")
> plot(mch)
The chart will show an obvious decline of ought to as a modal auxiliary over 20 years.
First, create a table with the frequencies of the modal verbs in R:
> should <- c(88452, 56658, 68235, 59858, 81660)
> ought <- c(10720, 3913, 2191, 2113, 2297)
> mod.reg <- rbind(should, ought)
> colnames(mod.reg) <- c("Spoken", "Fiction", "Magazine", "Newspaper", "Academic")
> mod.reg
Spoken Fiction Magazine Newspaper Academic
should 88452 56658 68235 59858 81660
ought 10720 3913 2191 2113 2297
> chisq.test(mod.reg)
Pearson's Chi-squared test
data: mod.reg
X-squared = 7804.823, df = 4, p-value < 2.2e-16
> chisq.test(mod.reg)$residuals
Spoken Fiction Magazine Newspaper Academic
should -16.74049 -2.063185 6.925206 5.730985 8.680305
ought 68.43570 8.434369 -28.310474 -23.428457 -35.485378
The residuals show that the modal ought to is very strongly overrepresented in the spoken data, and strongly underrepresented in the press and academic texts. For should, the reverse holds, although the residuals are overall smaller, which indicates a more even distribution. All this can also be seen in a mosaic plot:
> library(vcd)
> mosaic(mod.reg, shade = TRUE, varnames = FALSE)
The data from Table 20.2 can be entered manually or copied, saved as a text file with tabs as delimiters and read as a table in R (see Chapter 2).
> mod.reg.time <- read.table("YourDirectory/mod_reg_period.txt", header = TRUE)
> mch1 <- gvisMotionChart(mod.reg.time, idvar = "Register", timevar = "Year")
> plot(mch1)
The motion chart displays changes in the frequencies of both should and ought to in all registers. In the spoken subcorpus, the decline of ought to is especially dramatic. However, with the exception of fiction, the frequency of should is declining, too.