The ability to recognise basic emotions in music, such as happiness and sadness, is a universal skill that does not always depend on previous exposure to the musical style (Fritz et al. 2009). There is growing recognition of the variety of emotional states that music can express (Zentner et al – see my previous blog on complex musical emotions) and the speed at which we can correctly identify these states (as quick as 500ms!)
We know that the structures within music help to communicate happiness and sadness:
Sad music = soft dynamics, legato articulation, soft tempo and minor mode
Happy music = staccato articulation, louder intensities and major mode
A new article seeks to tackle the next logical question for music emotion studies – is there an additional influence of lyrics?
Most studies that present music in order to measure resulting perceptual, cognitive or affective responses stick to using instrumental works. The main justification for doing this is to avoid any confounding influences of activating the language system. This is a completely reasonable argument if your aim is to isolate the cognitive or neural processing of music. But it leaves us unable to draw conclusions about a large proportion of the world’s music – vocal music (e.g. pop, rock and folk music).
A new study by Elvira Brattico and her colleagues in Finland and Germany has looked at fMRI brain activation when people listen to happy and sad music with and without lyrics. They opted to move on from the typical use of western classical music and to use a range of music genres and timbres, as selected by 15 participants who had a wide range of musical training.
Participants selected 16 eighteen second excerpts of music: 4 sad and 4 happy that they liked and that they didn’t like. They listened to them while in a 3T fMRI scanner and rated them again for liking and emotion (happy or sad?)
RESULTS
1) Acoustic analysis: The authors analysed low level acoustic features of the music itself to determine if there were any patters that marked a piece as happy or sad. They focused on the attack slope and the spectral centroid (i.e. timbre, brightness) as well as tempo and mode. They found:
Happy music with lyrics: Faster attack than happy music without lyrics and all sad music. It also had the brightest timbres compared to all other categories.
Happy music: Brighter timbres and faster tempos compared to sad music. It was also more frequently in the major mode.
Music with lyrics: Brighter timbres than music without lyrics.
2) fMRI analysis
Sad music with lyrics: Unique activation in several brain areas that were not active when lyrics were absent, including parahippocampal gyrus, amgydala, medial and inferior frontal gyri (including Broca’s area) and the auditory cortex
Happy music with lyrics: Auditory regions alone
Happy music without lyrics: Limbic system and inferior frontal gyrus
CONCLUSION
There were few acoustic differences between music with and without lyrics. There were far larger acoustic differences between happy and sad music. The authors concluded that any differential affects driven by the presence of lyrics in the scanner would be as a result of the semantic impact of the words rather than their acoustic features.
- Lyrics are crucial for defining sadness in music. The presence of lyrics in sad music was associated with brain activations that have previously been reported in response to music chills (see previous blog), judgments of beauty, demanding speech tasks and the human “mirror neuron” system.
- By comparison, acoustic features are key to defining happiness in music. Instrumental happy music triggered more strong activations in the emotion-related limbic regions, in comparison to lyrical music.
Happy music was also associated with more left hemisphere activity, whether or not it contained lyrics. The authors explain this finding as being due to the acoustic features of happy music, including the faster attack and brighter timbres. There is growing evidence to support the theory (Zatorre et al. 2002) that the left hemisphere is not so much language dominant but dominant for sounds that contain fast specto-temporal transitions (including language but also happy music)
There were many other specific findings within the paper but that summary gives you a flavour of the major results and their interpretations. Overall, the paper gives insight into the effects of lyrics on the neural processing of human emotion in a range of musical styles and opens the door for a greater understanding of the potential effect of song, not just instrumental music, on our minds.



