Béla Bartók’s Romanian music collecting

Béla Bartók collected numerous traditional melodies in villages throughout the old “Hungary” in his pursuit of the old layer of Hungarian song. In the years before WW1 he made visits to many Romanian villages in Transylvania and Banat with the purpose of collecting material as he found that in the Romanian villagers they knew only the old layer of songs and not the newer art songs that dominated elsewhere.

There are many inconsistencies in Bartók’s place names as he used Hungarian spelling for Romanian names. These are mostly clarified by Tiberiu Alexandru [1].

Bartok separates Romanian music into 4 categories by function:

  1. Colinde
  2. Funeral songs
  3. Instrumental dance music
  4. Melodies not linked to a particular circumstance

Bartók’s observations and notations give an insight into the dance music in Romanian villages, and differences between the Romanian traditional music landscape compared to that Bartók had encountered before in Hungarian and Slovak regions.

It happened to us to be able write down several hundred melodies in a single village and even get that many from one singer. In another place – even at a distance of 100 km. for example – we find the same local richness, but the set of melodies does not differ essentially from that of the previous territory. […] in the Romanian regions not only for two or three municipalities, but even for an entire district the musical material is much poorer. In a town in the county of Bihar, for example, there are barely more than forty to five forty melodies, moreover not very different from each other. And the same melodies are found in the neighbouring town. To discover others, we have to go in a region of different dialect character. There, on the other hand, the melodies from the previous region are completely unknown, and so on. Ultimately, it follows that, if we take into account throughout the Romanian territory, the Romanian folk melody is no poorer than those of the Hungarians or the Slovaks. We would be tempted to say that Hungarian or Slovak music is more rich and varied in a “vertical” direction, that of the Romanians, in a “horizontal” direction.

Bartók 1936:217-218 [2]

This theme of the difference in distribution and variety of traditional Romanian music compared to his previous Hungarian and Slovak work is discussed in Bartók’s writings on Romanian music. His search for the old Hungarian pentatonic melodies has a consistency throughout old Hungary, in the mutual exchange with the Slovaks. Whereas, Romanian music has dialects that are almost independent, and mostly (the exception being central Transylvania) show no loans from their long time Hungarian neighbours.

My particular interest is in dances documented in Banat

  • Sannicolau Mare (Jan-1910);
  • Murani (Mar-1912);
  • Igriș, Saravale, Vălcani (Nov-1912);
  • Banloc, Ghilad, Jebel, Livezile, Alibunar, Seleuș, Vladimirovac (Dec-1912);
  • Cenadu Mare, Cornești, Foeni (Feb-1913).

The popup in the map below on each location shows the dance melodies recorded at that location.


Bartók adopted and continuously perfected the musical theory of Ilmari Krohn which orders the collected material based on categories, metrics, in a sequence from simple to the complex, from the number of melodic lines, the main cadence and the cadences of the other melodic lines [3].

Bartók understood the process of “tradition” and “variation” so employing this method allowed material to be classified relative to the other material.

The melodies transmitted from generation to generation change more or less, depending on the region, and give rise to variations of melodies. On the other hand, melodies, originally of different constructions, are transformed and become similar to each other: the result is homogeneous musical styles with a well-defined character.

Bartók 1936:197 [2]

The dance melodies are put in two classes: [4]

  • Class A includes those of determined structure (consisting mostly of four melody sections),
  • Class B includes those in which motifs (mostly of two or three bars) are repeated or follow each other without apparent system.

His method, might be termed caesura tone scheme and syllabic scheme [5], but is actually quite simple. Most traditional songs are in 4 phrases. Bartók documented by considering the ending note of each phrase relative to the final note of the melody in the belief that this is one of the most stable features of a melody type. Hence he categorised by the sequence of phrase ending notes (mostly 3) in numerical interval from the final note.

Secondly, the melodies are classified most broadly into “isometric” tunes (tunes with the same number of syllables per line) and “heterometric” tunes (tunes that set text lines of different length). These syllables are not representing a rhythm, just the number of syllables in the song words, and assumes some stability in the number of syllables per phrase, although recognising that developments can merge two, or divide one into two, through evolution.

Bartók’s conclusions

Bartók’s concluding comments are deeply interesting in the positioning of Romanian cultural communities in the shared space of old Hungary.

He discusses the Romanian regions he researched in broad outlines, the regions from Banat, from Bihor, from Hunedoara, from central Transylvania (Mezőség/Câmpia) and Maramureș (including Oaș).

The Romanian melodies are dissimilar across geographically regions, compared to the old Hungarian melodies that appear to be consistent throughout old Hungary (and Bartók attempts to further link these to the Finno-Uralic Mari of the Volga region in Russia). Bartók finds the musical repertoire totally different in each zone, but later he adds the confusion of intermediate forms between zones. Regarding the dance music, the local dance melodies of Bihor, Hunedoara, Banat and Maramureș have nothing in common with the popular Hungarian melodies [2].

Bihor melodies with text do not indicate even the remotest Hungarian influence. In fact, the whole area looks as if shut off by a “Chinese wall” from the adjacent Hungarian territory. (Bartók, 1967, 48).

Bartók 1967:48 [4]

The situation is completely different in central Transylvania (Mezőség/Câmpia). Here Bartók considers the Romanian song uses the old pentatonic melodies of the Szekler people. As Romanian songs are nearly always composed of 8 syllable verses, when a 10 or 11 syllable melodies are borrowed from the Hungarian repertoire some extra syllables like “tra-la-la” are added to the end of the text [2].

Some of the Romanian dance tunes from central Transylvania are very similar to the Hungarian melodies called verbunkos, and part of this fund of melodies are designated by the word “Țiganească” (Roma), which would seem to indicate that these melodies were propagated by Rom musicians [2]. The more recent Hungarian melodies with a “punctuated rhythm” and neo-Hungarian melodies have no influence on Romanian melodies. This shows a sharing of the old layer of Székelys songs, but no borrowing of later Hungarian melodies.

Regarding Maramureș, Bartok has an unresolved difficulty trying to place Maramureș melodies in his framework based on the concept of old Hungarian traits and widespread sharing in mixed ethnic regions. He notes a few potential Hungarian borrowings, but has problems attempting to relate this via Ruthenians back to Hungarian. He also notes a diffusion of these Maramureș songs into Transylvania.


When using Bartók’s invaluable work we have to remember that he only had sparse samples of music from the areas he visited. Such an sparse catalogue of examples can lead to incomplete assumptions (such as the discussion on hora lunga type songs which his visit to the Brăilou archives in Bucharest in 1934 resolved) [6]. His discussion on asymmetric rhythms is logical, but from a position of not previously having encountered such rhythms.

As an outsider it is clear that his musical knowledge and understanding was different to that of  the villagers of Pădureni (Hunedoara) which leads to some interesting quotes showing the complexities of researching within a foreign community:

This means that pieces with the same motifs may be performed either as De doi or as De danț. The tempo of both genres is identical. Does all this indicate that there is no musical difference between De doi and De danț music? How can dancers then distinguish them if they are musically identical? Nonetheless, they are able to do so, since choreography of the two genres is completely different!

Bartók 1967:51 [4]

The fluier pieces of Hunedoara.—This section is, unfortunately, an anti-climax because of the dullness of the material it describes. […] Most of them have such inconsistancy of form that each of four or five renditions of the same piece will show a different structure. Moreover, the structure itself of the single renditions has frequently such unsymmetrical, almost incomprehensible form, the melodies such awkward, dull, and I would say senseless patterns, that one is at a loss how to evaluate them. Had I found only one player playing such melodies, I would have thought him a rather bad and unreliable performer and would have, perhaps, omitted his contribution. But as a score or so of performers played in the same manner (or with the same mannerisms), I had to accept the phenomenon as characteristic.

Bartók 1967:51 [4]


  1. Alexandru, Tiberiu. 1957. Despre localităţile din care Bela Bartok a cules muzica populară românească. Revista de Folclor Anul 2 Nr 1-2:169-178.
  2. Bartók, Béla. 1936. "La musique populaire des Hongrois et des peuples voisins." Archivum Europae Centro-Orientalis II:197-244.
  3. Câmpeanu, Mircea. 2010. "Béla Bartók şi etnomuzicologia românească." In Revista Traditii Clujene - Tezaure Umane Vii Nr.5, 40-58. CJCPT Cluj.
  4. Bartók, Béla. 1967. Rumanian folk music. Volume 1: Instrumental melodies. Vol. 1. The Hague: Martinus Nijhoff.
  5. Gollin, Edward. 2008. "On Bartók's Comparative Musicology as a Resource for Bartókian Analysis." Intégral 22:59–79.
  6. Brauner, Harry. 1979. Sa auzi iarba cum creste. Bucharest: Editura Eminescu.