On initial state, initial stages, other fantastic beasts, and where to find them


UC Berkeley – April 9, 2022

Keywords: initial state, initial stages, second language acquisition, third language acquisition, falsifiability

The concept of initial state is crucial for first language acquisition, but it has been adopted by second and third language acquisition fields as well (e.g., Rothman 2011, 2015, Rothman and Cabrelli Amaro 2010). Recently, it was amended to “initial stages” (Rothman et al. 2019). The purpose of this paper is to argue that (a) initial state for nLA is not directly related to the initial state for L1 and actually is a conflation of two distinct notions, and that (b) in the way most current literature uses the notion of “initial stages”, it is not viable as a concept per the falsifiability considerations.

A: initial state for nLA does not share much with the similar concept in first language acquisition
Chomsky (1980:69) wrote that “[UG] is taken to be the set of properties, conditions, or whatever that constitute the “initial state” of the language learner, hence the basis on which knowledge of language develops”. That is, the theory of UG is the theory of the initial state for first language acquisition. However, in SLA/TLA, “initial state” (S0) came to be a conflation of two other notions:

The two are perhaps somewhat related, but unmistakably distinct. It is also not immediately clear which of (a) and (b) is related to Chomsky’s initial state. Perhaps, neither:

Chomskian tradition (see, e.g., Flynn 1985, 1988, 2021) finds the very idea that the UG is not accessible affront to minimalism, as it is stipulating that the language being acquired is some non-human language. Chomsky’s genetic endowment (UG) is stipulated as necessary for acquisition of any human language. Chomsky’s “initial state” is UG, not its accessibility from (a) since, trivially, UG remains unaltered quantitatively or qualitatively throughout lifespan (save exceptional cases).

B: initial stages is not viable as a concept per the falsifiability considerations
Recently, trying to cope with the failure of “initial state” for L2 to gain empirical or theoretical support, Rothman et al. (2019) introduced the concept of “initial stages”. It was implicit in Chomsky (1981) that S0 is a state, not a process, and therefore is not durative, in the sense that it does not denote action or alterations. Similarly, the discussion above dealt with a concept of state being stative (as opposed to durative). Extending “initial state” to “initial stages” (Sd, d for durative) in Rothman et al. (2019) as an ad hoc build-on to move from indefensibility of the S0 appears to be just as infertile. The concept of Sd presupposes that there is some specific period of time when the state of grammar-lexicon has some unique properties, and such properties are only observable during that period. Otherwise, the period is unmotivated. Yet, none such properties are observably manifested (e.g., in the form of a “sensitive period” perhaps). Crucially, Stringer (2021) argued that these “initial stages”: “[give] the TPM [Typological Primacy Model of Rothman et al. (2019) – author] wiggle-room to provide alternative accounts of any counterevidence of property-by-property transfer. Citing the need for ‘cognitive economy’ to explain wholesale transfer just for those stages of acquisition after the initial stage but prior to more advanced acquisition falls short of actual evidence against property-by-property transfer throughout the L3 acquisition process”.

Concluding, we hope to have scrutinized the notion of “initial state” and “initial stages” in Ln acquisition, illuminating some elements of these notions that are – to our view – inadequate for generative inquiry. We hope to have shown that the concepts of “initial state” and “initial stages” for Ln acquisition are indeed magical beasts which ought not be pursued in generative inquiry. Rothman and colleagues call for empirical evidence against “initial stages”, yet no such evidence could be provided since the very concept is unfalsifiable. It is a convenient position, yet a position which is affront to generative acquisitionist inquiry.


(please note that the references are not meant to be comprehensive)