This dissertation examines the processing and mental representation of the sociolinguistic variable ING (thinking~thinkin'). Sociolinguists have asked questions about the locus of the ING variable using naturalistic speech data, which has resulted in a debate on whether the variable is phonological or morphological. These accounts of ING are not well-defined, and it is hard to isolate these representational properties in conversational data. I propose that locus of variation questions can be thought of as questions about the mental representation of variation, and that it would be fruitful to explore them using a highly-controllable tool from psycholinguistic research: primed lexical decision experiments. This tool is used to show that semantic, phonological, and morphological aspects of representation facilitate processing in different ways. I integrate sociolinguistic knowledge of variable ING with psycholinguistic knowledge on researching mental representations to ask: how are the socially meaningful variants -ing and -in' mentally represented, and which aspects of shared representation contribute to how they are processed? Based on a framework of relevant aspects of representation, I establish a baseline understanding of the mental representation of -ing and -in' across six experiments. Chapter 4 shows that -ing and -in' prime both themselves and each other in words with unrelated stems (e.g. jumping-thinking), and uncovers an asymmetrical priming pattern between -ing and -in' targets; -in'-in' prime-target pairs enjoy a processing boost over -ing-ing, -in'-ing, and -ing-in' pairs. Chapter 5 finds that this "-in' boost'" is temporally weak. Chapter 6 establishes that surface phonology does not contribute to the -in' boost. Chapter 7 shows that the -in' boost is insensitive to shared representation between prime-target stems. The results show robust and replicable affix priming for -ing and -in'. They also show a processing difference between the variants, and demonstrate properties of the -in' boost. Taken together, the -in' boost can be interpreted under a representation-based account, which suggests that -ing is the underlying phonological form, and that this can change to -in' via application of a phonological rule. Finally, I propose future avenues of research that test this account and elaborate our understanding of the mental representation of variable ING.