*Introduction*: Language games (ludlings) involve systematic manipulation of the phonology of words; generally the purpose of a ludling is to provide a ‘secret language’ for a group of initiated speakers (Bagemihl 1988). As such, ludlings must strike a balance between comprehensibility and phonological distortion in order to provide a feasible communication system for the initiated while preventing the uninitiated from understanding. This property allows us to use ludlings to study how severely various phonological transformations impede word recognition. Reduplicative ludlings are the most common type of ludling cross-linguistically (McCarthy 1991). They involve inserting material in a word, some of which is copied from that word, e.g. ‘zombie’ might become [za *za*mbi]. *We ask whether ludlings that reduplicate onsets disrupt word recognition more than those that reduplicate nuclei*. We hypothesize that the former will be harder to comprehend than the latter. Two things motivate this: First, onset-reduplicating ludlings are typologically unattested while nucleus-reduplicating ones are common (McCarthy 1991, Nespor et al. 2003). Second, there has been much work suggesting that consonants may be more important than vowels for word recognition (e.g. van Ooijen 1996 et seq; see Nazzi & Cutler 2019 for review). If this is the case, we may expect the superfluous consonants in an onset-reduplicating game to derail lexical access more than the superfluous vowels in a nucleus-reduplicating one. This prediction is explicitly made by Nespor et al. 2003 but has not been tested. *Method*: We invented 3 ludlings that inserted a C(C)V sequence after the first nucleus of the word (see Table 1). In the *onset ludling*, the V was fixed and the C(C) was a copy of the preceding onset; in the *nucleus ludling*, the C was fixed and the V was a copy of the preceding nucleus. [f] and [u] were the default onset and nucleus. A fixed *syllable ludling* in which ‘foo’ [fu] was inserted served as a control. Materials comprised 72 English words. We adapted the ludling-learning task used by Nevins & Endress (2007). Participants were assigned to learn one of the three test ludlings and completed 3 phases with 24 trials each: *Familiarization*: Participants passively listened to pairings of words and their encoded counterparts (e.g. [zambi]… [zazumbi] for ‘zombie’ in the onset ludling). *Recognition*: They heard a word followed by that word encoded in either the target ludling or one of the two unfamiliar ludlings. Participants rated the encoded form as correct or incorrect. *Comprehension*: Participants heard a word encoded in their ludling and tried to identify it. 96 monolingual English speakers participated. Syllable Nucleus Onset *chrome* kro*fu*m kro*fo*m kro*kru*m *zombie* za*fu*mbi za*fa*mbi za*zu*mbi *plastic* plæ*fu*stɪk plæ*fæ*stɪk plæ*plu*stɪk *Table 1*: Example words encoded in the three language games. *Results*: Accuracy in the recognition and comprehension tests were analyzed by fitting generalized linear mixed effects models using the lme4 package in R. Fixed factors included the type of ludling learned, number of syllables, and word frequency. No comparisons were significant in the recognition phase. In the comprehension phase there was a significant effect of target Ludling at the nucleus and onset levels (see Table 1): Participants who learned the nucleus ludling were better at decoding words in their ludling than were participants who learned the syllable ludling. Conversely, participants who learned the onset ludling performed worse than those who learned the syllable one (& consequently those who learned the nucleus one as well). *Discussion*: This study indicates that manipulating onsets disrupts word recognition more than manipulating nuclei, which is consistent with psycholinguistic literature arguing that consonants are more important for word recognition than vowels. It also indicates that the lack of an attested, naturally-occurring onset-reduplicating ludling is due to this functional asymmetry. Finally, it highlights two open questions: What mechanisms drive the asymmetry between consonants and vowels, and how do we implement it in models of word recognition? *References* Bagemihl, B. (1988). *Alternate phonologies and morphologies*. Doctoral dissertation, University of British Columbia. McCarthy, J.J. & Plénat, M. (1991). L'infixation réduplicative dans les langages secrets. *Langages* 101:11-29. Nazzi, T., & Cutler, A. (2019). How consonants and vowels shape spoken-language recognition. *Annual Review of Linguistics*, *5*, 25-47. Nespor, M., Peña, M., & Mehler, J. (2003). On the different roles of vowels and consonants in speech processing and language acquisition. *Lingue e linguaggio*, *2*(2), 203-230. Nevins, A. & Endress, A. 2007. The edge of order: Analytic bias in ludlings. Harvard Working Papers in Linguistics 12:43-53. Van Ooijen, B. (1996). Vowel mutability and lexical selection in English: Evidence from a word reconstruction task. *Memory & Cognition*, *24*(5), 573-583.