On The Language Closet, we have taken a look at some of the more unusual vowel systems that can occur in the world’s languages. From languages that have been purported to have fewer than two phonemic vowels to languages with multiple phonation types for each vowel, there is one main similarity that these languages primarily share. That these languages distinguish vowels by vowel height.

We have talked about vertical vowel systems before, where vowels are only phonemically distinguished by the extent to which the body of the tongue is raised when articulating the said vowel. There are several examples of languages with vertical vowel systems across the world, from Upper Arrernte in Australia to debatably, Kabardian in the Caucasus, these languages only distinguish vowels along a singular axis.

However, there is another axis that does not seem to have received much attention. The horizontal axis.

This axis marks the frontness and backness of vowels, as shown by the position of the tongue in the mouth. Try pronouncing the vowel sounds [i] and [e]. Do you feel where your tongue is? These are indeed examples of front vowels. Conversely, [u] and [o] are back vowels. In this logic, a language that uses a horizontal vowel system would phonemically distinguish between vowels like [i] and [u], but not necessarily [i] and [e], as [i] and [e] differ in their vowel heights. Phonotactic processes would be involved in producing what could be interpreted as surface vowels instead.

Pouring through phonological inventories of various languages across the world, I am unable to find an attested example of a natural language that distinguishes vowels using only this axis. Languages with two vowels tend to distinguish them along the vertical axis (i.e. vowel height), while languages with three vowels tend to distinguish them along the same vertical axis, or some variation of /a i u/ or /a i o/. Searching up for publications suggesting explanations behind this pattern has also returned fruitless, so I still have this question still open.

However, while a language using a horizontal vowel system has not been identified today, it does not necessarily imply that such languages simply do not exist. Perhaps we just need to peel back the curtain and take a look at a hypothetical language which could have been spoken several millennia back.

The hypothesised Proto Indo-European language, or PIE, is perhaps the only example I have come across that seemed to suggest a plausible example of a horizontal vowel system. In essence, PIE is a hypothetical reconstructed language that suggests what the common ancestor of all Indo-European languages might have sounded like, with a heavy emphasis on the word ‘might’. Established using evidence from its daughter languages like Sanskrit, Latin, and Hittite, PIE is perhaps the most thoroughly studied of all proto-language reconstructions, with the first records of a proposed proto-language unifying languages like Latin and Greek dating back to the 17th century, and the first recorded attempts of characterising PIE dating back to 2 centuries ago.

Reconstruction of PIE’s features included taking a lot of consideration from the languages which had amongst the earliest attestations, or the earliest attestation of a certain branch. For example, the Anatolian and Tocharian languages are hypothesised to be the first attested branch of Indo-European languages which diverged from PIE, with Ancient Greek, Sanskrit, and Latin following thereafter. As such, for some PIE features, these languages may offer more weight, as they are more likely to reflect some features, or some modification patterns, that would trace back to what PIE might have had.

PIE has been argued to have at least four phonemic vowels, and more could have existed. However, what is generally common amongst these reconstructions for PIE is, the *u and *i sounds are generally treated as semivowels *w and *j respectively. PIE is also argued to distinguish vowels by length, giving us the vowel inventory *e, *o, *e:, and *o:. This is pretty much what I mean by a horizontal vowel system, where the *e and *e: are front vowels, and the *o and *o: are back vowels. There does not seem to be a distinction by vowel height, though this part is rather contentious.

Whether or not the /a/ vowel could have existed in PIE is under debate. Arguments against its existence point towards one of the leading theories underpinning PIE phonology, that is, the laryngeal theory. If you look up words that ultimately lead back to PIE reconstructions, you would find some unusual notations in the words, which go by *h1, *h2, and *h3. These are the laryngeal consonants that are thought to have had existed in PIE, though the true nature of how they are articulated is not known, just that we know that there could have been multiple types of these consonants. This theory gained traction as the Hittite language got deciphered, and was found to derive many of its words ultimately from PIE. In Hittite itself, there were some sounds that could be thought of as possible laryngeals leading back to PIE, further lending support for the laryngeal theory.

So, how does the laryngeal theory shape the debate over the phonemicity of the *a sound?

The laryngeal consonant in question is *h2. This is occasionally referred to as the ‘a-colouring’ laryngeal. Some hypotheses over that *h2 could have been included the pharyngeal consonant [ħ], as in the Semitic languages, this consonant would also cause some a-colouring as well. It is thought that the /a/ sound could itself be a surface vowel of the *e sound, which has been a-coloured by the *h2 sound. As such, the *a sound could be reduced to *e. Counterarguments however, suggested that the *a sound could not in fact, be traced back to any laryngeal consonant, and would have been independent of *h2. Examples included the reconstructed word for ‘white’ include *albʰós, which depicts the *a sound as its own phoneme which existence is independent of the *h2 sound. As such, this relative lack of predictability of the *a sound would suggest that the *a sound is indeed a phoneme, rather than an allophone or a surface vowel of sorts.

These sounds would have gone through some certain sound changes over time, and as languages and branches diverge from PIE, these sounds could have merged, split, or gone through other processes. Nevertheless, Anatolian languages like Hittite have been treated as significant due to the substantial number of PIE features they are thought to conserve. Being hypothesised as one of the first divergences from PIE, the decipherment of Hittite provided further understanding in the reconstruction of PIE.

Whether or not these sounds could have existed, if PIE did exist in the form we think, is unlikely to be fully elucidated. After all, the common ancestor of the Indo-European languages preceded all writing systems, leaving behind the abundance of descendant languages from Hittite to Albanian. As such, if the reconstruction of PIE is indeed accurate, this might have been the one example of a language we know of which uses a horizontal vowel system.

Further reading

Beekes, R. S. P. (1995) Comparative Indo-European Linguistics, Amsterdam, Benjamins.

Byrd, A. M. (2015) The Indo-European Syllable, Brill.

Lindeman, F. O. (1987) Introduction to the ‘Laryngeal Theory’, Oslo, Norwegian University Press.

Mayrhofer, M. (1986) Indogermanische Grammatik. I, 2, Halbband, Lautlehre, Heidelberg, Carl Winter Universitätsverlag.

Pooth, R. A. (2015) ‘A typological overview of Proto-Indo-European’.

Vennemann, T. (1989) The New Sound of Indo-European: Essays in Phonological Reconstruction, Berlin, New York, De Gruyter Mouton.