Want to keep learning?

This content is taken from the Lancaster University's online course, Corpus Linguistics: Method, Analysis, Interpretation. Join the course to learn more.

Skip to 0 minutes and 11 seconds Now I would love to give you another example. So far we looked at the large patterns through the lens of corpus linguistics. In this example, I would like to give you a traditional, variationist approach, and show how this actually works in the corpus linguistic context. Again, I have a question for you. Which of these sentences sounds more natural to you? And I have two options. Option one or Option two. At this stage, you might like to pause the video again and think about these.

Skip to 0 minutes and 49 seconds OK, is it one, or was it two for you? Just to look at what the sentences do, it is actually one same utterance with a slight modification. In the first utterance, we have a possessive construction, farmer’s field. Whereas in the second construction, we have a possessive construction field, of the farmer. In linguistic terms, we call the farmer, the person that possesses something, a possessor. Whereas the field, the object of possession a possessum. In the first construction, we have a sequence of possessor plus possessum, whereas in that second construction, we have a possessum, a field, followed by the possessor, the farmer. So the question is, which one we would prefer and in what context?

Skip to 1 minute and 44 seconds And as you can probably see, these are typical examples of Labovian sociolinguistic variable. These two sentences mean the same thing. The question is, which contexts and constraints would prefer the first option over the second option?

Skip to 2 minutes and 10 seconds So we designed a study that was exactly targeting the research question that I asked you before. So what are the contexts that favour one as opposed to the other variant? This is a research study that I carried out together with Miriam Meyerhoff, Susan Reichelt, and Dana Gablasova. We were looking at constraints, both external, social, such as gender, age, but also internal, that is, the linguistic context in which the “‘s” or “of” possessive constructions appear. We searched the two corpora, the British National Corpus 1994 and the BNC 2014 and obtained a dataset with over 5,000 target variables, of which 2,219 were the “of” constructions, whereas 2,792 were the “‘s” constructions.

Skip to 3 minutes and 14 seconds We coded these manually, some of these categories, and some of them were coded automatically and checked manually afterwards, so semiautomatic coding. We obtained a very, very large dataset. And you can see just small part of it. You have to do a lot of scrolling to be able to see the entire dataset. And it takes really weeks to process this dataset and weeks of intensive labour to be able to do all the coding. When we processed this dataset statistically, we arrived at some very interesting patterns. First of all, we looked at the contexts, the internal contexts, in which the possessive constructions appear. Now let us look at some of the results.

Skip to 4 minutes and 8 seconds What you can see here is a range of options that we have, some of which prefer the “‘s,” some of which prefer the “of” construction. The “‘s” construction is the lighter colour. The “of” construction is the darker colour. What we can see are contexts in which the possessor is animate human and the possessum is animate human, such as “Peter’s friend.” Peter is the possessor and friend is the possessum in the linguistical terminology. To the other pole, where we have the inanimate possessor and inanimate possessum, “the roof of a house.” In this context, the “of” genitive is much more frequent in these types of constructions. So again, we have the range of linguistic contexts in which these can appear.

Skip to 5 minutes and 8 seconds When we look at some of the social, external variables, such as variable of age, we can see that there’s a very interesting pattern indeed. We are looking at age groups 75 to 95 to 0 to 14. And we are also separating these two corpora to see whether there’s any development over the period of 20 years, with the 2014 corpus being the line at the bottom with these circles here, following the same pattern as the top line, but being lower on that “‘s” genitive. This is very interesting, because just at the first sight, we might assume that the “‘s” genitive is no longer as popular as it used to be 20 years ago. But is this really the case?

Skip to 6 minutes and 5 seconds When we look at the data, we actually see that this is not the case, because in the subset from 1994, 59%, so almost is 60%, include the human possessor– and we know that these contexts favour the “‘s” genitive– while only 41% of the sample from 2014 includes the human animate possessor context. So again, the reason for this discrepancy is that we are using data with different internal constraints on the use of these two constructions, which is very, very important to note. So what we did in the next step was to build a complex statistical model that would account for all the variables that are relevant and their interactions.

Skip to 7 minutes and 3 seconds And here are some of the results that I would like to show you. Again, you don’t need to understand the table and all the details. What is important to know is some of the highlights that we came up with as part of our research. First of all, we can see that the “‘s” genitive clearly favours animate contexts of which the typical possessive construction is of ownership and kinship. So that’s the “‘s” genitive and the animate context, the “Peter’s friend” type.

Skip to 7 minutes and 41 seconds We also looked at the phonological contexts and the pronunciation of these, because that plays a role as well. We know that sibilants– /s/ /z/ /ʃ/ /dʒ/ those are consonants that actually disfavour the “‘s” genitive for a simple reason, because if we have too many Ss in a row, these are very hard to pronounce. “Socrates’ death” is harder to pronounce than “the death of Socrates.” There’s also a very interesting effect of gender in the 1994 corpus with female speakers preferring the “‘s” genitive over the “of” construction, but this effect of gender disappears in the 2014 corpus. No other external social variables play a role in this variation, which is very interesting indeed as well.

Skip to 8 minutes and 44 seconds What we can however see a very interesting trend with more blurred distinction between the “‘s” and the “of” genitive constructions, with the “‘s” genitive moving into the territory which was previously inhabited by the “of” genitive in the 2014 data. So there’s this blurred distinction. There are still very clear constraints in operation we can see both in 1994 and 2014, but the distinctions are not as dramatic as they used to be 20 years ago.

PART 3: Case study 2. Possessives in spoken English

Vaclav Brezina presents another case study on the use of possessives ‘s vs. of in Spoken BNC1994 and Spoken BNC2014.

The case study demonstrates the advantages of the variationist approach with a clearly defined Labovian sociolinguistic variable.

Share this video:

This video is from the free online course:

Corpus Linguistics: Method, Analysis, Interpretation

Lancaster University