Reference Words in Text and Speech
In text and speech, we often use interesting linguistic concepts called endophora and exophora. Endophora is the process of referring to something within the text or speech. A proper noun, for example, may be mentioned once in a sentence, and in the next one, instead of repeating the same proper noun, a pronoun is used. Both the noun and pronoun actually refer to the same entity.
Exophora is the act of referring to something not mentioned in the text or speech. We can understand exophoric references by context clues or the physical environment around us. A lot of times, pronouns such as these, that, those, and this are exophoric in nature. For example, a statement such as “This is horrible!” can refer to the state of things in the present moment, but exactly what is horrible is omitted from the utterance. The speaker uses physical cues or context to convey the desired meaning.
We use endophora and exophora to avoid repetition and to vary our speech or writing. Under these broader labels are more specific ones: anaphora, cataphora, and homophora. A chart that defines the relationships between these concepts is shown below.
Anaphora is the act of referring back to an expression in a text. For example, take this sentence:"Sammy likes ice cream, and his favorite flavor is mint chocolate chip."
Notice how the possessive pronoun "his" is used to refer back to "Sammy" in this context. The concept of anaphora is often used to avoid repetition in speaking or writing (repetition in this case would be saying "Sammy" twice).
So, the order in which the proform and the entity appear in text or speech is entity first, then proform.
Anaphora can also be a temporal or spatial reference. Examples of these could include: then, last week, next year, there, here, etc. These words are often anaphoric in nature because they usually refer to a previously established time or location stated before in text or speech. Thus, anaphora binds two or more elements or phrases together in sentences.These types of phrases are fairly complex and aren’t always straightforward to language learning models. Humans naturally use and understand these phrases as essential parts of our everyday language, and with games like Lingotorium and Phrase Detectives available on the Lingo Boingo web portal, researchers can collect annotations and agreements between players on how to correctly use these concepts in text or speech, which can then be used as corpora to train language learning models.
Cataphora is similar to anaphora, but the order is reversed. In anaphora, the noun or main entity appears first, before the proform. In a cataphoric expression, the proform occurs before the noun/ main entity. This concept is more uncommon than anaphora in text or speech. An example of cataphoric expression is as follows: When she got home, Josefina went right to sleep.
Homophoric expressions are general phrases that refer to a more specific entity, and that entity is commonly known without defining it or giving context. An example of one homophoric expression is “The Queen''. This phrase usually refers to the Queen of England, and typically Queen Elizabeth.
These types of phrases present challenges to language learning models because while the understanding of entity-proform relationships and context needed to understand endophora and exophora come quite naturally to humans, it’s not so obvious to computers. To better train language learning models to process this concept, researchers need to study how humans understand this concept. With games like Lingotorium and Phrase Detectives available on the Lingo Boingo web portal, researchers can collect annotations and agreements between players on how to correctly use these concepts in text or speech, which can then be used as corpora to train language learning models.
Head to Lingo Boingo to participate in many different kinds of language-based Games With a Purpose, and make sure to check out Lingo Boingo's Twitter, Instagram, and Facebook.