SynLinks: Links between representation and processing in syntactic acquisition
The field is closer than ever to constructing its first set of large-scale theories of syntactic acquisition, synthesizing the insights from theoretical, experimental, and computational approaches. Importantly, these theories will recognize the links between children’s developing representations and their developing processing abilities, as well as the links between syntactic representations and other representations.
This all-day workshop brings together 6 experts who use a range of techniques to investigate syntactic acquisition and build theories of exactly this kind. Our goal is to highlight the links between syntactic theory, acquisition, and processing theory, and to foster broad group discussions that move us forward in building an integrated theory of syntactic acquisition.
The workshop will be held Saturday 09.17.16
The workshop will be held at the University of Connecticut, in Storrs, CT. The conference room is Oak Hall (the home of the Department of Linguistics), room 112.
- Misha Becker - University of North Carolina at Chapel Hill
- Kamil Deen - University of Hawaii at Manoa
- Jeff Lidz - University of Maryland, College Park
- Akira Omaki - University of Washington
- Lisa Pearl - University of California, Irvine
- Jesse Snedeker - Harvard University
- Diane Lillo-Martin and William Snyder will serve as discussion leaders during the two panel discussions.
Each talk will be 40 minutes, followed by 10 minutes of discussion, followed by a 10 minute break between speakers.
|8:30 - 8:55||Light refreshments: coffee, tea, breakfast snacks|
|8:55 - 9:00||Welcome and acknowledgment of NSF support (Lisa Pearl and Jon Sprouse)|
|9:00 - 10:00||Akira Omaki||Sources of syntactic creativitiy: Parsing to learn binding and wh-dependencies|
|10:00 - 11:00||Kamil Deen||Voice, Word Order and Relative Clauses in Tagalog|
|11:00 - 12:00||Jeff Lidz||On the Nature and Origins of Principle C|
|12:00 - 2:00||Catered Lunch||Discussion panel with all six speakers|
|2:00 - 3:00||Lisa Pearl||Integrating conceptual and structural cues: Theories for syntactic acquisition|
|3:00 - 4:00||Misha Becker||Sentence-level and Environmental Cues to Lexical Representation of Abstract Predicates|
|4:00 - 5:00||Jesse Snedeker||Clean Mapping: How conceptual structure might serve as the developmental and phylogenetic starting point of syntax|
|5:00 - 7:00||Catered Dinner||Discussion panel with all six speakers|
Sources of syntactic creativitiy: Parsing to learn binding and wh-dependencies Akira Omaki
Much developmental research has provided evidence for early mastery of grammatical knowledge, but there are puzzling reports of delays in grammar development where children appear to have mis-adopted a grammatical option from another language. For example, English-speaking children seem to allow wh-scope marking structures (e.g., What do you think who the cat chased __ ?)) that are ungrammatical in English but grammatical in other languages (e.g. Russian). Relatedly, Japanese children appear to treat the long-distance reflexive 'zibun' as an English-like local anaphor, even though it is compatible with both local and long-distance binding in Japanese adults' grammar. In this talk, I will discuss on-going studies in my lab that are beginning to explore whether these cases of apparent mis-learning may result from constraints on the parser of language learners, which may cause misanalysis and misrepresentation of information in the input. While our findings are still inconclusive, our studies provide useful methodologies for investigating the sources of syntactic creativity.
Voice, Word Order and Relative Clauses in Tagalog Kamil Deen
The tension between processing theories and grammatical theories of acquisition is often mediated by notions such as frequency and distance/length effects. Highly frequent forms, the thinking goes, become routinized quickly, and therefore automatized, thus reducing their processing burden. Lengthy dependencies are thought to exert heavier burdens on processing systems, and should therefore be disfavored, and perhaps acquired later by children. In this talk, I discuss the voice system in Tagalog, which allows for a variety of arguments to be elevated into a position of syntactic prominence. This prominence allows the targeted argument to be relativized, and it also has word order consequences. I discuss a series of studies that investigate the acquisition of word order, relativization, and various semantic factors such as definiteness, specificity and animacy. I show that factors such as frequency and distance/length make predictions not borne out in the data, but that common principles of grammar may account for the data.
On the Nature and Origins of Principle C Jeff Lidz
Integrating conceptual and structural cues: Theories for syntactic acquisition Lisa Pearl
The process of syntactic acquisition draws on a variety of information sources, including conceptual and structural cues. In this talk, I investigate what’s been called the Linking Problem, which concerns determining where event participants appear syntactically. I evaluate existing theories that synthesize conceptual and structural cues in different ways by embedding them in a computational model of the acquisition process and seeing if acquisition then proceeds the way we observe in children. This computational evaluation involves specifying several aspects of acquisition concretely, including (i) what children know already, (ii) what data children are learning from, and (iii) what children seem to learn when. I discuss current findings about the theoretical proposals that are best able to match children’s acquisition behavior.
Sentence-level and Environmental Cues to Lexical Representation of Abstract Predicates Misha Becker
We know from experimental work supporting the Syntactic Bootstrapping Hypothesis that as predicates become more abstract in meaning, sentence-level cues to the predicates' meaning and argument structure properties become more reliable than environmental cues (what is going on when a predicate is uttered) (Snedeker & Gleitman 2004, Gleitman et al. 2005, Becker & Estigarribia 2013). But environmental cues are not unimportant. This talk takes a closer look at how sentence-level and environmental types of cues interact in order to yield accurate representations of new lexical items. The talk will have two parts. Starting with my older work on the learning of raising verbs ('seem', 'tend') and 'tough'-adjectives ('easy', 'hard'), I show how both sentence-level and environmental cues lead children to make inferences about the grammatical properties of novel predicates that do not denote observable states or properties. I also describe a little-used methodology for tapping into speakers' intuitions about the grammaticality of sentences containing novel abstract predicates. This methodology relies on reaction time as an indicator of speakers' degree of surprise at encountering a particular linguistic form, where greater surprise (longer RT) could suggest a lower degree of grammaticality.
The second part of my talk turns to another class of abstract predicates, namely emotion adjectives. These adjectives occur in some of the same sentence frames as adjectives denoting concrete physical properties (NP is Adjective), but they also occur in frames ungrammatical for adjectives denoting physical properties (NP feels Adjective about something). While to some extent the denotation of emotion adjectives may be observable (e.g. through facial expressions), their true denotation is that of an internal mental state. I describe an experiment aimed at teasing apart sentence-level and environmental cues to the meanings of these types of adjectives.
Clean Mapping: How conceptual structure might serve as the developmental and phylogenetic starting point of syntax Jesse Snedecker
In this talk, I will entertain an old and powerful idea: that language reflects the structure of thought. This hypothesis has several attractive properties. First, it leads to a theory of language development that puts meaning (and thus communication) front and center, without attempting to reduce structure to function. This framework allows us to begin integrating work from infant cognition, semantic bootstrapping, syntactic bootstrapping, and statistical learning. This approach also provides a more satisfying evolutionary account, calling into question the underlying assumption of minimalism (that language is the result of just one evolutionary change). I will naively and optimistically conclude that this is good direction to go looking in, and then you all will tell me why I'm wrong.
If you are flying (thank you!) the closest airport is Bradley International (BDL), but it still requires either a rental car or a two-way taxi/uber ride. Sometimes cheaper flights can be found to Boston's Logan airport, followed by a Peter Pan bus ride.
We can offer limited crash space for graduate students attending from other universities (thanks to the graduate students in the UConn Department of Linguistics). If you would like to request crash space, you can indicate this on the registration form below.
The on-campus hotel at UConn is called the Nathan Hale Inn. The hotel has offered a special rate of $144 per night (plus tax) for workshop attendees. You must call to book using that rate. Please ask for the rate for "Linguistics Attendees Room Block" when you call. The hotel is not holding rooms for this (we are too small) so please book asap if you know you would like a room. There are no other hotels within walking distance, but several options within a short drive.
Please register here: registration form.
Registration is free. The workshop is funded by the National Science Foundation grants BCS-1347028 and BCS-1347115 to Lisa Pearl and Jon Sprouse. But we we do ask that you register. This will allow us to plan for the two working meals (lunch and dinner), and to create nametags. Please register by 08.17.16 so that we can plan accordingly.