Thursday, April 3, 2008

Hungarian Vowel Harmony

Hungarian is a language with vowel harmony. According to Gussenhoven and Jacobs (1998), it is a subclass of long-distance assimilation and excludes certain combinations of vowels in the word. In my MA thesis, I related the findings in the literature regarding Hungarian vowel harmony to the results of my sociolinguistic experiment with bilingual Hungarian speakers.

Part of my experiment involved the analysis of the neutral vowels. In addition to back vowels and front vowels, Hungarian has phonetically front vowels which for the purposes of vowel harmony are classified as neutral. This simply means that the front vowels "e" (low front vowel and mid front vowel written with an acute accent) and high front vowel "i" (long vowel written with an acute accent and short vowel) can combine with both front and back vowels. In a root with non-neutral vowels, they do not seem to trigger vowel harmony. Thus they are transparent (they let vowel harmony pass through them).

I also analyzed doublets, proper names and recent loanwords which allow either front or back suffixes. Doublets with back suffix variants have transparent neutral vowels because they allow vowel harmony to pass through them. On the other hand, doublets with front suffix vowels have opaque neutral vowels because they block the back vowel from spreading a back vowel to the suffix. In other words, they trigger a front vowel in the suffix. An example is the doublet "hotel". The phrase "in the hotel" can be "a hotelban" or "a hotelben". In the phrase "a hotelban", the neutral vowel "e" is transparent. The back vowel "o" triggers the back suffix variant -ban. However, in the phrase "a hotelben", the neutral vowel "e" acts as a harmonic front vowel. We can say it is opaque because it blocks the back vowel "o" from triggering a back vowel in the suffix. The suffix variant -ben has a front vowel.

Another part of my experiment involved toponyms with neutral vowels. These included "Los Angeles", "Seattle" and "Vancouver". Some linguists claim that the vowel "e" is no longer a neutral vowel but actually behaves as a front vowel. However, my results did not confirm this. They merely confirm that the neutral vowels exhibit different degrees of neutrality. The neutral vowel "i" is the most neutral and the low front unrounded "e" is the least neutral.

I tested the use of the inessive suffix -ban/-ben in 30 participants who attached a suffix to a toponym three times. This resulted in 90 tokens with each toponym. I obtained the following results with "Los Angeles", "Seattle"and "Vancouver":

Los Angeles -ben 87.8% -ban 12.2%
Seattle -ben 31.1% -ban 68.9%
Vancouver -ben 70% -ban 30%

In Hungarian magazines and newspapers, I had previously seen "Vancouverben" (in Vancouver), so I was surprised that in 30% of cases, participants used the form "Vancouverban". In the response "Vancouverban", the neutral vowel "e" is tranparent because a back vowel occurs in the suffix. However, it may be that in the response "Vancouverban", some participants used a schwa or a vowel similar to the mid front rounded vowel of French, i.e., "fleur" (flower). If this was the case, it is not surprising that they attached a back suffix variant to "Vancouver".

With "Seattle", a root which consists of a front vowel followed by a back vowel and front vowel, the back suffix variant was used 68.9% of the time. This is a very high percentage. If the vowel "e" is now a harmonic front vowel, the back suffix vowel should not have been used so frequently. However, it may again be the case that the final vowel of Seattle was pronounced by some participants as a schwa or a mid front rounded vowel. In such cases, the back suffix variant would be expected.

"Los Angeles" has two back vowels followed by two front vowels. Despite the presence of the front vowel "e" in the final two syllables, the back suffix variant -ban was used in 12.2% of cases. It may be that some participants pronounced one or both of these vowels as schwas, but nevertheless, I did not expect that the back suffix variant would be used to this extent. For the participants who attached a back suffix variant, the "e" vowel was clearly neutral.

How can we explain the reasons the neutral vowel "e" is either neutral or opaque? I offer a few possible explanantions.

In those cases in which the form "Vancouverben" (in Vancouver) is produced, the vowel "e" is a harmonic front vowel. Also, we can claim that recent cues predominate over earlier cues. Though the difference in the time of the utterance of the vowel "e" versus the earlier back vowels is in milliseconds, this may be significant for the purposes of vowel harmony. Another possibility is the adjacency condition which claims that the vowel of the suffix needs to share the same feature as that of the vowel in the adjacent syllable. In the form "Vancouverben", the vowel "e" of the final syllable and the vowel "e" of the suffix are both front vowels.

In those cases in which the form "Vancouverban" (in Vancouver) is produced, the vowel "e" is a neutral vowel. It is transparent because it does not participate in the vowel harmony process. Also significant may be the count effect which claims that the number of vowels in a word is significant. The word "Vancouver" has two back vowels and one front vowel. (The first vowel in "Vancouver" is a back vowel in Hungarian). It may be that a back suffix variant is more common in roots that have a greater number of back vowels. If this theory is correct, the back suffix vowel would be less common in a root that had the combination of one back vowel followed by one neutral vowel and more common in a root that had three back vowels followed by one neutral vowel. My experiment did not confirm whether or not this theory is correct. Another explanation may be stress. In Hungarian stress is fixed on the first syllable of a word. Since the stressed syllable is Vancouver contains a back vowel, this theory predicts that the suffix should contain a back vowel. However, my experiment did not confirm the accuracy of this theory and to the best of my knowledge, most linguists are not convinced this is valid.

To further my research, I would like to determine the reason the neutral vowels exhibit different degrees of neutrality. I would also like to determine the significance of the count effect and the role of stress in Hungarian vowel harmony. Also fascinating would be the answer to the question of why languages have vowel harmony.

No comments: