Christopher D. Manning · page 4
Papers
- Parsing Models for Identifying Multiword Expressions (2013 · 110 citations)
The paper treats the detection of multiword expressions (MWEs) as part of syntactic parsing. Earlier work handled this as a separate step that classifies n-grams using co-occurrence statistics. That approach misses connections between variants such as part of speech and parts of speech , and it suffers from data… - Word Segmentation of Informal Arabic with Domain Adaptation (2014 · 101 citations)
Splitting clitics off Arabic words helps downstream tasks such as parsing and translation. However, existing segmenters either target formal Modern Standard Arabic (MSA) only or depend on hand-built, dialect-specific morphological analyzers. The authors extend the character-level CRF segmenter of Green and DeNero… - Truncation Sampling as Language Model Desmoothing (2022 · 95 citations)
Sampling directly from neural language models often yields long texts that drift into nonsense. Truncation methods such as top-k and top-p help by zeroing out unlikely tokens, but it has been unclear what they are actually trying to achieve. The authors treat any neural LM as a blend of the true distribution with a… - Labeled LDA: A supervised topic model for credit attribution in multi-labeled corpora (2009 · 87 citations)
Many text collections, from newswire to social bookmarking sites like del.icio.us, give each document several human tags, but a tag rarely fits every part of a document equally well. - BioMedLM: A 2.7B Parameter Language Model Trained On Biomedical Text (2024 · 79 citations)
Frontier systems such as GPT-4 and Med-PaLM 2 perform very well on biomedical NLP. However, they are huge, expensive, reachable only through remote APIs (a problem for patient privacy) and trained on undisclosed data. The authors ask whether a much smaller, domain-focused model can be competitive. They built… - RNNs can generate bounded hierarchical languages with optimal memory (2020 · 65 citations)
RNN language models capture natural-language syntax well in practice, but the theory behind this has been thin. - Was It Good? It Was Provocative.” Learning the Meaning of Scalar Adjectives (2010 · 56 citations)
Indirect answers to yes/no questions often do not say "yes" or "no" outright. "Was it good? It was provocative" leaves the answer open, while "It was great" does not. The authors mine two kinds of evidence. First, scale order between gradable adjectives comes from online IMDB reviews: a word's expected star rating… - A Reasoning-Focused Legal Retrieval Benchmark (2025 · 55 citations)
Legal AI tools increasingly pair LLMs with retrievers over case law and statutes, but realistic English benchmarks for testing such pipelines are scarce. - More Constructions, More Genres: Extending Stanford Dependencies (2013 · 47 citations)
The Stanford scheme was built on newswire and gives little analysis to constructions that are rare there but common in web text: tough adjectives, free relatives, comparatives, small clauses, questions, imperatives, fragments and lists. - Sentences with Gapping: Parsing and Reconstructing Elided Predicates (2018 · 40 citations)
In "Paul likes coffee and Mary tea" the verb is elided, so standard parsers attach the remnants badly and downstream relation extraction misses the predicate. - Arc-swift: A Novel Transition System for Dependency Parsing (2017 · 39 citations)
Transition-based parsers reach a distant attachment only through a sequence of local shifts and reduces, and every step is a place to make an error. - An Emulator for Fine-Tuning Large Language Models using Small Language Models (2023 · 36 citations)
Instruction-tuned LLMs come from two stages: broad pre-training and narrower fine-tuning. It is often assumed that pre-training supplies knowledge while fine-tuning mainly shapes behavior, but this has rarely been tested directly. The authors propose emulated fine-tuning (EFT), which treats a fine-tuned model as the… - NP Subject Detection in Verb-initial Arabic Clauses (2009 · 26 citations)
Arabic often places the verb before its subject, so translating into English means moving the subject, and sometimes a long nested iDafa noun phrase, ahead of the verb. - Gapping Constructions in Universal Dependencies v2 (2017 · 25 citations)
This is a design paper rather than an experiment. It asks how gapping (the omission of a repeated verb, as in "John likes tea, and Mary coffee") should be represented in UD v2. The authors survey the cross-linguistic patterns of gapping, including languages where the overt verb sits in the last conjunct, and compare… - Biomedical and Clinical English Model Packages in the Stanza Python NLP Library (2020 · 24 citations)
General-purpose NLP toolkits handle biomedical and clinical English poorly, and domain-specific tools often lack modern neural models or native Python support. - Does Universal Dependencies need a parsing representation? An investigation of English (2015 · 24 citations)
UD makes content words the heads and attaches function words (auxiliaries, copulas, subordinators, adpositions) to them. - Presupposed Content and Entailments in Natural Language Inference (2009 · 18 citations)
The paper addresses a gap in the NatLog natural logic system for NLI. NatLog runs negation, monotonicity and implicatures through one projection mechanism, but presuppositions behave differently. They survive negation, yet some operators block them and conditionals can cancel them. The authors add a separate… - Multi-word expressions in textual inference: Much ado about nothing? (2009 · 12 citations)
The paper asks whether multi-word expressions (MWEs) matter for recognising textual entailment (RTE). It uses the manual Microsoft Research word alignments for the RTE2 data and counts the MWEs that appear as multi-word alignment links. Only 81 (dev) and 60 (test) true MWEs occur, in 8.2% of the 1,600 sentence pairs… - MSCAW-coref: Multilingual, Singleton and Conjunction-Aware Word-Level Coreference Resolution (2024 · 8 citations)
Strong multilingual coreference systems usually enumerate candidate spans and then connect them, so their cost grows as O(n^4) in document length. - PragmatiCQA: A Dataset for Pragmatic Question Answering in Conversations (2023 · 8 citations)
The dataset has 6,873 question-answer pairs in teacher-student conversations about Fandom wiki topics. For each student question the teacher marks literal answer spans, pragmatic spans that answer the follow-up the student is likely to ask, and a paraphrased final answer. Crowdworkers choose topics of mutual interest… - Mechanisms vs. Outcomes: Probing for Syntax Fails to Explain Performance on Targeted Syntactic Evaluations (2025 · 4 citations)
The authors test whether the quality of syntactic probing can predict how a language model performs on targeted grammatical tests. - Do “English” Named Entity Recognizers Work Well on Global Englishes? (2023 · 4 citations)
Standard English NER benchmarks such as CoNLL 2003 and OntoNotes are drawn almost entirely from British and American newswire. - Universal Dependencies for Sindhi (2025)
Sindhi is an Indo-Aryan language with split ergativity, spoken by about 40 million people in Pakistan and India.
This page in Markdown → https://md.research.elamur.ai/free/en/authors/christopher-d-manning/page/4/