A good movie line is the best kind of earworm — but have you ever wondered how these quotes get stuck in your head in the first place?
By analyzing upwards of 2,000 oft-quoted movie lines, a team of computer scientists says it's uncovered how word choice and sentence structure give rise to cinema's most memorable soundbites. Their findings, they say, could even be used to fine tune the catchphrases of tomorrow.
The researchers' results are presented in an entertaining and approachable article titled "You had me at hello: how phrasing affects memorability." It is definitely worth checking out when you have the time, but the gist of the study is as follows:
Using IMDB as a source for catchy quotes, Cristian Danescu-Niculescu-Mizil and colleagues at Cornell University rounded up memorable lines from around 1,000 movies and paired them with other lines spoken by the same characters in the same movies, spoken at roughly the same time. "Highly cited quotes," the researchers write, "tend to have been delivered under compelling circumstances or ﬁt an existing cultural, political, or social narrative"; comparing memorable quotes against non-memorable ones, explain the researchers, allowed them to control for the cinematic context in which the lines were delivered.
The researchers then presented volunteers with quote pairs from movies they hadn't seen, and asked them to identify which of the two phrases was the "memorable" one (you can try the test out for yourself on the researchers' website). They found that people picked the memorable quote 78% of the time.
The researchers then compared the memorable phrases against a text collection of typical American English (called the Brown Corpus) to identify what sets them apart from "common language."
The researchers noted several interesting patterns. For example, memorable quotes tend to mix atypical word choice with simple, straightforward sentence structure. They also noticed that more memorable phrases were "easier to use outside the specific context in which they were uttered." In other words, they possess the quality of "generality." They found a quote's generality to be dependent on three main qualities:
1. The fewer personal pronouns the better (second-person pronouns excluded). Personal pronouns imbue a quote with specificity by referencing a person introduced earlier in the discourse.
2. Utterances containing indefinite articles (like "a" and "an") are more memorable than those containing definite articles like "the."
3. Avoid the past tense. Like personal pronouns and indefinite articles, verbs in the past tense are more likely to reference something specifically, in this case a previous event.
Finally, the researchers designed a computer model that could analyze sentences based on their generality, their uncommon word choice, and their straightforward sentence structure. When they ran the computer through its paces, they found it could distinguish between memorable and non-memorable quotes an impressive 64% of the time.
It's easy to envision how a program such as this might one day be used by screenwriters to craft the perfect one-liner, or by advertising executives to fine-tune a slogan. (When Danescu-Niculescu-Mizil and his colleagues examined catchphrases from advertisements, they found they had the same properties of generality, unusual word combinations and simple syntax as the memorable movie quotes.) But is the human mind really that predictable? Can the political focus group be replaced by a program? Will the clever quips and snarky comebacks of tomorrow's summer blockbusters be churned out by a computer algorithm?
As Captain James T. Kirk once said: "We humans are full of unpredictable emotions that logic alone cannot solve." Perhaps he was mistaken.
The researchers' findings are available free of charge over on arXiv. Image via DeviantART.