If you hear spoken Swedish and Norwegian, you would find that these languages tend to sound considerably more melodic compared to many of their Germanic counterparts. This difference is predominantly attributed to their ‘pitch-accent’ feature, where certain stressed syllables tend to come with some form of tone, or more rather, accent.

To recap, in Swedish, and by extension, Norwegian, syllables can be stressed or unstressed, but only the stressed syllables carry an accent. This accent comes in one of two flavours, described by linguists as the acute and grave accents, or more boringly, just accents 1 and 2 respectively. There is considerable variation in how these accents occur, with Swedish spoken in Finland generally lacking distinction between these accents. Stockholm, for example, generally has accent 1 being a low tone, and accent 2 being a high tone. However, some southern Swedish variants have these tones reversed, with accent 1 being a high tone and accent 2 being a low tone instead.

Interestingly, this pattern of accents also corresponds to the patterns observed in Danish phonology, called the stød. Briefly put, the stød is a laryngeal or creaky phonation that occurs at the end of a syllable affected by this feature. This feature distinguishes between Danish words like hun (she) and hund (dog), which are pronounced as /’hun/ and /’hunˀ/ respectively. Furthermore, just like how the accents vary between varieties of Swedish and Norwegian, this stød also varies between varieties of Danish. Generally speaking, the accent 1 in Norwegian and Swedish corresponds to words with stød, while accent 2 corresponds to words without stød.

This led me to ask the following questions, how did we get here? What features and developments contributed to the development for such a melodic-sounding system?

Linguists have proposed a period in which such a phonemic distinction may have arisen, at around 1000 – 1200 AD, right about when Old Norse was spoken in Scandinavia. The traditional explanation of the Swedish and Norwegian pitch-accents involves two important processes, one that involves the formation of the clitic definite article ‘the’, and epenthesis in words that end in a syllabic consonant, like ‘n’, ‘r’, or ‘l’. These processes produce disyllabic words, or words with two syllables.

The logic is, in Old Norse, definite articles follow the word that is modified, which later became fused at the end of the word stem. For words like and (mallard duck), this creates a disyllabic word, from and hinn to anden. Many words that were initially monosyllabic would have become disyllabic without changing their pitch accents. This would lead to accent 1 being dominant in words like this, in contrast to ande-n (the spirit), which follow accent 2 instead.

Similarly, the formation of epenthesis before syllabic consonants would render some initially monosyllabic words as more disyllabic. This would also correspond to accent 1 today, although it does not really apply to all words that have evolved this way.

However, we can go even further back in time to understand where all of these come from. It must be noted that such a accent distinction is not really seen in other North Germanic languages like Icelandic and Faroese, which suggests that the pitch-accents heard in Swedish and Norwegian would have preceded the settlement of Iceland, or right around the 9th century AD.

This would have occurred after what linguists referred to as the ‘syncope period’, estimated to be around 550 – 850 AD. Briefly put, syncope is a process where some sounds are lost in a word. In Old Norse (and further back, Proto-Norse or PN), this would most likely manifest as losses of unstressed vowels.

Additionally, in PN, stress usually fell on the first syllable, which could be marked by loudness or duration of that syllable. Whether or not this stress came together with pitch contours in PN depends on which linguistics paper one consults. For example, Hamp suggested that PN had some form of pitch-accent system, which could imply that Icelandic and Faroese would have lost this system to a great extent. However, there are also linguists like Bye who argued that such a distinction did not arise until Old Norse was in use.

Riad’s hypothesis, published in 1998, provides a comprehensive and propelling argument concerning how the pitch contour system entered mainstream in Swedish and Norwegian. This hypothesis involves several processes, which can be sequentially listed as such, syncope, stress clash, and resolution.

The loss of unstressed syllables during the syncope period would have lead to stressed syllables occurring on adjacent syllables, resulting in this problem Riad referred to as ‘stress clash’. This necessitates resolution, which can occur via several methods, including stress shift (just move one of the stresses to another syllable). In these Germanic dialects at the time, the norm would have been to just remove the secondary stress, which would occur in the form of removing the loudness and duration carried in the syllable. While the secondary stress has been removed, the pitch contour which was once tied with this stress was preserved, and thereby separated from the information provided by stress. This would later give rise to accent 2.

Riad’s hypothesis, however, does not really quite answer where this pitch contour comes from, but more rather, demonstrated the process in which this pitch contour took over as a potential distinguishing feature between anden and anden.

This is where we arrive at Bye’s 2004 publication, which built on previous hypotheses on tonogenesis in Swedish and Norwegian, using Lorentz’s proposal of peak delay to build arguments around. This results in what Bye called target delay, which concerns how pitch accents are aligned with a given syllable. What needed to happen first was the rise of some form of allophonic variation in these pitch accents.

In words with more than one syllable, the high tone peak may be skewed to the right, while remaining unmoved in words with one syllable. This creates some form of delay, which later became lexicalised. If the peak occurs early in the stressed syllable, it would become accent 1 we hear today. Conversely, if the peak occurs late in the stressed syllable, it would become accent 2.

These accounts have provided a comprehensive summary surrounding the different hypotheses through which pitch-accents have arisen in Swedish and Norwegian in particular. Despite being of particular academic interest for several decades by now, there seems to be no consensus on when exactly this pitch-accent distinction could have first arisen. What I hope to have shown, though, is how such a distinction became relatively ‘mainstream’, in contrast to many other Germanic languages which still depend on stress for distinction between lexical terms.

Further reading

Bye, P. (2004) ‘Evolutionary typology and Scandinavian pitch accent’, Kluwer Academic Publishers.

Hamp, Eric P. (1959) ‘Final Syllables in Germanic and the Scandinavian Accent System’, Studia Linguistica, 13(1–2), pp. 29–48.

Kock, K. (1901) ‘Die alt- und neuschwedische Accentuierung’, Karl J. Trübner, Strassburg.

Liberman, A. (1982) ‘Germanic accentology’, The Scandinavian Languages, University of Minnesota Press, Minneapolis.

Lorentz, O. (2001) ‘Tonal structure and tonal function’, Ms., University of Tromsø.

Lorentz, O. (2002) ‘Delayed peak and tonal crowding in Scandinavian tonogenesis’, Ms., University of Tromsø.

Riad, T. (1996) ‘Remarks on the Scandinavian Tone Accent typology’, Nordlyd, 24, pp. 129-156.

Riad, T. (1997) ‘Stød, curl and generalized accent 2’, Ms., Stockholm University.

Riad, T. (1998) ‘The origin of Scandinavian tone accents’, Diachronica, 15(1), pp. 63-98.