What do people ‘like to say’ – and what do they ‘like saying’?Posted by Heng-ming Carlos Kang on April 04, 2012
This post comes from guest blogger Heng-ming Carlos Kang of the Graduate Institute of Linguistics at National Chengchi University in Taiwan. It is based on a talk that Carlos gave at the recent Asia-Pacific Corpus Linguistics Conference in New Zealand.
Have you noticed any difference between what people like to say and what they like saying? Now think again: can these two patterns be used interchangeably? In most English grammar books, gerunds (V–ing) or infinitives (to-V) after the verb like are treated as two ways of delivering the same message. But when you look more closely at the data, the semantic differences between gerunds and infinitives become clearer. In this post, I am going to briefly introduce previous scholars’ opinions on the difference at sentence level, and then present my own investigation of like to say and like saying at discourse level.
There are two viewpoints previous scholars hold. Wood (1956) claimed that the difference between gerunds and infinitives lies in generality. For example, Lying is wrong refers to lying in general, but To lie is wrong refers to a specific event of lying with a specific time or location. Another perspective by Dixon (1995) and Givón (2001) is about whether the event is realized or not. Take this example: John started doing the dishes but then changed his mind. The gerund doing here suggests that the event is realized: John was in the process of doing the dishes, but then changed his mind. But if doing is replaced with to do (John started to do the dishes but then changed his mind), the infinitive – according to this view – signifies that John changed his mind almost immediately after he actually started. However, when these patterns are investigated at discourse level, using corpus data, another picture might emerge.
A Discourse Explanation
To learn about the discourse behaviour of gerunds and infinitives, the Corpus of Contemporary American English (COCA) was used to gather data. The scope of my search was narrowed down to the patterns like to say and like saying, both with lemmatized like (i.e. like in any of its verb forms) in the sense of enjoying. The corpus returned 211 valid examples of like to say and 33 for like saying.
Three features were investigated. The first of these is discourse arrangement. When the verb say is used, there will usually be something to indicate what is said. This information can appear in three possible ways: before, after, or both before and after like to say or like saying. What the corpus data shows is that when the pattern is like to say, the information about what is said occurs after the phrase in around 64% of instances. But when the pattern is like saying, around 72% of examples have the content information (what is said) both before and after the phrase. This suggests that like to say tends to occur with information that is being introduced for the first time (new information), whereas like saying tends to occur with ‘old’ information which has been mentioned already.
The remaining two features relate to syntax. First, the subjects of like to say/like saying. Like to say has a strong preference for having as its subject a personal name (such as Brown or Calvin), rather than a pronoun, with 64% of subjects being named. Conversely, like saying has more pronoun subjects (85%). The final feature is the difference in the object forms of these two patterns. Like to say usually has a clause as its object (79% of instances) , but the most frequent object of like saying is a pronoun (61%). The high percentage of named subjects and clause objects in like to say implies that the saying event appears for the first time and is new to the reader, and information is thus given in its full form. In contrast, the large number of pronouns as subjects and objects of like saying signifies that the saying event is not new in context, and that the content of what is said (and who said it) has already been mentioned in the previous context, thus becoming old information.
As we can see from these results, like to say tends to occur with new information, while like saying typically appears with information that has already been mentioned. These results, based on a discourse approach, are quite different from previous scholars’ perspectives at the sentence level. When teaching English to non-native speakers, teachers may want to present to students how gerunds and infinitives work both at sentence level and at discourse level.
Dixon, R. M. W. (1995). Complement clauses and complementation strategies. In F. R. Palmer (Ed.), Grammar and meaning: Essays in honour of Sir John Lyons ( pp. 175-220). Cambridge: Cambridge University Press.
Givón, T. (2001). Syntax: An Introduction, II. Amsterdam: Benjamins.
Wood, Frederick T. (1956). Gerund versus infinitive. English Language Teaching 11: 11-16.Email this Post
I loooved this post!
I learned this very subtle difference in meaning when I was 12 and I could not understand why on earth most grammar books tend to overlook this significant change in meaning for both combinations. Two very clearly different meanings.
Many kudos on this post.
Keep up with the great job!
What I have inferred after reading the post is that like doing sth may refer to a habit by the subject.
I really like and appreciate your contribution. I am into the same kind of thing and I also made a short exercise exemplifying the difference between the two. Fascinated by corpora work and time permitting would like to do more of it. my exercise was based on a small specialised corpus for students of tourism that I compiled myself for the purpose of my master of science thesis. Patterns mean all for a language and make all the difference to the meaning and students’ awreness of the fact needs to be raised by a lot of examples and practical work that corpus exercises enable.
PS The web I suppied is my faculty’s web and may not mean a lot but the field was required so …