Shared or different: How linked are word production and comprehension?

How people produce and perceive words is one of the core questions in language research, as words are the basis of communication about the world. While it seems simple, producing and comprehending words is actually quite a complex task with multiple steps involved. When thinking about what to say we need to translate an idea into a concrete ‘word’ and tell our vocal cords to produce the word sounds in a particular order. To understand a word, we must take a continuous stream of sound and carve it up into ‘words’ that we understand. Currently, there is a rich source of debate between language researchers on how these different steps are organized, and how similar they are between word production and comprehension. In this article we focus on to what extent word representations are shared between word production and comprehension.Before discussing whether representations are shared, we first broadly state the processing steps involved in each behaviour. In general, words can be broken down into three different linguistic processing levels: the semantic level (i.e., what does the word mean?), the phonological level (i.e., what is the sound form of the word?), and the articulation or sound stream level (i.e., how do we say this, or what word do these sounds relate to?). Of these three levels, there are different ways to approach the question of whether word representations are shared. In the article, we acknowledge the importance of clarifying the terms ‘representation’ and ‘shared’. We argue that defining the first term, ‘representation’, is difficult because of a) the ambiguities in what researchers mean when they use the term ‘representation’, and b) of technical limitations to exactly what we can measure. For some researchers, representation refers to a holistic word representation that has a concrete ‘location’ in the brain, but for others it is constructed ‘on the fly’ during communication (or a mixture of the two). Without sufficient clarity in definition, it can be challenging to test whether representations are shared. We argue that a strict definition of a shared representation instantiated in the brain would be that the same neurons would fire for the same reason in both production and comprehension. Given that we do not have the technology to measure that yet, we rely on other approaches. The approach we take here has three prongs: word representations are shared if they 1) share the same time course of processing between different processing levels, 2) if the same regions of the brain are used for specific word processes in production and comprehension, and 3) if, when carrying out production and comprehension in parallel, there is interference between the tasks which we can locate to different levels. Relating to our first approach, there is mixed existing evidence on whether or not the time course of processing in word production and comprehension is similar. By time course, we mean whether the different levels are accessed simultaneously or in a sequence, for instance whether the semantics of the word are accessible prior to the sounds of the word, or whether all ‘parts’ of a word are accessible at the same time. While some studies have found a pattern which follows sequential activation of these stages (Dufour et al., 2013; Salmelin et al., 1994), more current studies, including by our team, suggest that all levels of word processing are simultaneously active very early on. For example, using EEG we found that lexical-related and phoneme-related information are simultaneously active in word production and comprehension (Fairs et al., 2021). This suggests that the representations underlying word production and comprehension are strongly shared.Relating to our second approach, whether or not the brain regions underlying word production and comprehension are shared, again the evidence is mixed. Some theories argue that different areas of the brain are involved in word production and comprehension, with temporal areas involved in both but frontal areas only involved in production (Hickok & Poeppel, 2007). Other theories argue that the same word representations are used for production and comprehension and hence the same brain regions are involved in both (Pulvermuller, 1999; Strijkers et al., 2017). In a preregistered study which is currently underway, Fairs et al. (2020) aim to address this issue, by testing whether specific areas of the brain already established as being active for certain semantic and phonetic categories are active during both word production and comprehension. This study has the potential to truly shed light on how interconnected production and comprehension are.Relating to our third approach, we review a body of dual-tasking experiments that have tested whether word production and comprehension can be carried out simultaneously, and whether and how they interfere with one another. Fargier & Laganaro (2016; 2019) found word level interference between listening to syllables and producing words, suggesting that the stage of word form encoding is shared between production and comprehension. Fairs and colleagues (2019) found a large amount of processing overlap between word production and comprehension which suggests that not only the word form encoding stage but potentially more stages are shared between the two language behaviours. However, other studies have suggested that the linguistic system can flexibly adapt word production and comprehension processes (Fairs et al., 2018) such that the issue is not that there is overlap between representations, but when this overlap occurs. In the last section of this article, we support the paradigmatic turn to study how words function in context and highlight some of the research avenues. Words are the core basis on which conversations are built, and understanding how people have conversations is the ultimate goal of psycholinguistic research. Current research at a conversational level suggest that there is a large amount of overlap between production and comprehension both within a person (Menenti et al., 2011) and between people (Silbert et al., 2014). We nonetheless conclude that only with more research investigating the foundational blocks of language processing – words - will we truly understand how production and comprehension work and how shared word representations and processes are.