Scarica Morphosyntax and Experimental Studies on Language Acquisition: A Grammar-Based Approach - e più Appunti in PDF di Morfologia e Sintassi solo su Docsity!
Morphosyntax and experimental studies on language
HO.
Calendar November Mo 25 14 - 16 room 149 Tue 26 9 - 11 room 401 We 27 9 - 11 room 349A January Tue 7 9 - 11 We 8 9 - 11 18 - 20 room 349B Th 9 14 - 18 room 15 (presentations) Fr 10 9 - 12 room 356 (presentations) 14 - 17 room 15 (presentations) December Mo 2 14 - 16 room 149 Tue 3 9 - 11 room 401 We 4 9 - 11 room 349A Mo 9 14 - 16 room 149 Tue 10 9 - 11 room 401 We 11 9 - 11 room 349A Mo 16 14 - 16 room 149 Tue 17 9 - 11 room 401 We 18 9 - 11 room 349A Contents Main domain that will be investigated: Language acquisition with some occasional reference to different modes of acquisition, such as: Monolingual, bilingual, L2, SLI, and other types of atypical language growth/ (or manifestations of problematic behaviour through language, e.g. Dyslexia, HI), with some comparisons with other forms of language pathologies, such as e.g. (agrammatic) aphasia. Although comparative considerations on multilingualism are developed in some better detail in the first year class. General leitmotif : The role of the internal grammar in the process of language acquisition (grammar based approach: how does the grammar deal with unexpected behaviour?) Presentations of results and discussions based on often current or anyway recent research. ◦ Slides of the classes available at: http://www.ciscl.unisi.it/didattica.htm ◦ Belletti, A. & M.T. Guasti (2015) The Acquisition of Italian, John Benjamins ◦ Guasti, M.T. (2017) Language Acquisition, MIT Press ◦ some integrated references (later on, for presentations in class).
Cfr. also some of the contributions in: ◦ Roberts, I. The Oxford Handbook of Universal Grammar, Oxford University Press, 2017 ◦ Lidz, J., W. Snyder, J. Pater, The Oxford Handbook of developmental Linguistics, Oxford University Press, 2016 Themes Types of object A’-dependencies and their acquisition:
- Relative clauses: Subject Relatives vs Object Relatives
- A different object A’-dependency: Object-topicalization in ClLD
- Aspects of the acquisition and derivation of (Romance) clitic pronouns
- Another type of (Object) A’- dependency: Subject vs Object Clefts and their presence in answering strategies
- Aspects of the acquisition of (types of) passive, also in comparison with ClLD and the syntax of causatives and their relation with passive, with impersonal and reflexive constructions
- On a-marking of direct objects in a comparative perspective: Differential Object Marking/DOM and Left-peripheral (a)-topics 7. Poverty of the stimulus and children’s grammatical creativity
- Theoretical hypotheses at the base of the research questions are supported and refined through experimental results (beyond grammaticality judgments) The experimental approach: some general considerations As linguists we are required to gather data in order to conduct experimental researches rough an experimental approach and there are different ways to do that If we have an experimental attitude, we should always keep in mind the three properties of the speaker. The easiest and cheapest way to discover how a speaker behaves is by asking questions: grammaticality judgment. We have a certain sentence and ask if it is possible in that specific language or not or also ask the speakers to rate that sentence in a scale. John spoke Mary with. Is it possible in English? No → the answer is a Grammaticality Judgement. How bad is this sentence in a scale from 1 to 5? 4 → Rating the sentence task They are experiments if they are minimally comparative: not asked in the vacuum but in comparison and asked to answer a specific question we as researchers have in mind.
speaker to answer a question (i.e. we ask a question to the speaker and we fix a condition such as please answer with a full clause). ◦ Comprehension → picture/videos/scenarios to sentence matching tasks ◦ Production → eliciting the production of particular structures through some task, most typically through: Production experiments focus on the elicitation of the production of the structure under investigation. ⎯ answering a question (Recall experiment discussed last year on acquisition of VS order, S=new information subject in different multilingual populations) chi ha parlato? → a question on subject such this one produces an elicited production of the word order VS Can we then investigate what will happen with a clitic? Eliciting question: cos’ha fatto con la penna? The expected answer would imply: - No repetition of the final object DP
- The use of an object clitic
- Null subject Note that languages with clitics and strong pronouns have a different distribution and some related features such as the animacy, where strong pronouns display animacy and clitics inanimacy. Answer: L’ha appoggiata sul tavolo. → no subject noun repetition + CL - V order She has put it on the table → no subject noun repetition + V-CL order If we use the pen , we have an unfelicitus sentence: no repetition of the noun. The first thing to notice in this comparison is the fact that there is something wrong in both languages in repeating the noun – no repetition of the subject noun, this is called noun penalty to repeat a noun subject in contexts in which it would be felicitus- and indeed in both sentences we find a pronoun. ** Ha messo lei sul tavolo. ⎯ completing a sentence ⎯ repeating a sentence (e.g. word repetition) → The time dimension is also important There is something deeply wrong also here. This would be a pronoun after all but we do not accept that and prefer something like the clitic. And we find that through an elicitation production experiment.
Thanks to the work of Cardinaletti and Shaker, we discover that strong pronouns own stress and have not only therefore a different distribution but also some features that associate to this necessarily, that is the animacy [+AN]. Online experiments: time taken during the experiment; how long does it take to do this task. This method is able to tells us when the problem starts, and measurements take place while the task is unfolding. Offline experiments: the focus is if the sentence is well repeated. It tells us if there are problems, where are the problems and in which specific structure. Often the time dimension in this kind of task is not relevant because they focus in telling us where the problem start, namely in which kind of structure.
- Online tasks/procedures : ⎯ Reading time + accuracy measured Accuracy questions are usually asked to see if the structure has really been understood ⎯ Measuring time in fulfilling the task ⎯ Eye-tracking It measures the way experimental subjects look at the stimuli: the longer is the time, the harder are questions and sentences ⎯ …………………… ⎯ Types of neuroimaging (ERP, fMRI,…) ⎯ Reaction time Often reaction time can signal the presence of something complex indeed long time is interpreted as a marker of computational complexity (but that is already and interpretation). Online tasks are tasks that measures the timing of the answer so the time dimension here is crucial (usually offline but they can become online too). Often high reaction time is associated to high difficulty of the sentence. ORs are the harder to be processed, they take longer indeed to be repeated. When we have to repeat something, it can be an easy task, like the sentence the apple is good , or it can also be very difficult for children, like the sentences with this kind of construction the child the mum hugs is good (OR- object relatives). These second example sentences are harder than others and a way to find that out is through repetition tasks.
- Kayne: microcomparisons of closely related languages mimic the lab situation Kayne approach is emphasizing the microcomparative syntax: comparisons that have only one property varying for example among northern Italian dialects.
- Hence, some explicit theoretical hypotheses to put to test are necessary, they are a prerequisite for the beginning of the construction of an experimental task. The dialogue between theory and experimentation is crucial: Linguistic theory feeds the experimental work in generating research questions, which can be properly and explicitly formulated. At the same time, results from well controlled experiments can illuminate linguistic theory by suggesting possible (re)formulations and precise implementations of the explanatory principle(s) which helped raising the research question. The work on the role of locality in accounting for developmental paths, which we will study in detail, has precisely this property. The dialogue is crucial in the fact that the experimentation provides us a shortcut and then questions are generated and experiments conducted and then new theories are delineated or refinements of theoretical hypothesis. The problem with a sentence with object dependency like the first one resides in the presence of a lexical subject which is somehow overcome when the subject is null or functional. OR – lexical subject: the child that the mum is hugging <-> OR – pronominal subject: the child that she is hugging <-> In this case, the second sentence is better understood than the first one. There seem to be difficulties with the lexical NPs. We will also try to see other examples, e.g. the role of Binding principle C in guiding possible use and interpretations of lexical noun phrases The Baseline → a term of comparison is essential Any experiment must have a baseline, which is constituted by the control group. This is because any result acquires a meaning only in comparison with what is classified as being TD (typically development). The phonological impairments is immediately noticed while the syntactic and morphosyntactic impairment is less straightforwardly noticeable because it might be the case that children are just avoiding certain constructions. Baseline, a crucial notion. Any experiment must have a so called Control group (CO).
No experimental result on any population, on any specific construction has a value per se, beside pure description_._ It acquires one only if compared to what is independently determined as the typical behaviour of a defined population on the same empirical domain (e.g. a given linguistic construction). E.g:
- children of different ages in typical development (at 3 at 5 etc….)
- children of different ages vs adults (to check for development)
- children with (S)-SLI vs typically developing children of the same age (age – match)
- children with (S)-SLI vs typically developing children of the same language level (language-match; determined through some standardized test)
- agrammatic adults vs adults with no diagnosed pathology Agrammatism is an acquired type of impairment (aphasia) that touches on the functional lexicon (determiners, complementizers, pronouns and prepositions). It is not to be confused with developmental impairments such as SLI (specific language impairment, which since few years is known as DLD – developmental language disorder).
- monolingual children at age X vs bilingual children at the same age X
- bilingual children at age X vs L2 children at age the same age X ……………… In conclusion, in order to identify some potential deviation or to spot a developmental path it is necessary to know what we can expect in the domain under investigation in the relevant population against which the experimental subjects will be studied. A term of comparison in these cases is needed to identify the deviation or the different developmental path. One possible way is a grammar-based approach, which tracks how the grammar deals with perhaps unexpected behaviours.
Italian Q: Chi ha parlato? A:? E’ la mamma Q^1 : Chi è che ha parlato? A^1 : E’ la mamma The answer sounds more felicitous to the second question than to the first one because we have a cleft-like question and we are following that kind of structure in the answer.
- Aspects of the acquisition of (types of) passive, also in comparison with ClLD and the syntax of causatives and their relation with passive, with impersonal and reflexive constructions Q: Cosa succede al mio amico cane? A: pro viene lavato dal gatto → passive Lo lava il gatto → Pronoun or ClLD Il gatto lo lava → Pronoun or ClLD Il cane, il gatto lo lava → Pronoun or ClLD (+ topic: overtly expr only in some conditions) Adults: tend to answer with a passive Children: tend to answer with the pronoun and the ClLD
- On a-marking of direct objects in a comparative perspective: Differential Object Marking/DOM and Left-peripheral (a)-topics In Italian, children are expected to produce a ClLD, sometimes actually mark the object with a preposition (in regions in which the DO should not be marked by a preposition) saying things like: Al bambino, la mamma lo abbraccia → typical of Spanish LP “a-topic” obj cl In a “A-Topic” structure, it remains an object clitic: the “a” marker does not transform the clitic into a dative pronoun (while al bambino is dative). This is a construction typical of Spanish.
Differential Object Marking (DOM) In structures of this kind, objects are marked by a different preposition: *La mamma abbraccia al bambino. This sentence is ungrammatical in standard Italian but it is accepted in southern varieties. Most children tested only produced an a-marked object in ClLD constructions which shows that the variety they are speaking is not a differential object marking variety (they came from the province of Florence).
7. Poverty of the stimulus and children’s grammatical creativity Grammatical creativity means that in acquisition, children do not only test options at random; they explore possibilities that are not instantiated in their own language. In the LP, children explore a-marking.
- Theoretical hypotheses at the base of the research questions are supported and refined through experimental results (beyond grammaticality judgments) Two final general considerations : 1. Production and comprehension There is an intrinsic difference between production and comprehension tasks and has to do with choices: when we are asked to produce something, we have choices, even though we might not be always completely aware of that, while in front of a comprehension task we do not have choices to make. Eliciting a question requires us to make choices among the different types of structures available as answers to that specific question and our system is very efficient in selecting a particular structure. Clearly, we do not have something like that in comprehension tasks because the answers are already given. However, production and comprehension should work together in the fact that they both involve the computation of grammar rules, even if in different ways. Of course, production occurs in real time and it can be affected by external factors and this must be taken into account in processing the data collected for production experiments.
E.g.: Do we first decide on V and its arguments? Or do we start with the nominal arguments and then associate them to a V? Is there a difference between associating a nominal DP-object to V compared to a nominal DP-subject? Recent studies based on a V-final language like Japanese (by C. Phillips et al.) seem to suggest that OV is accessed to faster than SV. This is somehow consistent with the linguistic analysis (see h.o. of first year), according to which the relation between V and its Internal Argument/IA is stricter (first Merge) than the one between V and the External Argument/EA (mediated through ‘v’ and then through T, the subject of the clause). A lot of experiments elicited that OV is accessed faster than SV and it is interesting to understand the verbal argument structure. The theoretical analysis appears to be directly reflected in actual production. According to Philips, the conclusion that can be reached is that the structure OV (because the tests have been run in Japanese, with is an head final language) is accessed faster than the structure SV. According to this our grammas built the VO faster than the structure SV. It seems that we process information bottom up, looking at an X-bar tree, we build faster the VO than the SV.
2. The dialog theory-experimentation: Merge And Brain Imaging “According to most linguists, the syntactic structure of sentences involves a tree-like hierarchy of nested phrases, as in the sentence [happy linguists] [draw [a diagram]]. Here, we searched for the neural implementation of this hypothetical construct. Epileptic patients volunteered to perform a language task while implanted with intracranial electrodes for clinical purposes. While patients read sentences one word at a time, neural activation in left hemisphere language areas increased with each successive word but decreased suddenly whenever words could be merged into a phrase. This may be the neural footprint of “merge,” a fundamental tree-building operation that has been hypothesized to allow for the recursive properties of human language.” In brain imaging techniques, the subjects are exposed to one word at a time and therefore certain aspects of syntax are difficult to be tested. Matthew J. Nelson, Imen El Karoui, Kristof Giber, Xiaofang Yang, Laurent Cohen, Hilda Koopman, Sydney S. Cash, Lionel Naccache, John T. Hale, Christophe Pallier and Stanislas Dehaene PNAS 2017 May, 114 (18) E3669-E3678. https://doi.org/10.1073/pnas.
