Ch. 19 | Exercise 2

# Chapter 19 | Exercise 2

### Case study 2 ‘English causatives revisited’

Using the data set `caus` from Chapter 15, perform an MCA with causative constructions as supplementary points. Tip: to deal with missing values, use  `na.method = 'Average'`.

1.

Perform MCA of the data. Make plots of Dimensions 1 and 2, and 2 and 3.

2.

Interpret the dimensions.

3.

Compare the location of supplementary points with the clustering based on the Behavioural Profiles of the constructions in Chapter 15. Are the results similar?

```> library(Rling) > data(caus) > library(FactoMineR) > caus.ca <- MCA(caus, quali.sup = 1, graph = FALSE, na.method = "Average") > plot(caus.ca, invis = "ind", col.var = "darkgrey", col.quali.sup = "black") > plot(caus.ca, invis = "ind", col.var = "darkgrey", col.quali.sup = "black", axes = c(2, 3)) ``````> dimdesc(caus.ca) \$`Dim 1` \$`Dim 1`\$quali R2 p.value CrSem 0.6487981 7.485609e-104 Cx 0.4061690 1.553261e-45 CdEv 0.2458183 4.129346e-28 Poss 0.2287618 4.201520e-27 CeSem 0.1377642 4.096571e-15 Neg 0.0831522 4.577684e-10 Coref 0.0422062 1.118512e-05 [output omitted] ```

Dimension 1 corresponds to the contrast between animate and inanimate Causers, Dimension 2 contrasts mental with non-mental (social and physical) caused events, and Dimension 3 is related mostly to the presence or absence of coreferentiality between the Causer and other participants.

The clustering of constructions with regard to two first dimensions resembles that obtained in Chapter 15: cause_V, have_Ving and make_V are close to one another, and get_Ving is on its own. However, the distinct cluster of get_Ved and have_Ved is not found on the MCA maps. On the contrary, the constructions are even divided by Dimension 3.