An Empirical Examination of the Lay Listener Test
In Music Composition Copyright Infringment
Music recordings consist of two distinct copyright protections: (1) a copyright in the underlying composition (the “Composition Copyright”) and (2) a copyright in the sound recording (the “Recording Copyright”). The most popular test for Composition Copyright infringement, the Lay Listener Test, conflates the two by having jurors listen to sound recordings. Playing the sound recording in a Composition Copyright case invites the jurors to make the wrong comparison, comparing the sound recordings, rather than the compositional elements underlying each recording. To test this potential for prejudice, an experiment was conducted replicating the Lay Listener Test in a controlled setting. Experiment participants were presented two pairs of music from actually litigated composition infringement cases. The participants were asked to assess the similarity of the allegedly infringing compositions as would jurors performing the Lay Listener Test. One set of participants heard the songs performed similarly, i.e. same timbre, orchestration, tempo, key, and style. The other participants heard the identical compositions but performed differently, i.e. different timbre, orchestration, tempo, key, and style. Participants consistently rated compositions performed similarly as being more compositionally similar than identical compositions played dissimilarly, suggesting that the Lay Listener Test introduces prejudicial elements into the jury’s determination of substantial similarity.
Swirsky v. Carey
The first experimental group assessed the songs litigated in Swirsky v. Carey, 376 F.3d 841, 849-50 (9th Cir. 2004) for substantial similarity: "Thank God I Found You" and "Just One of Those Love Songs." Participants in the experiment were randomly assigned to either listen to pair of songs either played similarly or differently, that is with the same or different tempo, key, orchestration, and style/genre.
For instance once group listened to "Thank God I Found You" played this way:
And compared it to "Just One of Those Love Songs" played this way:
A second group listened to "Thank God I Found You" played this way:
And compared it to the original version of "Just One of Those Love Songs":
When asked whether they thought the songs were "substantially similar," based both on an ordinal scale finding of similarity (1-5 scale) and based on a standard jury instruction for the legal definition of "substantial similarity," participants were much more likely to find substantial similarity when the songs were played similarly, although the songs were legally compositionally identical. These findings suggest a systematic flaw in the results of the Lay-Listener Test for Music copyright infringement, at least as currently administered.
Gaste v. Kaiserman
The second half of the experiment required participants to analyze the songs from Gaste v. Kaiserman, 863 F.2d 1061, 1067-68 (2d Cir. 1988) for substantial similarity: "Feelings" and "Pour Toi." In Gaste, the jury made an actual finding of similarity, suggesting that the two songs are objectively very similar. Like the first half of the experiment, participants were randomly assigned to either listen to pair of songs either played similarly or differently, that is with the same or different tempo, key, orchestration, and style/genre.
Approximately half of the participants listened to "Pour Toi" played this way:
And compared it to "Feelings" played this way:
The other half of participants listened to "Pour Toi" played this way:
And compared it to the original version of Feelings:
When asked whether they thought the songs were "substantially similar," based both on an ordinal scale finding of similarity (1-5 scale) and based on a standard jury instruction for the legal definition of "substantial similarity," participants were significantly more likely to find substantial similarity when the songs were played similarly, although the contrast between the two groups of participants was not as striking as for the Swirsky v. Carey songs. These findings suggest that although method of performance strongly influences a finding of substantial similarity with two songs that are neither strongly similar or dissimilar, there is less room for gamesmanship when the songs are as nearly identical as "Feelings" and "Pour Toi."
