Language and musical capacity of the infant brain

During the last trimester of the pregnancy, hearing is developed to the extent that the foetus reacts to vowels.Reference Abrams and Gerhardt⁹ Consonants have higher frequency and less intensity than vowels and are, as far as we know, not detected. French newborns are able to discriminate between English, a stress-language with irregular rhythm, and Japanese, a pitch language with regular rhythm, but fail to discriminate between English and Dutch, which are both stress languages, suggesting that newborns use prosodic or – more specifically – rhythm information to classify speech segments into broad language classes.Reference Nazzi, Bertoncini and Mehler¹⁰ Sleeping newborns extract prosody from continuous speechReference Sambeth, Ruohio, Alku, Fellman and Huotilainin¹¹ and two-day old hearing infants with either hearing or deaf parents all prefer consonance over dissonance, presumably essentially an innate predisposition rather than dependent on special prenatal or early postnatal experience.Reference Masataka¹²

Neonates and infants show a right ear (left hemisphere) advantage for speech and left ear (right hemisphere) advantage for music.Reference Bertoncini, Morais, Bijeljac-Babic, McAdams, Peretz and Mehler^13, Reference Best, Hoffman and Glanville¹⁴ Most infants show an interest in music from their early days and they pay more attention to the singing than the speaking of their mothers.Reference Trehub^15, Reference Trehub¹⁶ They also have a bias for speech compared with non-speak sounds.¹⁷ Language input to infants has culturally universal characteristics such as louder and more extreme vowels, and is syntactically and semantically simpler than speech directed to adults. By six months of age, speech perception abilities are attuned to the vowels of the native languageReference Kuhl, Williams, Lacerda, Stevens and Lindblom¹⁸ and the infant’s consonant perception is attuned to the native language by 10–12 months of age.Reference Kuhl, Stevens, Hayashi, Deguchi, Kiritani and Iverson¹⁹ Similarly, lullabies share some cross-cultural features. Increasing musical exposure attenuates the effects of culture-general factors while amplifying the influence of culture-specific factors.Reference Schellenberg and Trehub²⁰ KuhlReference Kuhl²¹ has stressed the importance of emotional and social contact with the mother for optimal acquisition of language in infants. The recent demonstration that short sentences can activate Broca’s area in infants already at two month of age – before the babbling stage – implies that this is not a consequence of learning to speak but rather that this region may drive the learning of the complex motor sequences required for speech production when exposed to speech.²² It would be extremely interesting to have a corresponding study performed that substitutes sentences with short pieces of music.

Brain plasticity related to learning to read and musical training

Brain plasticity implies that the brain is adaptable and includes all the mechanisms responsible for the brain capacity to change in response to incoming stimulation, our activities and even thoughts, from genes to cortical network.Reference Pascual-Leone, Ameri, Fregni and Merabet^23, Reference Johansson²⁴ That neuronal circuits are tuned in close interaction between genes and the environment is evident from the global variations in customs, languages and culture. That one effortlessly learns to talk and to sing while being exposed to language and music is an example of developmental plasticity. Implicit knowledge of basic harmonic functions in music is evident and relatively complete among children of 6 and 7 years of age that have no musical training.Reference Schellenberg and Trehub²⁰ Although developmental plasticity needs no specific training, the competence will be highly related to input and environmental stimulation. In contrast, reading, writing and playing an instrument are cultural innovations that require specific training, and that involve training/learning-induced plasticity. In a study of American subjects whose ages ranged from 6 to 22 years, learning to read and write was associated with two patterns of change: increased cortical activity in the left-hemisphere middle temporal and inferior frontal gyri, and decreased cortical activity in the right infero-temporal areas. The activity in the left-posterior temporal gyrus of the youngest readers was associated with the maturation of their phonological processing capacity.Reference Turkeltaub, Gareau, Flowers, Zeffiro and Eden²⁵ Basic mechanisms for language processing are universal but there are cultural modifications related to the structure and orthography of the language, and in bilinguals changes are also related to when the second language was acquired.Reference Johansson²⁶ In studies of literate and illiterate Portuguese subjects, the illiterate groups were consistently more right-lateralized than their literate controls in the inferior parietal cortex but had a similar degree of left–right differences in early speech-related regions of the superior temporal cortex, suggesting that literacy influences the functional hemispheric balance in reading and verbal working memory-related regions. In addition, voxel-based morphometry showed differences in white matter intensities in the mid-body region of the corpus callosum and the inferior parietal and parieto-temporal regions (literate > illiterate). The lack of corresponding differences in the grey matter suggested that the influence of literacy on brain structure was affecting large-scale brain connectivity rather than grey matter per se.Reference Petersson, Silva, Castro-Caldas, Ingvar and Reis²⁷

Musical training involves complex motor, auditory and cognitive skills, often requiring memorizing long, complex bimanual finger sequences, and translating musical symbols into motor sequences during sight-reading. Continued practice of complicated bimanual motor activity leads to changes in the brain structure and the functional reorganization of cortical regions and connections between different regions. String players have an increased representation for the little finger of the left hand, which correlates with the age at which the person started training,Reference Elbert, Pantev, Wienbruch, Rockstroh and Taub²⁸ and musicians have a larger representation in the auditory cortex for tones, pitch and timbre than non-musicians.Reference Pantev, Oostenveld, Engelien, Ross, Roberts and Hoke^29, Reference Pantev, Roberts, Schulz, Engelien and Ross³⁰ The motor cortical excitability and synaptic plasticity is enhanced in professional musicians, and correlate with the hours of daily practise and age of starting to play.Reference Rosenkranz, Williamon and Rothwell³¹ An example of how extensive training can shape cognitive auditory processes in the adult brain is the superior auditory spatial tuning in conductors, compared with pianists and non-musicians.Reference Münte, Kohlmetz, Nager and Alternmüller³² Studies on brain organization in musicians have provided much further information on brain plasticity in the developing and adult brain.Reference Pascual-Leone^33–Reference Zatorre, Chen and Penhune³⁷

This review is based on data from healthy individuals. Although studies on individuals with brain lesions have given interesting information there are confounding factors that need to be considered. One is that unilateral brain lesions are associated with lesion-induced plasticity, a dynamic process leading to time-related cortical reorganization in the ipsi-and contra-lateral hemisphere that can be modified by training and other interventions.Reference Johansson²⁴

Auditory processing in the cerebral cortex

Cortical asymmetries may have been developed as a solution to the need to optimize processing of the acoustic environment in both temporal and frequency domains.Reference Zatorre and Belin³⁸ Temporal differences as small as 20 ms are needed for perception of consonants in language, whereas melodies with note durations shorter than about 100–150 ms are difficult to identify. The asymmetric sampling in time theory proposes that auditory cortices preferentially sample at rates tuned to fundamental speech units, with the left auditory cortex preferentially integrating auditory signals into 20–50 ms segments that correspond roughly to the phoneme length, while the right auditory cortex would preferentially integrate segments of 100–300 ms, thus optimizing sensitivity to slower acoustic modulation, such as voice and musical instrument periodicity, speech prosody and musical rhythms.Reference Poeppel³⁹

With the basic concept that the acoustic speech network must interface both with conceptual systems and with motor-articulatory systems, a dual-stream model for the cortical organization of speech processing has been proposed with a dorsal stream lateralized to the left hemisphere for sound to action and a bilateral ventral stream for sound to meaning.Reference Hickock and Poeppel⁴⁰ In an attempt to synthesize available data and critical elements of earlier models, Ben Shalom and Poeppel have proposed a functional neuro-anatomy model based on three different aspects of language processing, namely phonological, syntactic and semantic processing, and three different types of operations underling various aspects of language processing, namely memorizing (or retrieval of stored items), analysing and synthesizing.Reference Ben Shalom and Poeppel⁴¹ It suggests that the temporal lobe deals principally with memorizing (storing) lexical items, the parietal lobe with analyzing these items, and the frontal lobe with synthesizing the information. A further suggestion is that there is a superior-to-inferior gradient in each lobe, with phonetic/phonological information being mapped in more superior/dorsal fields, syntactic information in the middle, and semantic representations more inferiorly/ventrally.

Koelch and SiebelReference Koelsch and Siebel⁴² have presented a cognitive model of the neural basis of music perception, built on the time-relation of different components, as shown in different ERP-responses, and on their relation to meaning and emotion. The neurophysiologic data indicating that the semantic processing may be identical for music and language are clearly presented, and the role of sub-cortical analysis of the acoustic input is pointed out. Referring to studies on infants and to published data on overlapping and shared neural resources for music and language, the authors comment that it appears that the human brain does not treat language and music as strictly separate domains.

Congenital amusia

The differences in the sound processing between music and language are demonstrated in the effect of cochlear implants in children and adults. While the implants provide many deaf patients with good speech understanding, music perception remains a major challenge, although interestingly preliminary data indicate that some improvement may be achieved by intense melodic contour training, another example of brain plasticity.Reference Galvin, Fu and Nosaki⁴³ The right secondary auditory cortex has a finer pitch resolution than the left.Reference Hyde, Peretz and Zatorre⁴⁴ Congenital amusia, which affects 4–5% of the human population, has been attributed to a deficit in fine-grained pitch.Reference Patel^5, Reference Peretz, Ayotte, Zatorre, Mehler, Ahad, Pentune and Jutas^45, Reference Foxton, Dean, Gee, Peretz and Griffiths⁴⁶ Voxel-based morphometry has indicated a reduction of white matter in the fibre tract connecting the auditory cortex and the right inferior frontal gyrus implicated in musical pitch encoding and melodic pitch memory.Reference Hyde, Zatorre, Grifiths, Lerch and Peretz⁴⁷ However, individuals with amusia, popularly named tone-deafness, have difficulty also in discriminating intonation contours extracted from speech, suggesting that the dissociation between spoken and musical pitch perception in tone-deaf individuals is due to a deficit at a higher level than simple pitch-change detection.Reference Patel, Foxton and Griffiths⁴⁸ A further study has revealed significant co-variations between performance and grey matter volume in language areas in the left hemisphere in regions that correlated with performance on melodic and rhythmic subtests, and the authors proposed that congenital amusia could be an auditory-motor feedback disorder.Reference Mandell, Schultze and Schlaug⁴⁹ There is also evidence that congenital amusia is strongly related to a deficit in spatial processing, suggesting that the processing of pitch in music depends on cognitive mechanisms that are normally used to process spatial representations in other modalities.Reference Douglas and Bilkey⁵⁰ Apparently there are more problems with congenital amusia than those related to music.

Syntax and meaning

Syntax in language and music refers to the principle governing the combination of discrete structural elements into sequences. Comparisons between syntax in language and music, so far performed only on European Western tonal music, have demonstrated that violations of linguistic and musical syntax activate similar networks in the brain including Broca’s area.Reference Koelsch, Gunter, Wittfoth and Sammler^7–Reference Nazzi, Bertoncini and Mehler¹⁰ Thus, although language and music have distinct and domain-specific representations (chords versus words) they share neural resources for activating and integrating these representations during syntactic processing,Reference Patel⁸ a hypothesis that has been supported by the finding that children with specific language impairment are impaired also in music-syntactic processing.Reference Jentschke, Koelsch, Sallat and Friederici⁵¹ The rule-violation response that is left lateralized in language is often bilateral in music, sometimes with a trend to right dominance, and a differential hemispheric dominance in detecting unexpected errors in the lyrics (left) and melody (right) of memorized songs has been observed.Reference Yasui, Kaga and Sakai⁵²

Meaning in music is a controversial topic, and music clearly does not have meaning in the way language has. The broader definition that meaning exists when perception of an object/event brings something to mind other than the object/event itself, proposed by Nattiez,Reference Nattiez⁵³ allows for individual interpretations that could include the expression and appraisal of emotion, individual memories and associations, and the combination of linguistic and musical meaning in songs. Koelsch and SiebelReference Koelsch and Siebel⁴² provide further examples. Neurophysiologic markers (ERP responses) suggest that music can activate certain semantic concepts. Whereas musical and linguistic syntax mutually affect one another, only linguistic semantics affects musical semantics and not the other way round, suggesting that the semantic aspects of language are considerably more salient than those of harmonic irregularities.Reference Abrams and Gerhardt^9, Reference Steinbeis and Koelsch⁵⁴ A double dissociation between syntactic and semantic knowledge is well established in language. A similarly double dissociation between rules and memory, independent of musical training, has been demonstrated for music, suggesting that cognitive rule/memory dissociation extends from language to music.Reference Miranda and Ullman⁵⁵ An fMRI study of verbal and tonal working memory indicated that both rehearsal of verbal and tonal information (syllable and pitches), and storage of verbal and tonal information rely on strongly overlapping neuronal networks that appear to consist partly of sensori-motor-related circuits that provide resources for the representation and maintenance of information, and that are remarkably similar for the production of speech and song.Reference Koelsch, Schulze, Sammler, Fritz, Müller and Gruber⁵⁶

The role of the right hemisphere in language

The right hemisphere might play an important role when it comes to comprehending complex, natural language such as stories, text and conversations. Accumulated evidence from neurophysiology, neuro-imaging and anatomy suggests three roughly separate but highly interactive components of semantic processing; semantic activation, semantic integration and semantic selection (in posterior middle/superior temporal gyri; anterior middle/superior temporal gyrus and inferior frontal gyrus respectively), with each proposed to have bilateral components with the right hemisphere performing coarser computations for the same general process.Reference Jung-Beeman⁵⁷ The right hemisphere has been shown to be particularly important to metaphor comprehension when both hemispheres work in concert in a complex dynamic pattern.Reference Arzouan, Goldstein and Faust⁵⁸ The right hemisphere is more sensitive than the left to unfamiliar metaphoric relations but less so to familiar semantic relations.Reference Marshal and Faust⁵⁹ The hemispheric difference in ambiguity resolution during language processing is related to the fact that the right hemisphere maintains a broader and less differentiated range of related meanings, including both dominant and subordinate meanings of ambiguous words, than the left hemisphere. The differences in ambiguity resolution extend also to verbal memory.Reference Faust, Ben-Artzi and Harel⁶⁰ The specialization of the right hemisphere in the activation of a broader range of related meanings, including novel, non-salient meanings than the left hemisphere has been confirmed in a TMS study suggesting that the right hemisphere is critically involved in the integration of the individual meanings of two seemingly unrelated concepts into a meaningful metaphoric expression.Reference Pobric, Marchal, Faust and Lavidor⁶¹

Interaction between music and language

Music perception skills correlate with phonological awareness and reading developmentReference Anvari, Trainor, Woodside and Levy⁶² and singing facilitates language acquisition.Reference Schön, Besson, Boyer, Moreno, Besson, Peretz and Kolinsky⁶³ Music training facilitates pitch processing in both music and languageReference Magne, Schön and Besson⁶⁴ and is associated with an enhanced ability to perceive prosody in speech.Reference Thompson, Schellenberg and Husain⁶⁵ Both young and adult musicians detect pitch violation in a foreign language better than do non-musicians.Reference Marques, Moreno, Castro and Besson⁶⁶ Musicians have an enhanced sub-cortical auditory and audiovisual processing for both speech and music,Reference Musacchia, Sams, Skoe and Kraus⁶⁷ and musical experience shapes human brainstem encoding of linguistic pitch patterns,Reference Wong, Skoe, Russo, Dees and Kraus⁶⁸ which could be one explanation for the higher learning ability. Another explanation, culled from a cross-sectional and longitudinal study in children, could be that music training improves verbal memory,Reference Ho, Cheung and Chan⁶⁹ an important factor in language learning.

Interaction between language, music and the motor system

The effect of musical training on the motor system has already been referred toReference Elbert, Pantev, Wienbruch, Rockstroh and Taub^28, Reference Rosenkranz, Williamon and Rothwell^31, Reference Zatorre, Chen and Penhune³⁷ but also non-musicians extensively activate bilateral motor areas when listening to music.Reference Popescu, Otsuka and Ioannides⁷⁰ Considering the hypothesis that language evolved from manual gestures,Reference Liberman and Mattingly^71, Reference Coballis⁷² a comparison with the motor system is very relevant, and there is evidence for close connections.Reference Pulvermullezr, Hauk, Nikulin and Ilmoniemi⁷³ Listening to speech specifically stimulates the tongue muscles,Reference Fadiga, Craighero, Boccino and Rizzolatti⁷⁴ and language perception activates the hand motor cortex.Reference Floel, Eliger, Breitenstein and Knecht⁷⁵ A recent study, following a new approach, is compatible with this hypothesis. Unrelated to any specific input or output, neural oscillations mainly reflecting synaptic activity are abundant in ongoing spontaneous brain activity patterns. Recent research has shown that these oscillations are not random but, at least to a large extent, are related to known anatomical systems.Reference Fox and Raichle⁷⁶ The hypothesis that oscillations in speech relevant frequency bands are linked to regional activity in areas specialized for speech perception and production was tested by simultaneous registration of EEG and fMRI in resting individuals not subjected to any tests or auditory input. Across multiple time-scales, spontaneous variations within the gamma range (>30 Hz, phonemic rate) correlated best with left auditory cortical synaptic activity, while fluctuations within the theta range (4–7 Hz, syllabic rhythms) correlated best with that in the right auditory cortex. Power fluctuations in both ranges correlated with activity in the mouth pre-motor region, indicating coupling between temporal properties in speech perception and production.Reference Giraud, Kleinschmidt, Poeppel, Lund, Frackowiak and Laufs⁷⁷

Concluding remarks

The brain is a multi-sensory organ, and sensory integration is necessary and occurs at many different levels. Based on accumulating evidence for multi-sensory processes involving vision, audition, and somato-sensation across the neo-cortex it has been proposed that multi-sensory influences are integral to primary as well as higher order cortical operations.Reference Ghazanfal and Schroeder⁷⁸ Music and language clearly interact with other cognitive systems. Perception of auditory speech is improved when the speaker’s facial articulatory movements are visible.Reference van Wassenhove, Grant and Poeppel⁷⁹ In musical performance, many cognitive functions must be integrated at a high speed. A study on the brain activities in professional singers during overt and imagined singing of an aria provides a detailed and illustrative example of multi-sensory integration.Reference Kleber, Birbaumer, Veit, Trevorrow and Lotze⁸⁰ Historically, emotion and cognition have been viewed as separate entities, but complex cognitive-emotional behaviour emerges from the rich dynamic interactions between brain networks, and there is no truly separate system for emotion and cognition.Reference Pessoa⁸¹ Against this background it is hard to agree with the opinion that music is made up of multiple models that overlap minimally with other functions such as language.Reference Patel⁵ On the other hand, we know that individual neurons have a high degree of selectivity,Reference Conway, Moeller and Tsao⁸² and how the local networks are integrated in the multi-sensory processing is a key issue.

Neuro-anatomic evidence indicates that the brain evolved by preserving, extending and combining existing network components, rather than by generating complex structures de novo.Reference Sporns, Chialvo, Kaiser and Hilgetag⁸³ Experimental data on macaque networks have allowed reliable identification of hubs (nodes), many of which have previously been functionally classified as poly-sensory or multimodal. Using a computation approach relating structure to function, the model suggests that the cortical resting state contains rich and interrelated temporal structures at multiple time scales that are shaped by the underlying structural topology, and they illustrate the structural bias for the coexistence of functional segregation (modularity) and functional integration.Reference Honey, Kötter, Breakspear and Sporns^{84, 85} Research along this line has also led to a hypothesis regarding evolution with empirical predictions that are being testedReference Anderson⁸⁶ and could maybe sometime in the future tell us more about the relationship between music and language. A neuronal recycling hypothesis, according to which cultural inventions such as reading and arithmetic invade evolutionary older brain circuits and inherit many of their structural constraints, has been proposed for changes in cortical maps related to reading and arithmetic.Reference Dehaene and Cohen⁸⁷

Current research on language and music models are to a large extent based on non-musicians and musicians, with all subjects having been exposed to the training-induced cultural factor of reading, and the musicians to musical training. These studies have been very successful, and they have given us much information on different subcomponents in music and language. However, from an evolutionary point of view, the most interesting time for comparing music and language is, in my opinion, the early years before children learn to read or play music. As reported here, we do have evidence for similarities and interaction in infants and young children, and for the fact that early musical training might facilitate reading. Evidence from a recent review comparing sounds in infants and non-human primatesReference Masataka⁸⁸ would seem to support Darwin’s hypothesis that at an intermediate stage in human evolution there functioned a communication system that more closely resembles music than language and that may possibly have acted as a precursor for both current language and music. This is not a question we are likely to solve in the near future, but studies on the early development of language and music, as well as research on bilingual infants,Reference Werker and Byers-Heinlein⁸⁹ are likely to tell us much about the brain’s organization and post-natal development.