The Interplay of Synonymy and Polysemy

15 | 
Murphy (2003) 
Since Pustejovsky (1995) and Jackendoff (1990) do not fully discuss synonymy, the 
concepts proposed by Murphy (2003) provide an explicit way to judge synonymy within their 
systems. Murphy (2003) argues that synonymy (and other semantic relations) is not the type 
of information that is stored in the lexicon. One reason for arguing that synonymy is not 
stored in the lexicon is that synonymy can vary depending on the context. This is especially 
true since words are often polysemous, or at least have slightly different senses depending on 
the linguistic context (2003:29-30). For example, 
good
and 
bad
are antonyms, but this is not 
true in all their senses. 
Bad
in informal style can mean something that is good. In that specific 
case, 
good
and 
bad 
are in fact synonyms. A word can have several different semantic 
relations to another, depending on the sense under consideration (2003:36). 
It is also possible for a semantic relation to change, even when there is no obvious 
change in sense. That is, in one usage event words may be synonyms, while in another they 
are antonyms, homonyms, etc. Imagine two speakers discussing a group of people. 
(4) 
A: That’s quite the mob. 
B: That’s not a mob, just a crowd. 
A: Same thing. 
From A’s perspective the terms 
mob
and 
crowd
are close enough in meaning to be 
considered synonyms. This is not the case for B, who is instead highlighting their differences 
and not treating them as synonyms.
Another important characteristic of semantic relations is that they are productive: new 
relations can be derived (2003:27). When new slang words are created then new semantic 
relations are also created. For example, when 
ride
is used to mean ‘automobile’ the word also 
immediately has new semantic relations; the word will be a synonym with previously known 
words such as 
car, auto, 
etc. (2003:27).
Murphy uses the above facts to argue that semantic relations such as synonymy can be 
generated by using a principle. Her principle is called Relation by Contrast (RC) (2003:44). 
Her RC for synonymy states that synonyms are words that have all the same properties 
relevant for a given context and only differ in form (2003:134). For example, for speaker A, 
in the given context 
mob
and 
crowd
are equal in all relevant aspects, only being different 
phonologically. For this speaker, they are synonyms.
Murphy’s (2003) principle means that judgments on synonymy could vary across 
senses of a word, across usage events, and even across speakers. Her view allows for 

---

16 | 
creativity and productivity in synonymic relations. She does acknowledge that it is not 
necessary for semantic relations to always be generated. That is, sometimes a relation is 
common enough or salient enough to exist in memory. It is possible for synonymy to be part 
of our memory, but still she argues that this is non-linguistic knowledge and is not a part of 
the lexicon (2003:42).
The concept of a relation by contrast can be adopted into both Pustejovsky (1995) and 
Jackendoff (1990). In both of their theoretical models synonymy is judged by comparison, 
but they have little to say about how to compare two words. Murphy (2003) describes a 
principle that explicitly states the method of comparison that can be used to judge 
synonymy
1
.