Approaches‎ > ‎DATA DRIVEN LEARNING‎ > ‎

Comparing similar words: answer vs. respond

Explaining the differences between a pair of "synonyms" depends on your view of language. When asked what the difference between two words is, we grope for subtle differences in the meaning. If we can't find any or we aren't able to express it, we might use the UK-US usage justification, or opt for register: formal vs. informal. 

The unstated starting point is what the two words have in common, namely, following up a question.

Using corpora to answer such a question assumes Firth's view that we know a word by the company it keeps, as corpora provide data about the words being studied. Most corpus tools provide this data in the form of collocation lists which consists of words and numbers (frequency counts and significance scores). Other tools structure these lists according to the grammatical (syntactic) relationship between the target word and its collocates, for example, Just the Word, Stringnet Navigator and Sketch Engine.

Sketch Engine is named after its word sketches which are tables of collocates and numbers presented in columns of grammatical relationships. See this example of answer in the Brown corpus.

To compare words such as answer and respond, Sketch Engine has a related tool that presents word sketches of two words at the same time. The following examples are extracts of Sketch Differences using the British National Corpus. The words in green relate to respond and those in red to answer.

In this first extract, we see the different subjects of the two verbs. Cells, governments and bodies respond, while people and voices answerThe zeros are also of interest: they tell us that these words are never used as the subjects of their respective verbs, at least not in this corpus.


In the next two examples, we see their objects. Somebody or something answers an advertisement, knock, query, phone, etc. Interestingly, there is no green in the first one, indicating that there are no objects of respond.


Respond's objects come after to. We can now compare samples of what is answered and what is responded to. 
The data tells us that someone or something responds to a call, need, treatment, demand etc.


Although there are many more grammatical relationships in a Sketch Difference, one more example is given here, namely the and/or relationship, i.e. words co-occur in normal text joined by and or or often enough for this to be a significant part of knowing a word's company. Note again the zeros.


Sentence examples are obtained by clicking on the frequency numbers in the tables. Here are two of each.
  • Is it how we respond and react to old melodies and memories from the past?
  • How have people reacted and responded to your treatment and really erm how they feel about things now?
  • There are many such questions to be asked and answered.
  • How long this can and should, go on is the question which has to be asked, and answered.
Moving away from word data, another type of data available is text types. Certain grammatical structures, words and phrases occur more significantly in one type than another. This information informs our language choices when we are speaking and writing.

Here are the top nine text types for respond. The W at the beginning of each row indicates written language. None of the top nine here are from spoken language, which doesn't occur until the 17th row. Note that spoken language transcripts make up only 10% of the BNC.

Comparing the top items, respond occurs in fiction in the BNC 900 times and answer occurs 3,666 times.


Here are the top 15 text types for answer, where spoken language appears in 13th and 14th rows.


Conclusions
The question about the differences between one word and another can be answered by referring to the various companies that they keep, and by the text types that they are used in. 

However, the data complicates the idea that respond and answer are similar in that they follow up a question. While the data shows that they do, it also shows that the words work in other ways as well, e.g. respond to treatment and answer an advertisement. And this partly accounts for the difficulty people have explaining language – too many examples muddy our intuition. 

The important point to recognise here is that a word's meaning or reference changes in different company. The term meaning potential expresses the idea that words have a number of potential meanings which are not realised until they are used in context.  

When a language learner explores the range of words that are found as a verb's subject and object, they might be surprised and delighted to discover that answer and respond are used in such various ways, and especially if they didn't know that we say, respond to a challenge, answer to a name. Thus the benefits of investing so much time into answering a straightforward question include the serendipitous discoveries along the way, even if appears that we are using a sledgehammer to crack a nut.

It is worth bearing in mind the reason for asking the question. In practical circumstances, it relates to making the most appropriate choice when speaking and writing. The question might also arise when wondering if English actually needs both. Or when comparing the Germanic and Romance sources of English vocabulary, and answer/respond is a good example of this.

Finally, it is worth acknowledging that such linguistic processing is not for everyone. The process of studying language data is an empirical, analytical approach with which most people working in linguistics today are quite comfortable. However, there are also many people who come to foreign language study and teaching via the more aesthetic and holistic route that is literature. Bridging linguistics and literature is Stylistics, the branch of linguistics that studies literature linguistically.

Here are some pairs of words that are obviously similar. Some of them are worth investigating in the ways described above, but not all. Different questions demand different treatments.

iceberg – glacier
bicycle – tricycle
drawing – picture
direct – blunt
excellent – outstanding
leap – hop
data – information
knowledge – wisdom
ceiling – roof