Compound words are funny creatures. They exist as words themselves, but also contain smaller words which have their own unique meanings.  For example, the compound word butterfly refers to the pretty insect with wings, but it also contains the smaller words butter (yellow spread made from cream) and fly (move through the air).

But how do we process compound words? Do we retrieve them as one whole word or are we also affected by the smaller words inside? When we read butterfly, do we only think of the insect, or are we also affected by our knowledge of butter and flying?

In a recent study published in the Psychonomic Society journal Memory & Cognition, Mahmoud Elsherif, Jon Catling, and Steven Frisson (pictured below) examined how people process compound words. In particular, they examined how quickly people could read out loud various compound words.

There are lots of variables that affect how quickly people can read a word out loud – its length, how common that word is in the language, and even how difficult the sounds are to pronounce. However, one somewhat surprising influence is when the word is typically learned by children. Words learned early in development (bed, foot) are produced faster than words learned later (gale, birch). These age of acquisition effects occur even after controlling for other relevant factors such as word frequency and length.

For example, the words “animal” and “career” are used equally often in English, but “animal” is commonly learned at a younger age. When asked to read words out loud, adults will produce “animal” faster than “career” even though they are matched on word frequency and length.

Across two studies, Elsherif and colleagues examined how the age of acquisition affects word naming speed for compound words. When we read a compound word like “butterfly,” are we affected only by when we learned that word, or also by when we learned the component words?

In the first experiment, 48 British undergraduates were asked to read 236 compound words out loud as quickly as possible, without sacrificing accuracy. Naming speed was affected by when each word is commonly learned, even after controlling for variables such as the word length, word frequency, and the initial sound. People were faster to produce words that are commonly learned earlier in development.

However, naming speed was not affected by the individual words within the compound word. For example, “lunchtime”, a word learned early in development, would be pronounced faster than “bankroll”, a word learned much later. But naming speed was not affected by when children learn the words “bank”, “roll”, “lunch” and “time”.

The second experiment asked people to create their own compound words by combining two presented words. For example, participants would see “air plane” and say out loud “airplane”. Forty-eight new British undergraduates again went through the full list of 236 compound words.

Participants were just as fast to name the word in this new format. But, their naming speed was now affected by both the age of acquisition of the compound word and the age of acquisition of the internal word. When the words were visually separated, naming speed was affected by when children typically learn the whole word “airplane” and when they learn the two smaller words “air” and “plane.”

Overall, the results suggest that we naturally process compound words as a single entity. But, when visually separated, we recognize each of the internal words along with the compound.

This unitary perception often makes sense. It would likely be inefficient and confusing if we fully processed each of the smaller words within each compound word. As the authors mention, the meaning of a compound word often differs drastically from the meaning of each of the smaller words put together. For example, a “firefly” is an insect, but “fire fly” suggests that one is firing a fly from a gun or cannon.

Compound words are funny creatures, but, it turns out we may process them similarly to other more typical words.

The Psychonomic Society article considered in this post:

Elsherif, M. M., Catling, J. C. & Frisson, S. (2020). Two words as one: A multi-naming investigation of the age-of-acquisition effect in compound-word processing. Memory & Cognition, 511–525. https://doi.org/10.3758/s13421-019-00986-6