People greatly differ in the size of their social circle. In general, interacting with more people should lead one to receive more variable linguistic input. Variability has been shown to facilitate learning of new phonological categori es (e.g., Bradlow & Bent, 2009). Variability, however, might also be beneficial at other linguistic levels. Additionally, it may boost performance even in one's native language. We ex amined whether having a larger social circle improves individuals' lexical and sema ntic skills in their native language. In Study 1, we tested whether individuals' social circle size influences their linguistic skills at the lexical and semantic levels. In Study 2 we replicated the results of Study 1 using an experimental manipulation of social circle si ze, thus showing the causality of this effect. Study 1 tested the influence of social circle size on lexical and semantic skills using a lexical prediction task. We recruited 226 part icipants and asked them with how many people they interact in a typical week. We then pr esented them with a forced choice sentence completion task, and asked them to select the most common way that others would complete the sentence. The response choices were based on common responses in norms for these sentences (Lahar, Tun & Wingfield, 2004) and then further normed in a multiple choice format (N=70) to verify the dominant response. There were two types of items: (1) Lexical items, in which responses are synonymous in the context, e.g., She calls her husband at his ____ (a) job (b) office (c) work (d) workplace, and (2) semantic items, in which responses differ in meaning, e.g., Few nations are now ruled by a ____ (a) dictator (b) ki ng (c) president (d) woman. Results revealed an interaction between so cial circle size and linguistic level, such that larger social circle size predicted higher accuracy on the semantic items, but not on the lexical items. Study 2 used an experimental ma nipulation of social circle size. First, we elicited short reviews of chairs from 8 speakers. We then replaced the words horrible, bad, ok, good, and great in these reviews with 5 novel words (e.g., noral ). In total, we had a set of 160 reviews (4 reviews x speaker x rating level). We manipulated social circles size by assigning each participant (N=76) to one of two sampling conditions: receiving all the reviews from 2 randomly selected reviewers, or receiving 5 randomly selected reviews from each of the 8 reviewers (1 per rating level). In both cases, participants we re exposed to 40 reviews in total. Each review had an equal probability of appearing in each sampling condition. Reviews appeared with a cartoon that represented the reviewer, so participants could track the reviewer's identity. After this exposure stage, we tested participants on their semantic comprehension of new reviews with these novel words, and on their lexical choi ce prediction for these words (when meaning is held constant). Both tests used new reviews from new reviewers. Results replicated Study 1, showing that those in the large social circle condition (8 reviewers) did better than those in the small circle condition (2 review ers) on the semantic task, but worse on the lexical task. These studies show that individuals' social circle size influences their linguistic skills. Specifically, having a larger social circle improves semantic, but not lexical, skills. We hypothesize that the differential effect of social ci rcle size is due to properties of the linguistic level, such as the number of competitors and the ratio of intra- to inter-individual variability