A Report on the First Native Language Identification Shared Task
Joel Tetreault Nuance CommunicationsDaniel Blanchard Educational Testing ServiceAoife Cahill Educational Testing Service
Native Language IdentificationTask of automatically identifying
a speaker’s first language based solely on the speaker’s writing in another language
Applications:◦Authorship profiling (Estival et al.,
2007)◦Education: more targeted feedback
to language learners (Leacock et al., 2010)
Sample Essay 1 No risk no fun I agree the statement
"Successful people try new things and take risk".In my mind it is so, to. When you thing you like do new stuff you need a liddelbit the kick. That is the big point what I need. For exsample I like to go to a big city like New York. I was never in this town I dont no from the city. But I like go to the city. Thats fun I stay every time for proplems. I need eat a hood offer my head. The ather side I can go dow. I dont gat waht I need…Next exsample the wall street you put money in funds, well you this make a good job. Dont for get the risk look like lose money.German
Sample Essay 2 For example, if you take a look at an ordinary
school, you have different teachers for every subject. Your calculus teacher is different than your literature teacher. Each teacher must specialize in a specific subject in order to convey suffiecient and proper information to the students. However, that doesn't mean that the teacher is narrow-minded and has a limited perspective in life because to specialize in one subject doesn't hinder you or stop you from exploring other subjects.
Arabic
MotivationLots of work in NLI but…it has
been hard to compare different approaches:
1. ICLEv2 (Granger et al, 2009): de facto train/test data is small and has NLI-unfriendly idiosyncrasies
2. No consensus on evaluation:- Which L1’s / how many L1’s?- Train/test splits?- Best features?
ContributionsGoal to unify community and help
field progressProvide a larger, more NLI-friendly
corpus that improves upon ICLEv2Common evaluation framework
◦Everyone evaluates using same train/dev/test splits and same L1s
Corpus and scripts to be made public to further promote the field
OutlinePrior WorkDataShared Task OverviewResultsNLI Shared Task in the Future
Prior WorkTreat NLI as a classification taskKoppel et al. (2005): POS n-grams, content
and function words, spelling and grammatical errors
Syntactic features (Wong and Dras, 2011)Tree Substitution Grammars (Swanson and
Charniak, 2012)Adaptor Grammars (Wong et al., 2012)Data Size Effects (Brooke and Hirst, 2012)Word n-grams (Bykh and Meurers, 2012): LMs and Ensemble Classifiers (Tetreault et
al., 2012)
Data: TOEFL11 Corpus12,100 essays from the ETS Test of
English as a Foreign Language (TOEFL) 11 L1s:
◦Arabic, Chinese French, German, Hindi, Italian, Japanese, Korean, Spanish, Telugu, Turkish
◦900 train / 100 dev / 100 testSampled for equal representation of L1s
across topics as much as possibleIncludes 3-tier proficiency levelPublic release via LDC this summer?
Shared Task Description: 3 Sub-tasks1. Closed-Training: 11-way classification
task using only TOEFL11-TRAIN and DEV2. Open-Training-1: use of any amount or
type of training data excluding TOEFL113. Open-Training-2: use of any amount or
type of training data combined with TOEFL11
* All sub-tasks use TOEFL11-TEST for the final evaluation set
Shared Task DescriptionEach team allowed to submit up
to 5 different systems per taskTeams submitted a CSV file for
each system to NLI OrganizersEvaluation script automatically
compares each prediction file to gold standard and creates performance report and contingency tables
29 TeamsBobicev Eurac MITRE
“Carnie”UKP
Chonger HAUTCS MQ UnibucCMU-Haifa ItaliaNLP NAIST UNTCologne-Nijmegen
Jarvis NRC UTD
CoRAL Lab @ UAB Kyle et al. Oslo NLI VTEXCUNI (Charles University)
LIMSI Toronto
Cywu LTRC IIIT Hyderabad
Tuebingen
Dartmouth Michigan Ualberta
RESULTS
Sub-Task Participation Statistics
Sub-task # Teams Competing
# Submissions
Closed 29 116Open-1 3 13Open-2 4 15
Closed Sub-TaskSee Table 3 of Report for full
resultsNo statistically significant
differences between top 5 teamsTeam Name Abbreviation Overall
AccuracyJarvis JAR 0.836Oslo NLI OSL 0.834Unibuc BUC 0.827MITRE “Carnie” CAR 0.826Tuebingen TUE 0.822
Open Sub-tasksChallenge : finding new data to cover each L1
Data sources for HIN & TEL:◦ ICNALE Pakistani essays HIN (TUE team)◦ Bilingual blogs (TOR & TUE team)
Corpus Description
ICLE All L1s except ARA, HIN, TELFCE All L1s except ARA , HIN, TELICNALE CHI, JPN, KOR essays onlyLang8 All L1s, but mostly Asian L1s
Discussion of ApproachesMachine Learning
◦SVM overwhelmingly the most popular approach
◦4 teams also tried Ensemble classifiers
◦String kernels (BUC) using character level n-grams
Discussion of ApproachesFeatures
◦N-grams: word, POS, character, function
◦Syntactic Features: Dependencies, TSG, CF Productions, Adaptor Grammars
◦Spelling Features4 of top 5 teams used n-grams at
least 4-grams, some went up to 9-grams
2 of top 10 teams used syntactic features
Future of NLI Shared TaskIdeas to expand scope of task
◦ Use a new set of TOEFL essays for test◦ Expand genres: blogs? Tweets? ◦ Number of L1s◦ Do different L2
ItaliaNLP – preparing Italian NLI corpus with CNR Pisa Also a corpus of Finnish with L1 (Turku Uni)
◦ Add slavic languagesLogistics
◦ Hold another shared task in 2014? Or 2015?◦ Merge with PAN Shared Task?
Tell us your thoughts!
AcknowledgmentsDerrick Higgins (ETS)ETS TOEFL Patrick Houghton (ETS)BEA8 OrganizersAll the NLI Participants!
http://www.nlisharedtask2013.org/