Music structures influence sentence interpretation: musical primes shift relative-clause attachment in English

Music structures influence sentence interpretation

Although theorists have long argued about possible connections between language and music (Jackendoff 2006, Patel 2008, Katz and Pesetsky 2011), relatively few psycholinguistic experiments have examined whether music can influence language processing. More specifically, scientists have asked whether structural patterns in music affect how people process grammatical structures in sentences. If the two domains share cognitive resources, then hearing certain musical sequences should lead to corresponding effects in language.

Van de Cavey and Hartsuiker (2011) found early support for this idea. They presented series of sine-wave tones of varying pitch to Dutch speakers and observed priming effects on how people interpreted relative clauses. The new experiment extends that work.

Testing English relative-clause bias

The study had two central goals. The first was to see whether sounds that listeners easily recognize as musical — notes played on a piano — also trigger priming during sentence processing. The second was to check whether inserting a pause strengthens any priming effects, since relative-clause attachment is known to be sensitive to prosodic phrasing.

English was chosen because it shows a low-attachment bias for relative clauses (the clause attaches to the closest noun phrase), unlike Dutch, which favors high attachment (to a more distant noun phrase). Important research by Mitchell and Brysbaert (1998) established this cross-linguistic asymmetry. Testing English thus allowed a check on whether low-attachment and high-attachment languages differ in their primability by music.

Participants listened to short piano melodies and then completed sentence fragments. Their completions revealed which noun they thought the relative clause described.

Experimental design

The experiment used a 2x2 design with two factors:

• Attachment height: high attachment (HA) vs. low attachment (LA)
• Pause: pause present vs. no pause

Twenty participants saw 30 target items and 120 fillers. The musical primes were built using the Circle of Fifths, a standard harmonic framework where tones from various keys combine into permissible chords and called progressions.

Music prime structure

Each melody had eight notes. In high-attachment primes, the final two notes moved back toward the initial harmonic region. In low-attachment primes, an initial and a second section existed, and the last two notes attached back into the second harmonic domain. Notes were constant in duration and interval, except that in the Pause condition the final two notes were preceded by a rest; the No-Pause condition lacked this silence.

Linguistic target items

Participants read incomplete sentences such as Kevin counted the fans of the singer who and then had to continue them. The counts (singular or plural) of the two nouns were counterbalanced. The researchers examined whether completions attached to the complex noun phrase — the fans of the singer, high attachment — or to the lower noun — the singer, low attachment.

Priming effects and role of pauses

The results showed a clear effect: Musical HA primes produced noticeably more high-attachment continuations — 38% on average — compared with LA primes, which gave only 23% high-attachment completions (p < 0.01). No significant difference emerged between the Pause and No-Pause conditions, and the interaction between pause and attachment was not reliable (both p values > 0.3).

These findings demonstrate that explicitly musical stimuli — notes from a familiar instrument — can activate hierarchical structures that partly overlap with syntactic structures in language. The presence or absence of a pause did not strengthen or weaken the bias one way or the other. That runs counter to some earlier claims (Fodor 1998) about how prosodic phrasing affects relative-clause resolution. The implication is that rhythmic breaks are not required for the priming effect to appear; instead, what matters is the organization of the harmonic domains themselves.

Shared abstract representations

Taken together, this experiment yields strong evidence that music and language rely on domain-general mechanisms for representing hierarchical structural information. The outcome confirms claims made by Patel (2003, 2008) that the two human capacities share resources at an abstract, structural level. When listeners hear note sequences that imply one attachment structure, that structural pattern influences the grammatical choices they make when interpreting sentences — even though the rules of tonal harmony and the rules of syntax apply to very different material.