I attended an interesting talk at the Linguistic Society of America (LSA) conference today titled ‘Grammaticalization in Somali and the development of morphological tone’, by Nina Hagen Kaldhol (University of California, San Diego) and Sverre Stausland Johnsen (University of Oslo). Here’s a quick summary!

The Somali language is a member of the Cushitic language family spoken in the Horn of Africa. This talk is about Northern Somali in particular, which has become something of a standard. It’s also a tonal language, meaning that differences in pitch change the meaning of the word. Somali has two tones: high (H) and low (L). When writing Somali in a Latin writing system, H tones are indicated with an acute accent over the vowel: <á>.

One of the ways Somali uses tone is to mark grammatical gender. In the following examples, a H tone on the ultimate (last) vowel of the word indicates that the noun is grammatically feminine, while a tone on the penultimate (second to last) vowel indicates that the noun is grammatically masculine. (Technically these rules apply to moras rather than vowels. Short vowels have 1 mora, long vowels have 2.)

Historically, there were two rules about tones in Somali:

1. The H tone always occurs on the ultimate or penultimate mora. This means that tone in Somali nouns is demarcative—it demarcates the ends of words.
2. There may only be one H tone per word. When the number of tones per unit is limited in some way, we call this a cumulative tone system.

Another way Somali uses tone is to indicate grammatical differences such progressive aspect or negation. (Aspect is a technical term for the temporal structure of an event. Events can be repetitive, continuative, habitual, punctual, progressive, etc.) In the following examples, you can see how different tonal patterns are associated with certain grammatical differences. The progressive examples in (c) and (d), for instance, have a H tone on the stem cún- ‘eat’, while the other examples do not. (See the List of Abbreviations for explanations of the abbreviations.)

(1) cun-aa
    eat-PRES.POS
    'eat'

(2) cun-ó
    eat-PRES.NEG
    'not eat'

(3) cún-ay-aa
    eat-PROG-PRES.POS
    'eating'

(4) cún-ay-ó
    eat-PROG-PRES.NEG
    'not eating'

Did you notice a problem with these examples, however? They don’t obey the two tone rules listed above! Example (3) has a H tone at the left edge of word (violating Rule 1), and Example (4) has two H tones (violating Rule 2).

How did this happen? Why did the tone rules change? Kaldhol & Stausland Johsen propose that the system changed when certain two-word constructions became merged into one word over time, a process known as univerbation. (Note: Univerbation means two words becoming one, not necessarily two verbs becoming one. The Latin word verbum meant ‘word’, not ‘verb’.)

Here’s how Kaldhol & Stausland Johsen explain it: Somali originally had a verb hay meaning ‘have’. You could use it in phrases like cúni hayaa (literally ‘I have eat’) to mean ‘I am eating’. Over time, however, the two words became fused into cúnayaa, and speakers reanalyzed the -ay part as a suffix indicating progressive aspect. This process where full words get reanalyzed as grammatical affixes is called grammaticalization. The result is that the H tone that was originally on the first word now got stuck in the middle of the word rather than the end, so speakers just started to associate that H tone with progressive aspect. Now, the progressive aspect suffix -ay adds a H tone to the mora before it. When an affix adds a tone to something beside it, this is called a floating tone.

This same process of grammaticalization also resulted in cases where two H tones occur on the same word. The negative suffix in Somali is -ó with a H tone. When the word hayó ‘I do not have’ fused with cúni ‘eat’, the result was cúnayó ‘I am not eating’, with two H tones.

Kaldhol & Stausland Johsen give a number of other examples and pieces of evidence supporting their proposal. Overall, this was an interesting, well-presented talk with a clear and simple point, nicely illustrating one way that morphological tone arises over time. Thank you Nina Hagen and Sverre!

If you’d like to see the slides for this talk, Nina Hagen has generously made them available on her website here!

Note: This post is my best attempt at representing this research faithfully and in a concise way, but I cannot always guarantee its accuracy, since this is neither my research nor my area of expertise. My goal here is simply to spotlight research I find interesting! Please contact the original researchers with any questions about this research.

List of Abbreviations