Dissertation research, funded by a Doctoral Dissertation Improvement Grant from the National Science Foundation and internal grants from the Social and Behavioral Sciences Research Institute and the Graduate and Professional Student Council (UA). Public Dissertation Defense [slides] Dissertation[pdf] Presented at the 31st CUNY Sentence Processing Conference, 7 Mar 2018 [poster] Preliminary Maltese results discussed at UoM's Linguistics Circle, 19 Oct 2017 [slides] Presented at the Morphological Typology and Linguistic Cognition workshop, 22 July 2017 [semi-interactive poster] Presented at the Roots V workshop, 17 June 2017 [poster] Presented at Mental Lexicon 2016, 20 Oct 2016 [poster]
Learning root-and-pattern morphology involves tracking non-adjacent dependencies: In words ABCD and AHCF, A predicts the presence of C, but doesn't predict the presence of the second element in the string. Previous work shows that artificial grammars mimicking root-and-pattern morphology and vowel harmony are quite difficult for adults to learn. However, previous research also shows that when an artificial grammar has elements of a speaker's native language (such as vowel harmony), those speakers are better able to learn the artificial grammar with analogous structures. My research looks at this in root-and-pattern morphology across native English speakers, native Arabic speakers, and native Maltese speakers.
Drake, S. (2018). The form and productivity of the Maltese morphological diminutive. Morphology, 28(3), 297-323. doi:10.1007/s11525-018-9328-0 Presented at the 90th Meeting of the Linguistics Society of America, 10 Jan 2016 [slides] Presented at the 5th International Conference on Maltese Linguistics, 25 June 2015
Maltese is an especially interesting language morphologically speaking because its lexicon is split between Semitic words and Indo-European words, both of which take their respective morphological structures. I used a wug task to find out whether native Maltese speakers use a morphological diminutive (like 'wuglet') or a lexical diminutive (like 'little wug') when encountering novel words. I also used this task to find out whether native speakers would use Semitic morphology when encountering a novel word that sounded Semitic, and Indo-European morphology when encountering a novel word that sounded Indo-European.
Cascaded semantic activation
Bell, D., Forster, K., & Drake, S. (2015). Early Semantic Activation in a Semantic Categorization Task with Masked Primes: Cascaded or not? Journal of Memory and Language, 85, 1-14. doi:10.1016/j.jml.2015.06.007
If activation is cascaded, the assumption is that semantic properties of the input word are linked to other words in the lexicon; thus, semantically related words should show a congruence effect in a categorization task with a masked prime. Over several experiments, we show that semantic activation is typically not cascaded, and the categorization task alters how the masked prime is processed.
Manuscripts in preparation
Psycholinguistic theories of the lexicon vs Distributed Morphology
With Dr. Heidi Harley (University of Arizona) Presented at Mental Lexicon 2016, 21 Oct 2016. [poster] From dissertation.
Psycholinguistic theories of the structure of the mental lexicon may be more compatible with the framework of Distributed Morphology than at first glance. If the two frameworks are married, there is further potential for theoretical linguistics and psycholinguistics to continue to inform each other to create more accurate theories with greater explanatory ability. Current work explores Distributed Morphology in light of morphological impairments.
Errors in learning root-and-pattern morphology with Wug Tests
Production tasks in an artificial grammar demonstrate that error patterns in morphology differ based on native language. In my research, I suggest that L1 transfer to an L2 (or Ln) starts after a very short learning period and differs subtly even among closely related languages with productive root-and-pattern morphological systems.
Maltese broken plurals
With Dr. Rebecca Sharp (University of Arizona) Presented at the 6th International Conference of Maltese Linguistics, 8 June 2017. [slides]
Like Arabic, Maltese has a system of both sound (regular) and broken (irregular) plurals. We are building a parser that will predict the form of a broken plural using real and nonsense words, and those intuitions will be tested via acceptability ratings and a wug test with native Maltese speakers.
Effects of bilingualism on morphological variation in Maltese
Forthcoming in Coyote Papers: Working Papers in Linguistics, conference proceedings for Arizona Linguistics Circle. [lingbuzz] Presented at Arizona Linguistics Circle 10, 4 Dec 2016 [slides]
It is well known that bilingualism affects various types of metalinguistic knowledge. Further, dominance in a particular language can affect how morphological and syntactic structures are used in a language. I investigate this in Maltese, where individuals are bilingual in both Maltese and English, but are dominant users of one of those languages.
Lily words: Morphophonological dissimilation in English
With Dr. Lauren Ackerman (Newcastle University) Presented at the 32nd Annual CUNY Conference on Human Sentence Processing, Mar 2019 [poster] Presented at the 92nd Annual Meeting of the Linguistics Society of America, 5 Jan 2018 [poster][proceedings paper]
We are studying under what conditions the adjectival and adverbial -ly suffixes next to one another are acceptable (e.g., jollily vs. smellily vs. lovelily), using sentence acceptability judgments and lexical decision tasks.
Conversational bilingual speech
With many members and affiliates of the Douglass Phonetics Lab
We used production and perception studies to analyze how L2 English speakers differed in stop consonant reduction depending on their L1 (Japanese, Dutch, Spanish), and also whether there were any differences in reduction across dialects of L1 English speakers (US, Canadian, New Zealand).
Replicating psycholinguistic tests of the Metrical Segmentation Strategy
With many members and affiliates of the Douglass Phonetics Lab Presented at the 177th Annual Meeting of the Acoustical Society of America, 16 May 2019 (4pSC36). [poster]
Based on statistical regularities of their language, humans hypothesize word boundaries according to the stress and vowel quality of utterances. Cutler and Norris (1988) formalized this as the Metrical Segmentation Strategy, and previous experiments (e.g., Norris et al., 1995) have borne out these predictions. Here, we present a partial replication of Norris et al. (1995), with some possibilities suggested for what may have occurred for the differences we find.
Distributed Morphology feature geometries in acquisition
Using the corpora in the CHILDES database, I am examining the order of acquisition of morphosyntactic features by infants, and whether that order is accurately predicted by feature hierarchies in the Distributed Morphology framework. The current work is based on English, but future studies will involve analyses of languages with greater morphological complexity, children with morphosyntactic impairments, and the preservation of morphosyntactic features of adults with grammatical impairments.