Producing and understanding reduced words
Prof. Mirjam Ernestus, Professor in Linguistics
In spontaneous conversations, speakers often don’t fully pronounce words and phrases. The word yesterday is typically pronounced like yeshay and the phrase supposed to see may sound more like sussesee. This phenomenon is called reduction. “The interesting thing about reductions is that both speaker and listener are unaware of this process,” says Professor Mirjam Ernestus. “Then what mechanism underlies how speakers reduce words and listeners understand the reduced spoken language so smoothly? And why are people good at understanding reduced words in their native language but not in a foreign language? These are the central questions in our research."
To answer these questions, CLS researchers study collected recordings of spontaneous conversations between people. They discovered that the degree to which a word gets reduced is influenced by how predictable the subsequent words are. In the sentence The children swam, the word swam is highly predictable, but in the sentence The children smoked, the word smoked is highly unpredictable. Ernestus, “It’s difficult to pronounce smoked because people are unfamiliar with pronouncing the combination of children and smoked. And because this is difficult, speakers need more time to prepare the utterance and they buy time by avoiding any kind of reduction. Hence, by listening to spontaneous speech, we obtain information about the processes underlying word reduction.”
Speakers were found to use reductions because it is easier to pronounce reduced words and it reduces the time it takes to produce an utterance. In doing so, the listener’s needs and expectations are taken into account: the degree to which speech is reduced depends on the context in which speaker and hearer exchange information. Reductions are less frequent in formal meetings and public speeches compared to informal, spontaneous conversations. “Even though it’s an unconscious process, reducing speech is a sophisticated compromise between what a speaker wants and what a listener needs,” says Ernestus.
As a result of reduction processes, a given word can have many variants. In a series of experimental studies, the CLS researchers were able to demonstrate that people store information about the relative occurrences of many of these variants in their memory. That information is used in producing and understanding speech. For example, the French word pelouse (lawn) can be pronounced fully (pelouse) or in a reduced form (plouse). Native speakers of French have strong and correct intuitions about how often these variants occur. Their respective frequencies affect the speed with which speakers start pronouncing the word after being instructed to pronounce it as either pelouse or plouse. Moreover, these frequencies affect the speed with which hearers recognize the word when it is produced as either pelouse or plouse.
Ernestus, “Results like these not only provide fundamental insights into how language is processed, but they are also of major societal relevance. We found that second language learners find it difficult to understand spontaneous speech with reductions. In comparison with native speakers, second language learners lack the input of these reduced forms and have no clear intuition about how often these reduced forms occur. They are also less sensitive to subtleties in speech signals, whereas native speakers automatically notice that even in reduced words, a tiny part of the reduced sound may actually be present.” These findings are useful for the development of educational materials and courses for second language learners.
The successes achieved in recent years are a direct result of the department’s unique approach to spoken language. “Studying everyday language in spontaneous conversations has led to discoveries we wouldn’t have been able to make in a laboratory environment where people tend to produce speech more formally and often unnaturally,” says Ernestus. In the years to come, the researchers will continue their work on language in interaction by studying rhythm, stress, and intonation. Ernestus, “Our ultimate aim is to unravel how language works in interaction, how language works in the mind, and above all: how everyday language works.”