Monday, November 17, 2014

The First Hilbert Question: Adriana Belletti

This is the first of our Hilbert-Question (HQ) posts. Adriana Belletti kicks the series off with the following query:
How much of detected developmental stages can be attributed to the internal complexity of formal/grammatical factors and how much to general mechanisms? Are there really general mechanisms?
There are structures and constructions that appear to be particularly hard in acquisition (Crain & Thornton 1998, Guasti 2002, 2007 for a background on issues in theoretically oriented language acquisition studies). Typically, this is the case cross-linguistically, in domains in which constructions in different languages are closely comparable (e.g relative clauses, passive …). For instance, it has been claimed that Passive is delayed in several languages and this has been attributed to late development of the different aspects of the formal computation involved (A-chain, Borer & Wexler 1987; by-phrase Fox & Grodzinsky 1998, smuggling à la Collins 2005 as in Hyams and Snyder 2005). However, recent and less recent (e.g. Demuth 2010, Crain et al. 1987/2009) contributions indicate that under appropriate conditions – strictly formal, discourse related and possibly linked to the richness in the input – passive is not that hard for even young children, across languages.  Similarly, in the acquisition of complex object relatives/A’-dependencies recent evidence has shown that not all types of object A’-dependencies are hard for young children. Appeal has been made to the principle of locality regulating intervention (Rizzi 1990, 2004) to account for the selected difficulty (Friedmann, Belletti, Rizzi (2009). This leads to the following more general query: do general mechanisms ever play any comparable role in providing refined explanations of these developmental milestones?


Borer, H. & K. Wexler (1987) “Maturation in Syntax”, in Parameter Setting, T.Roeper     E.Williams eds. Reidel, 123-172
Crain, S. & R. Thornton (1998) Investigations in Universal Grammar, MIT Press
Crain, S., R. Thornton, K. Murasugi (1987/2009) “Capturing the evasive passive”, Language        Acquisition, 16.2, 123-133
Collins, C. (2005) “A Smuggling Approach to the Passive in English”, Syntax, 8.2, 81-120
Demuth, K., F. Moloi, M.Machobane (2010) “Three year-olds’ comprehension, production and generalization of Sesotho passive”, Cognition, 115.2, 238-251
Fox, D. & Y.Grodzinsky (1998)  “Children’s passive: A view from the by-phrase”, Linguistic      Inquiry, 29: 311-332
Friedmann, N., Belletti, A., Rizzi, L., 2009. Relativized relatives: Types of intervention in the acquisition           of A-bar dependencies. Lingua 119, 67–88
Guasti, M.T. (2002) Language Acquisition, MIT Press
Guasti, M.T. (2007) L’acquisizione del linguaggio, Raffaello Cortina Editore
Hyams, N. & W.Snyder (2005) “Young children never smuggle: Reflexive clitics and the universal            freezing hypothesis”, GALANA, University of Hawaii
Rizzi, L.  (1990) Relativized Minimality, MIT Press.

Rizzi, L.  (2004) “Locality and the left periphery”, in A. Belletti ed., Structures and beyond: The cartography    of syntactic structures, Vol. 3. OUP, 223–251


  1. So I have a question: is the idea that feature sets regulating minimality might themselves be acquired piecemeal? And if so, what is the envisaged inventory of relevant features and how might they be acquired? I ask this for depending on the kind of answer you are considering the emergence of the relevant features might be developmental as in Wexler or acquired/learned as in Crain/Demuth etc. In other words, even IF minimality plays a role, the developmental work looks like it arises from the feature etiology.

  2. I would dispute the contrast that is implicit in Adriana's question. (Yes, party pooper, I know.) Fully formed developmental accounts of children's abilities will not involve only grammatical features, nor will they involve only "general" mechanisms. They will make reference to (i) mechanisms for constructing and encoding linguistic representations, (ii) a set of features that are used in encoding those representations, and (iii) mechanisms for accessing information in those representations, and for updating the representations when things go wrong. (ii) is clearly linguistic, (iii) is plausibly not specifically linguistic, and (i) could be a mix of both. So any well developed account will surely need to make reference to language-specific and language-independent components.

    And even accounts that focus primarily on language-independent mechanisms, e.g., errors that children make due to delayed development of inhibitory control mechanisms, still make crucial reference to the specifics of the linguistic representations.

    Also, I think that it is important to distinguish "things that children just don't know at age X" from "things that children find hard at age X". Comprehension of object relative clauses likely falls into the second category.

  3. I have a clarification question before I go and mutate the mutanda for Ph for our blog readership. I'm not sure which of the pieces AB mentions is supposed to be the "general mechanism." This says "Passives are hard, but X, Y, and Z make them easier. Do general mechanisms ever play any comparable role in providing refined explanations of these developmental milestones?" Comparable to what? Are X, Y, and Z here the general mechanisms, or are they supposed to be specific (these ones could be either depending on what you take as your baseline)? And if not, why bring up X, Y, and Z here? Is the question "will we ever find anything at all that's not specific to the language faculty that explains milestones"? Because there the answer is going to be trivial unless you specify the Hilbert question a bit more precisely (e.g., "presence or absence of any input at all" is going to fit the bill, but that's obviously not what we want).

  4. Adriana sent me these comments:

    "If I may react to the comments so far with one single reaction, here it is: I agree with Colin Philips that in order to understand development we need to make appeal to different mechanisms at the same time, both formal/grammatical ones and language independent ones. The point of my question is simply that many detailed distinctions can only emerge and some research questions can only be asked if they are grounded in an integrated formal grammatical theory. Moreover, it is well known that children, in some developmental stages, make use of structures which are not really (or productively) available in the language(s) they are exposed to, but which are available in other languages (a crucial point already made by Steven Crain many many years ago and which keeps emerging every time one works on development). This suggests that the role of input factors, which are obviously relevant, is not trivial to underpin and needs to be qualified (this may be a relevant consideration for the third comment by ewan). And yes, Norbert, I (and my coauthors in the papers on the acquisition of relatives) think that what is crucial is precisely the “inventory of relevant features” that the locality/Relativized Minimality principle looks at/is sensitive to. Our current research program now is precisely to try to make explicit the (different) role of different (/same) features (as in e.g. our paper on the gender feature appeared in Lingua 2012). It is an independent very central and relevant question to ask what exactly matures: the feature itself (it could hardly be so, though, judging from children’s easiness in the mastery of agreement morphology in a highly inflected language e.g. Italian), or the mastery of a possible feature geometry, or the computational ability to calculate set theoretic relations involving features (as we have formulated the proposal), or the mastery of the role of a particular feature with respect to the principle."