Browsing by Subject "Description"

Now showing 1 - 4 of 4

"Advanced" Arabic : investigating learners’ lexical richness in the context of an oral interview
(2013-08) Loomis, Summer; Raizen, Esther, 1951-; Brustad, Kristen; Plakans, Lia; Raz, Adi; Mohammad, Mohammad
This study used recordings produced in the American Council on the Teaching of Foreign Languages’ (ACTFL) Oral Proficiency Interviews (OPIs) to investigate the quantity and lexical richness of second language (L2) Arabic speakers’ lexical production. The study focused on 28 full-length tests and 53 sub-samples of narration and description, selected from an initial data set of 115 OPIs. The research questions were: 1) What are the average words and words per minute (WPM) produced by Advanced-Mid rating level test takers in this data set? Do Intermediate-Mid rating level test takers produce fewer words and WPM than Advanced-Mid rating level test takers? Do Superior rating level test takers produce more words and more WPM than Advanced-Mid speakers? 2) What is the lexical variation in the Advanced-Mid samples as measured by type-token ratio (TTR)? Is this variation higher or lower than the lexical variation of test taker samples at the Intermediate-Mid and Superior rating levels? 3) How many shared words produced by learners at the Advanced rating levels are from beyond the 2,000 most frequently used words in Arabic according to Buckwalter and Parkinson’s frequency dictionary (Buckwalter and Parkinson 2011)? 4) What qualitative observations can be made about test takers’ narration and description attempts at the Advanced rating levels? How do these attempts compare to narration and description attempts by test takers at the Intermediate and Superior rating levels respectively? The WPM and TTR values for the Advanced-Mid rating level differentiated this test taker speech from the Intermediate-Mid rating level speech. However, the WPM and TTR measures did not distinguish between the Advanced-Mid rating level and the Superior rating level test takers. In regards to word frequency, learners at the Advanced-Mid rating level did not produce shared words that were beyond the 2,000 most frequently used words in Arabic. However, the qualitative observations of the Advanced rating levels’ descriptions and narrations appeared to show a difference between this group’s lexical resources and those of the Intermediate and Superior rating levels. These findings and related suggestions for future research on the advanced L2 speaker of Arabic were also discussed.
Natural-language video description with deep recurrent neural networks
(2017-08) Venugopalan, Subhashini; Mooney, Raymond J. (Raymond Joseph); Grauman, Kristen; Stone, Peter; Saenko, Kate; Darrell, Trevor
For most people, watching a brief video and describing what happened (in words) is an easy task. For machines, extracting meaning from video pixels and generating a sentence description is a very complex problem. The goal of this thesis is to develop models that can automatically generate natural language descriptions for events in videos. It presents several approaches to automatic video description by building on recent advances in “deep” machine learning. The techniques presented in this thesis view the task of video description akin to machine translation, treating the video domain as a source “language” and uses deep neural net architectures to “translate” videos to text. Specifically, I develop video captioning techniques using a unified deep neural network with both convolutional and recurrent structure, modeling the temporal elements in videos and language with deep recurrent neural networks. In my initial approach, I adapt a model that can learn from paired images and captions to transfer knowledge from this auxiliary task to generate descriptions for short video clips. Next, I present an end-to-end deep network that can jointly model a sequence of video frames and a sequence of words. To further improve grammaticality and descriptive quality, I also propose methods to integrate linguistic knowledge from plain text corpora. Additionally, I show that such linguistic knowledge can help describe novel objects unseen in paired image/video-caption data. Finally, moving beyond short video clips, I present methods to process longer multi-activity videos, specifically to jointly segment and describe coherent event sequences in movies.
The osteology of Sarahsaurus aurifontanalis and geochemical observations of the dinosaurs from the type quarry of Sarahsaurus (Kayenta Formation), Coconino County, Arizona
(2013-05) Marsh, Adam Douglas; Rowe, Timothy, 1953-
Sarahsaurus aurifontanalis is the most recent sauropodomorph dinosaur to be discovered and named from the Early Jurassic of North America. The dinosaur is represented by a mostly complete and articulated holotype specimen that preserves a unique manual phalangeal count of 2-3-4-2-2 and accessory pubic foramen adjacent to the obturator foramen. The holotype of Sarahsaurus comprises a braincase and isolated cranial elements, but the skull previously referred to this taxon, MCZ 8893, can only be provisionally referred to Sarahsaurus until additional crania are found associated with postcranial material. Sarahsaurus comes from the middle third of the Kayenta Formation, which is considered to be Early Jurassic in age despite the absence of a radiometric date from that unit. A new technique used to obtain a U-Pb radiometric date from the type quarry of Sarahsaurus in the Kayenta Formation was influenced by secondary uranium enrichment in the open system of the fossil bone. That suggests that uranium within the Kayenta Formation may be the result of the movement of groundwater during the Laramide orogeny in the Late Cretaceous and Early Eocene, and lends support to the hypothesis that the uplift of the Colorado Plateau began relatively early in Late Cretaceous to the Eocene.
Vosmaer beschryving
(P. Meijer, 1778) Vosmaer, A. (Arnout), 1720-1799