Kate Bellamy

Of the archaeological evidence suggesting prolonged, meaningful interaction between the Andean region in South America and West Mexico, metallurgy appears the most convincing. Metalworking technology and knowledge seems to have been transferred from the former, and probably also from Colombia, via a mainly maritime route starting around 600CE. Similarly, a number of linguistic studies (notably Swadesh, 1967) have claimed a deep relationship between Quechua in the Andes and Purépecha in West Mexico, although the lexical data provide insufficient evidence for this grouping to have gained general acceptance. Nonetheless the archaeological record suggests that this is a link worth pursuing further.
In the absence of documentation dating from the pre-Columbian (pre-1500CE) period in both regions, we must turn to alternative sources to try and uncover the linguistic and migratory past of these complex societies. Through a comparative study of the lexical domain of metallurgy, its materials, processes and tools, I am investigating whether the proposed contact can be further clarified. My study covers languages spoken in the so-called ‘West Mexican Metalworking Zone’ (Hosler, 2009) and the known metalworking regions in Central and South America, but with a focus on the language isolate Purépecha, the clearest alleged recipient of the Andean technology.
References
Hosler, Dorothy. 2009. The Metallurgy of West Mexico: Revisited and revised, Journal of World Prehistory, 22: 185–212.
Swadesh, Morris. 1967. Lexicostatistic Classification, in Robert Wauchope and Norman A. McQuown (eds.), The Handbook of Middle American Indians, Vol. 5: Linguistics, Austin: University of Texas Press.

Objects in the physical world occur in different quantities and configurations. To determine how many objects are present in an array humans make use of two distinct cognitive systems: analogue and digital numeracy (Feigenson et al., 2004). Analogue numeracy constitutes the ability to estimate an imprecise amount, while digital numeracy comprises the exact assessment of an amount, either by subitisation, the seemingly effortless perception of up to four items, or calculation for larger amounts. Previous research (e.g. Nan et al., 2006) has largely dissociated analogue numeracy from language, assuming it to be pre-linguistic, constituting simply the ability to discriminate the larger of two amounts. However, the verbal expression of estimation (rather than discrimination) is necessarily through the numeral systems available in language.
These numeral systems vary cross-linguistically, in terms of their base, specific dimensionality, extent and morphosyntactic structure (Bender & Beller, 2011). Since language has been demonstrated to influence the perception of other domains, such as colour, event perception, gender, object individuation (Frank, 2011) and navigation and spatial relations (Levinson & Wilkins, 2006), I decided to investigate whether variation between languages can also affect analogue numeracy. Framed in neo-Whorfian terms, I tested empirically whether a specific, but cross-linguistically varying element of linguistic structure – the counting system base – affects how speakers of different languages estimate quantities. 
In a novel, limited exposure experiment, British English (base 10; n=26) and Georgian (base 20; n=26) speakers were briefly presented (1.5 seconds) with 50 slides containing numerosities from 1-400 and asked to state (within 5 seconds) how many objects they had seen. Supporting the principal hypothesis, response type and language were moderately related variables, Pearson χ² (7, N = 2600) = 231.396, p (asymptotic) = .000, Cramer’s V = .298. Both participant groups favoured multiples and powers of their numeral bases as estimation responses, in 63.7% of all cases (n = 962) for the English group and 44.54% (n = 962) for the Georgian group. Both groups also showed a secondary preference for ‘other’ responses (i.e. not related to the base), the tendency being much higher in Georgian (31.81% of responses) than in English (16.63%). The weaker preference for estimating in terms of the base and its multiples among the Georgian group may relate to the strong cultural and linguistic influence of Russian, a base ten language. Over  80% of the Georgian participants spoke Russian to at least an intermediate level, whereas the English participants demonstrated far less bilingualism or L2 proficiency. This difference also raises questions regarding the mental representation of number and counting systems in bilinguals. In sum, these findings indicate that participants from both groups make use of the same cognitive process (analogue numeracy, or estimation) when presented with amounts that cannot be readily subitised or calculated but also suggest that a language’s numeral base is associated with its speakers’ estimation preferences.