Imagine I do a priming experiment with the two types of word pairs, as shown in the example below:
- Related: KICKED ... KICK
- Unrelated: PURPLE ... KICK
In which condition do you think people will respond faster to the target word KICK?
Based on what we have learned so far, you should be able to predict that people respond to KICK faster in the related condition (when the prime is KICKED) than they do in the unrelated condition (when the prime is PURPLE). That's the priming effect.
Now, in what way do you think KICKED and KICK are related?
Morphological relationships
KICK and KICKED have a special relationship. A pair of words like BUTTERFLY and BEE has related meanings, but they obviously are different words. With KICK and KICKED, however, you might have the feeling that they're the same word, just with some different grammar suffixes attached: KICKED is the past tense of KICK. Another example is KICKS (the verb); there is absolutely no difference between the meaning of KICK and the meaning of KICKS, English just has a stupid grammar role saying that we have to say "I kick" but "He kicks".
So the important point is, KICK and KICKED are related in a special way, such that they feel like they're "the same word" in some way. We need some formal concept, though, to explain what's special about this relationship.
The concept we can use is the morpheme.
What is a morpheme?
A morpheme is the smallest unit in a language that has meaning on its own. This definition has two parts: a morpheme has to have meaning on its own, and it has to be the smallest possible thing.
To have "meaning on its own" (we can also call this an independent meaning), something must have some meaning that you can explain or describe. A word like cat has independent meaning: even without any context, you can explain what cat means (you can find something in the world that is a cat, and you can point it out; or, you can verbally describe what makes something a cat). Run also has independent meaning: you can explain it, you can demonstrate it, etc. 開 has independent meaning, and 車 has independent meaning. But things that are less than a word can also have independent meaning! In the word cats, the -s has independent meaning, because you can explain what it means all by itself ("-s is a suffix that makes a word plural; cat means one cat, but if you put an -s on the end then it means more than one cat"). The -er in runner also has independent meaning (can you think of how you would describe what it means?). And thinking of Chinese, there are things that aren't considered a word but do have independent meaning: for example, 們 (how would you explain what 們 means?). Some things do not have independent meaning. For example, I can't define what the t in cat means; it doesn't mean anything. Nor can I explain what the s in bus means; the word "bus" does not mean "more than one bu" (unlike "cats", which does mean "more than one cat").
The other part of the definition is that a morpheme is the smallest unit that has independent meaning. That means it can't be broken down into parts. So, for example, teacher is not a morpheme, because I could break it down into teach + er. I could define what teacher means, so it does have independent meaning, but it's not a smallest unit of independent meaning because I can break it into smaller meaningful parts. Likewise, the sentence Yesterday I went to the shopping centre is not a morpheme, because I can break it down into many parts.
Can you think of some more examples of morphemes, and examples of things that are not morphemes?
Morphological priming
Now let's return to our example of KICKED and KICK. We now have a way we can explain their relationship: these words are related because they share the same morpheme. KICK is one morpheme (we can define what it means, and we can't break it down into smaller pieces). KICKED can be broken down into two morphemes: KICK, plus the past tense morpheme -ED. Therefore, both of these words include the morpheme kick. We say they are morphologically related—i.e., they're related because of a shared morpheme.
In this sense, these words are more closely related than BUTTERFLY and BEE are. Thinking about the word networks we discussed in activity #3, BUTTERFLY and BEE are different words that are linked (or connected) in the network. On the other hand, KICK and KICKED actually include the same word (or, technically, the same morpheme) in them.
We can imagine that the vocabulary in your mind is organized as shown in the image below. In your mind you have a storage of all the morphemes you know. So morphemes like kick, as well as suffixes like -ed, are all stored in your mind. Complex words like kicked or kicking or kicker don't need to be stored in your mind. If you hear the word kicked, you can understand it by breaking it down into its parts (kick and -ed), activating both of those morphemes in your mind, and figuring out what it means when they go together.
In a priming experiment, you first see the prime KICKED, and when you see that prime you activate the morpheme -ed and the morpheme kick. A moment later, you see the target KICK. You had already just activated the morpheme kick when you saw the prime; now that you see KICKED, it should be super easy to activate again. Therefore, morphologically related pairs of words generally cause a huge priming effect (the target with a related prime is generally processed much faster than the target with an unrelated prime).
An alternative view
But wait a minute: do we really need these complicated abstract concepts? Maybe "morpheme" is just some abstract thing linguists made up and talk about in linguistics classes. Normal speakers, using language in a natural way, might not have any idea of this sort of thing. Can we explain this priming effect more simply, without assuming there are abstract concepts like "morphemes"?
There is indeed another proposal to explain why morphological priming happens without assuming that "morphemes" have any special status in our minds. Maybe we don't store abstract "morphemes" in our minds; maybe we just store all the words we know. If that is the case, the storage or words in our mind might look like the image below:
Here, the person's mind includes the words kick, kicked, and kicker; there's no need to store abstract morphemes like -ed.
Then how can we explain why KICKED...KICK gets such a big priming effect (i.e., why KICKED...KICK is responded to so much faster than PURPLE...KICK)? Easy! Word pairs with related meanings (like BUTTERFLY...BEE) are only related in terms of their meaning. Word pairs like KICKED...KICK are related in two ways: they have closely related meaning, but they also have closely related spellings (or closely related pronunciations, if we're speaking/listening). You could also argue that the meanings of KICK and KICKED are more closely related than the meanings of BUTTERFLY and BEE are.
In other words, according to this view, words that appear to be "morphologically" related are not actually morphologically related; they just have a close meaning relationship plus a close spelling relationship. That combination of relationships creates what we call a "morphological relationship", but actually it's nothing special; it's just meaning plus sound. On a strong interpretation of this view, you might even argue that "morphemes" are not a real thing; the whole idea that there is some morpheme like -ed, shared by many words, is actually just an illusion caused by the fact that many words have the spelling -ed in them and also have a meaning related to past tense.
This raises the question, then: are morphemes even real? Can words really be "morphologically related"? We've seen one perspective which believes morphemes are real, and one perspective which believes they are not real. How can we decide?
Using reaction-time methods to test whether morphemes are a real thing
Here we have a question about linguistics (a question about how our language—in this case, our words—is organized in our minds). But we can use psycholinguistic techniques, such as reaction time measurement, to try to solve it.
There have been a lot of priming experiments attempting to examine whether morphemes are a real thing. In general there are two approaches this research can take.
- First, some experiments attempt to see if morphological priming is bigger than meaning priming plus spelling (or pronunciation) priming. Imagine that the morphological priming effect in some experiment is 60 milliseconds (meaning that people respond to KICKED...KICK 60 milliseconds faster than they respond to PURPLE...KICK). But maybe in the same experiment, the meaning priming effect is 30 milliseconds and the spelling priming effect is also 30 milliseconds. If so, that would be consistent with the idea that morphemes aren't special: the "morpheme" priming effect is just as big as you'd expect based on having a meaning relationship plus a spelling relationship! Or, what if the meaning priming effect is 20 milliseconds and the spelling priming effect is 5 milliseconds, and the morphological priming effect is still 60 milliseconds? In an experiment like that, the meaning plus spelling relationship would only lead to 25 milliseconds of priming, but the morphological relationship caused more than 25 milliseconds of priming; if the results came out like that, this would suggest that there really is something special about morphological priming, and thus morphemes might be a real thing.
- Another approach to this research is to see if morphological priming occurs in conditions where we would not expect meaning priming at all, or conditions where we would not expect spelling priming at all. Think of English words like IS and ARE. These words have very closely related meanings (they're really just different forms of the same word, BE) but they are spelled completely differently from each other, and pronounced completely differently. If "morphemes" are really a thing, we might expect a priming effect for this word pair, since they are two different versions of the same morpheme. On the other hand, if there's no such thing as morphemes, and all that matters is meaning relationship and spelling/pronunciation relationship, then maybe these won't have an especially big priming effect, because they have related meaning but not related spelling. Or think of English words like HONEYMOON and HONEY. These words have a spelling/pronunciation relationship (both have honey in them), but the meanings are almost totally unrelated. If morphemes are a real thing, we'd say these both have the morpheme honey in them, so they should have a big priming effect. On the other hand, if morphemes don't exist, then we might not expect much priming effect for this word pair, since they have a spelling relationship but no meaning relationship.
In the class discussion session, introduce the students to this debate and the two possible approaches to testing it via morphological priming. Then have the students, in small groups, examine evidence from different experiments.
Below I have posted links to several experiments that test this question. Some of them support the morpheme idea and some support the whole-word idea. Have students break into small groups and have each group browse one of these papers (they can browse more if you want, but they must browse at least one) so they can understand what the researchers did to test this question, what results they found, and which theory their results support. You could assign a paper to each group (you could make every group discuss the same paper, or make different groups discuss different papers) or let them choose on their own. Make sure to check in with each group (or, alternatively, have everyone come back together for a wrap-up at the end of the discussion) to make sure they have identified a conclusion from their paper.
Of course, 15-30 minutes is not enough time for people to read and understand a whole paper. But if you have appropriately explained the debate and the logic behind this kind of test, students should be able to figure out the idea of any one of these papers fairly quickly, without needing to read the whole thing. They basically just need to see what kinds of priming conditions were included in the experiment (most papers include a table showing this in the Methods section) and check the results to see which view the results support. Make sure you explain this strategy to the students.
- Coughlin, C. E., & Tremblay, A. (2015). Morphological decomposition in native and non-native French speakers. Bilingualism: Language and Cognition, 18, 524–542. (You only need to worry about the results for native speakers)
- Coughlin, C., Fiorentino, R., & Spinelli, E. (2015). Morphological processing of regular verbs in native French speakers. Kansas Working Papers in Linguistics, 36, 34–58. Kansas Working Papers in Linguistics.
- Rastle, K., Davis, M., & New, B. (2004). The broth in my brother's brothel: morpho-orthographic segmentation in visual word recognition. Psychonomic Bulletin & Review, 11, 1090-1098.
- Fiorentino, R. & Fund-Reznicek, E. (2009). Masked morphological priming of compound constituents. The Mental Lexicon, 4, 159-193. (You only need to worry about Experiment 1)
- Longtin, C., Segui, J., & Hallé, P. (2003). Morphological priming without morphological relationship. Language and Cognitive Processes, 18, 313–334. (You can choose to browse just Experiment 1, or to browse just Experiment 2.)
- Gonnerman, L., Seidenberg, M., & Andersen, E. (2007). Graded semantic and phonological similarity effects in priming: Evidence for a distributed connectionist approach to morphology. Journal of Experimental Psychology: General, 136, 323–345. (You can choose to browse just Experiment 1, or browse just Experiment 4.)