All entries for Sunday 30 September 2018
September 30, 2018
Non-linguists often use the term "filler words" as a convenient catch-all for utterances that, for various reasons, don't meet arbitrary prescriptivist rules for "correct" speech.
"Pseudo-scientific" is overly kind. The study was apparently conducted by a marketing firm on behalf of a website, so it's really just a company taking advantage of the press for free advertising. This is a cursory list of obvious problems with the study.
Perceptions of language ≠ productions of language
The "study" surveys people's perceptions of their usages of linguistic features and presents them as facts about their productions of linguistic features. The absurdity of this should be obvious, since the "study" finds that 23 percent of people "don't realise they're using [filler words]" at the same time that it reports situations and demographics where people realise they're using "filler words." In other words, the "study" says, "People don't know when they say err, and those same people say they say err when they're on a date." You can't study what people actually do by asking them what they do. You must study them doing the thing.
The "study" design doesn't meet basic scientific standards
The "study" apparently showed people two videos of the same person giving talks about Brexit, one factual speech with "filler words" and one false speech without "filler words." As a basic methodological principle, you cannot tell people what an experimental condition is and then examine their responses to it. The method in this "study" of language is analogous to giving a person a trial medication and a placebo, telling them which is which, and then asking them whether they feel like the trial medication or the placebo worked better. That's obviously bad science. Beyond that, we'd need to know how the researchers identified the "filler words" included and whether there were other cues to the speakers' perceived confidence (like body language), not to mention conflating factors like informant demographics and attitudes toward Brexit. And the genre of speech matters, too: people like respond differently to someone using "filler words" in a situation where prepared remarks are expected (like a speech on major international policy) than they do in a spontaneous conversation.
The researchers don't know language
The "study" problematically conflates many different kinds of language under the label "filler words," revealing that they have no knowledge of the tremendous body of science about language. For instance, there are many different versions of the word like:
- He was like forty years old.
- I was like, "This is not good."
- Like, what do you want to do?
In (1), like carries an approximer function, so that the man is 'approximately, but not exactly, forty years old.' In (2), like introduces a quotation that captures the spirit and general idea of what was said, but may not be literally word-for-word. In (3), like functions as a focuser, which points the hearer toward the importance of the clause that immediately follows it.
These are not fresh insights into like. There's a whole Wikipedia page about like with a great bibliography of some of the work linguists have done on it. The fact, though, that this marketing gimmick conflated these very different forms of like under the heading "filler words" reveals the lack of knowledge about language and linguistics.
Indeed, there's real science behind every word that this "study" calls "filler." umm, for instance, is a hard-working discourse particle that helps govern conversation. People say umm in part to signal that they're still talking and that it's not yet someone else's "turn" to talk. Linguists refer to this as "floor management"--managing who "holds the floor" during conversational turn-taking. innit isn't "filler" at all. It is an innovation in British English that functions like a "tag question," but also indicates a speaker's strong feelings about the statement they've just made. It's lumped in as "filler" by the "study," because innit is associated with young people and the speech of young people is socially stigmatized.
"filler words" and future study
I take three points from reading the reports of this "study" in British media:
- Media must remain vigilant to being duped into handing out free advertisements to companies on the basis of phony science.
- Linguists have much more work to do to share scientific knowledge about language with the public.
- There are no "filler words" in language. Every linguistic utterance means something and is doing something to make language work.