Who commands within a word?
There are vowels that do not limit themselves to occupying their place: they condition others, often without speakers realizing it. This is what is known as vowel harmony, a phenomenon that helps us to better understand how certain hierarchies are distributed within words.
PalmaWhat would happen if, within a word, not all vowels were equally autonomous? If one decided how the others should sound? If we listen carefully to how some speakers in the Valencian Country pronounce words like ‘dona’ or ‘cosa’, or ‘conill’ or ‘sortir’ in certain areas of Mallorca or the Ebre Lands, something shifts with respect to what we would expect. On the one hand, where we would expect an 'a' at the end, we can hear a vowel that approaches the stressed vowel: ‘c[o]v[o]’ instead of ‘c[o]v[a]’. On the other hand, a vowel that we would expect to be more open closes due to the influence of another: ‘c[u]nill’ instead of ‘c[o]nill’.
These forms respond to a well-documented mechanism in Catalan and many other languages: vowel harmony. In general terms, it means that one vowel transfers some of its features to another. Saying it like this, however, is staying on the surface. What matters is not just that there is assimilation, but how it is organized: which vowel influences the others, in which direction, and how far it reaches.
Hierarchy
In the best-known cases, the hierarchy seems quite clear. In some Valencian dialects (especially southern, but not exclusively), the open vowels [è] and [ò] of the stressed syllable can modify the final vowel 'a', which is unstressed and often morphologically irrelevant. Thus, 'terra' can end with a vowel close to [è], and 'cosa' with one close to [ò]. A similar phenomenon is found in some areas of Mallorcan and in a large part of Western Catalan: a stressed 'i' vowel can cause the pre-tonic vowel to close, as in 'c[u]nill' or 's[u]rtir'.
In both cases, the movement goes from a strong position to a weak one. The stressed syllable, being more perceptible, projects its features towards an unstressed position. It is a pattern that fits with a widespread intuition: what is heard best tends to prevail.
However, this intuition is not always confirmed. There are systems in which the direction is reversed and it is the weak vowels that condition the strong ones.
In Central Catalan, for example, there are contexts in which the stressed vowel depends on the post-tonic vowel. This is the case with forms like 'euro' or 'ESO'. If the final vowel is realized as [u], the stressed vowel tends to be open: '[è]ur[u]', '[è]s[u]'. On the other hand, if it remains as [o], the stressed vowel closes: '[é]ur[o]', '[é]s[o]'. What is relevant here is the direction of harmony: a vowel located after the stressed one (and therefore less prominent) ends up determining its quality.
This type of behavior is also found outside of Catalan. In some Asturian dialects, for example, a high final vowel can cause the stressed vowel to close: 'caldíru' ('calder', singular) versus 'calderos' ('calders', plural). The peripheral vowel, despite being less prominent, gains weight and extends its features backward.
The direction of harmony, therefore, is not fixed. There are systems in which features propagate from left to right, others in which they do so in reverse, and varieties with mixed behavior. Tortosí is a good example: 'melic', pronounced 'milic', shows a backward influence, while cases like 'ginecòleg' (pronounced 'ginecòleg') indicate assimilations between unstressed vowels in the opposite direction.
When we move outside the Romance sphere, the phenomenon can be even more systematic. In languages like Finnish, vowel harmony is not a punctual effect, but a structural principle. Vowels are divided into two groups – front ('ä', 'ö', 'y') and back ('a', 'o', 'u') – with neutral vowels (e, i) that can appear with both series. This division determines the form of suffixes: if the root contains back vowels, the suffixes will also have them; if it contains front vowels, the suffixes will adapt to them. Thus, 'kaura' ('oats') becomes 'kauralla' ('with oats' or 'on oats'), while 'käyrä' ('curve') becomes 'käyrällä' ('on the curve'). The suffix is not invariant: it systematically adapts to the vowels of the base.
Regarding the causes of harmony, common explanations revolve around two axes. On the one hand, articulation: maintaining similar configurations of the speech organs (i.e., that all vowels in the word are pronounced similarly) can facilitate production. On the other hand, perception: extending a feature can make it more audible, especially if there is a risk of loss.
Eastern Andalusian illustrates this well. The final '-s' often weakens or disappears, and this loss is compensated by the opening of the preceding vowels (for example, 'los', pronounced ['lòh']). This feature can extend backwards, up to the stressed vowel, as in 'hombres', pronounced 'h[ò]mbr[è]'. It does not necessarily simplify articulation, but it reinforces a contrast that would otherwise be lost.
Homogeneous sequences
In other cases, as in some Catalan dialects, both factors can coincide: harmony creates more homogeneous sequences and, at the same time, reinforces relevant features. There is not a single motivation, but a balance between what is easier to say and what is easier to understand.
Overall, vowel harmony forces us to look at the word as a system of internal relationships. Vowels not only affect each other and show that, even in words, there are hierarchies. Sometimes, a single vowel ends up setting the rules for the whole.