Brighter than Gold: Figurative Language in User Generated Comparisons

Abstract

Comparisons are common linguistic devices used to indicate the likeness of two things. Often, this likeness is not meant in the literal sense-for example, "I slept like a log" does not imply that logs actually sleep. In this paper we propose a computational study of figurative comparisons, or similes. Our starting point is a new large dataset of comparisons extracted from product reviews and annotated for figurativeness. We use this dataset to characterize figurative language in naturally occurring comparisons and reveal linguistic patterns indicative of this phenomenon. We operationalize these insights and apply them to a new task with high relevance to text understanding: distinguishing between figurative and literal comparisons. Finally, we apply this framework to explore the social context in which figurative language is produced, showing that similes are more likely to accompany opinions showing extreme sentiment, and that they are uncommon in reviews deemed helpful.

Learning to Recognize Affective Polarity in Similes

“…First, the interpretation of a simile can be highly context-dependent and subjective, depending on the speaker or the perceiver. To illustrate, Second, we hypothesized that the polarity of a simile might interact with the distinction made in previous work between figurative and literal uses of similes (Bredin, 1998;Addison, 1993), for example Niculae and Danescu-Niculescu-Mizil (2014) showed that sentiment and figurative comparisons are strongly correlated. Thus our expectation was that most literal comparisons would be neutral while most figurative comparisons would carry polarity.…”

“…We found that, 1) the distinction between positive and negative similes in our data is orthogonal to the figurative vs. literal distinction, 2) some similes are used both figuratively and literally, and cannot be differentiated without context, 3) even in cases when all sample uses were literal, it is easy to invent contexts where the simile might be used figuratively, and vice versa, and 4) for a particular instance (simile + context), it is usually possible to tell whether a figurative or literal use is intended by examining the simile context, but some cases remain ambiguous. Niculae and Danescu-Niculescu-Mizil (2014), whose annotation task required Turkers to label comparisons on a scale of 1 to 4 ranging from very literal to very figurative. Even with Master Turkers, a qualification task, filtering annotators by gold standard items, and collapsing scalar 1,2 values to literal and 3,4 values to figurative, the inter-annotator agreement with Fleiss' κ was 0.54.…”

“…A simile consists of four key components: the topic or tenor (subject of the comparison), the vehicle (object of the comparison), the event (act or state), and a comparator (usually "as", "like", or "than") (Niculae and Danescu-Niculescu-Mizil, 2014). A property (shared attribute) can be optionally included as well (e.g., "He is as red as a tomato").…”

Neural Multitask Learning for Simile Recognition

“…Simile is a special type of metaphor with the comparator and it is relatively easier to locate metaphorical parts. Niculae and Danescu-Niculescu-Mizil (2014) aimed to distinguish a comparison from figurative or literal in product reviews using a series of linguistic cues as features. It is similar to simile sentence classification.…”

Automatically Inferring Implicit Properties in Similes

“…A simile typically consists of four key components: the topic or tenor (subject of the comparison), the vehicle (object of the comparison), the event (act or state), and a comparator (usually "as", "like", or "than") (Niculae and Danescu-Niculescu-Mizil, 2014). For the simile "the room feels like Antarctica", "room" is the tenor, "feels" is the event, and "Antarctica" is the vehicle.…”