Indo-European ablaut

In linguistics, the Indo-European ablaut is a system of apophony in the Proto-Indo-European language.
An example of ablaut in English is the strong verb sing, sang, sung and its related noun song, a paradigm inherited directly from the Proto-Indo-European stage of the language. Traces of ablaut are found in all modern Indo-European languages, though its prevalence varies greatly.

History of the concept

The linguistic phenomenon of ablaut was first recorded by Sanskrit grammarians in the later Vedic period, and was codified by Pāṇini in his Aṣṭādhyāyī, where the terms and vṛddhi| were used to describe the phenomena now known respectively as the full grade and lengthened grade.
In the context of European languages, the phenomenon was first described in the early 18th century by the Dutch linguist Lambert ten Kate, in his book Gemeenschap tussen de Gottische spraeke en de Nederduytsche.
The term ablaut is borrowed from German, and derives from the noun "sound", and the prefix, which indicates movement downwards or away, or deviation from a norm; thus the literal meaning is "sound derivation". It was coined in this sense in 1819 by the German linguist Jacob Grimm in his Deutsche Grammatik, though the word had been used before him. In particular, the 17th-century grammarian Schottelius had used the word negatively to suggest that German verbs lacked the sophistication of the classics, but there is no hint of this disdain in Grimm or in modern scholarly usage.
In English, the term became established through the 1845 translation of Bopp's Comparative Grammar.

Ablaut and vowel gradation

Vowel gradation is any vowel difference between two related words or two forms of the same word. The difference does not need to be indicated in the spelling. There are many kinds of vowel gradation in English and other languages, which are discussed generally in the article apophony. Some involve a variation in vowel length, others in vowel coloring and others the complete disappearance of a vowel.
For the study of European languages, one of the most important instances of vowel gradation is the Indo-European ablaut, remnants of which can be seen in the English verbs ride, rode, ridden, or fly, flew, flown. For simply learning English grammar, it is enough to note that these verbs are irregular, but understanding why they have unusual forms that seem irregular requires an understanding of the grammar of the reconstructed proto-language.
Ablaut is the oldest and most extensive single source of vowel gradation in the Indo-European languages and must be distinguished clearly from other forms of gradation, which developed later, such as Germanic umlaut or the results of modern English word-stress patterns. Confusingly, in some contexts, the terms 'ablaut', 'vowel gradation', 'apophony' and 'vowel alternation' are used synonymously, especially in synchronic comparisons, but historical linguists prefer to keep 'ablaut' for the specific Indo-European phenomenon, which is the meaning intended by the linguists who first coined the word.

Ablaut grades

In Proto-Indo-European the basic, inherent vowel of most syllables was a short e. Ablaut is the name of the process whereby this short e changed, becoming short o, long ē, long ō, or sometimes disappearing entirely to leave no vowel at all.
Thus, ablaut results in the alternation of the following sounds:
If a syllable had a short e, it is said to be in the "e-grade" or "full grade". When it had no vowel, it is said to be in the "zero grade". Syllables with long vowels are said to be in "lengthened grade".
A classic example of the five grades of ablaut in a single root is provided by the different case forms of two closely related Greek words. In the following table, an acute accent marks the syllable carrying the word stress, a macron marks long vowels, and the syllable in bold is the one illustrating the different vowel gradations.

Ablaut grade	PIE	Greek	Greek	English translation
e-grade or full grade	ph₂-tér-m̥			"father"
lengthened e-grade	ph₂-tḗr			"father"
zero-grade	ph₂-tr-és			"father's"
o-grade	n̥-péh₂-tor-m̥			"fatherless"
lengthened o-grade	n̥-péh₂-tōr			"fatherless"

In this unusually neat example, the following can be seen:

A switch to the zero-grade when the word stress moves to the following syllable.
A switch to the o-grade when the word stress moves to the preceding syllable.
A lengthening of the vowel when the syllable is in word-final position before a sonorant.

As with most reconstructions, however, scholars differ about the details of this example.
One way to think of this system is to suppose that Proto-Indo-European originally had only one vowel, short e, and over time it changed according to phonetic context. Thus, the language started to develop a more complex vowel system. It has often been speculated that an original e-grade underwent two changes in some phonetic environments: under certain circumstances, it changed to o and in others, it disappeared entirely.
However, that is not certain: the phonetic conditions that controlled ablaut have never been determined, and the position of the word stress may not have been a key factor at all. There are many counterexamples to the proposed rules: deywós and its nominative plural deywóes show pretonic and posttonic e-grade, respectively, and wĺ̥kʷos has an accented zero grade.

Lengthened grades

Many examples of lengthened grades, including those listed above, are not directly conditioned by ablaut. Instead, they are a result of sound changes like Szemerényi's law and Stang's law, which caused compensatory lengthening of originally short vowels. In the examples above, Szemerényi's law affected the older sequences ph₂-tér-s and n̥-péh₂-tor-s, changing them to ph₂-tḗr and n̥-péh₂-tōr. Thus, these forms were originally in the regular, unlengthened e-grade and o-grade. Such lengthened vowels were, however, later grammaticalised and spread to other words in which the change did not occur.
Nevertheless, there are examples of true lengthened grades, in which short e alternates with long ē. Examples are the verbs with "Narten" inflection, and nouns like mḗh₁-n̥s "moon", genitive méh₁-n̥s-os. Alternations of this type were rare, however, and the e ~ o ~ ∅ alternation was the most common by far. The long ō grade was rarer still and may not have actually been a part of the ablaut system at all.

Zero grade

The zero grade of ablaut may appear difficult for speakers of English. In the case of ph₂trés, which may already have been pronounced something like, it is not difficult to imagine it as a contraction of an older ph₂terés, pronounced perhaps, as this combination of consonants and vowels would be possible in English as well. In other cases, however, the absence of a vowel strikes the speaker of a modern western European language as unpronounceable.
To understand, one must be aware that there were a number of sounds that were consonants in principle but could operate in ways analogous to vowels: the four syllabic sonorants, the three laryngeals and the two semi-vowels:

The syllabic sonorants are m, n, r and l, which could be consonants much as they are in English, but they could also be held on as continuants and carry a full syllable stress and then are transcribed with a small circle beneath them. There are many modern languages who show these sounds in syllable nuclei, including Indo-European ones. In English, these only occur very marginally in stressed position, but are common allophones of schwa when unstressed ; there are no direct counterparts, however.
The laryngeals could be pronounced as consonants, in which case they were probably variations on the h sound and so normally transcribed as h₁, h₂ and h₃. However, they could also carry a syllable stress, in which case they were more like vowels. Thus, some linguists prefer to transcribe them ə₁, ə₂ and ə₃. The vocalic pronunciation may have originally involved the consonantal sounds with a very slight schwa before and/or after the consonant.
In pre-vocalic positions, the phonemes u and i were semi-vowels, probably pronounced like English w and y, but they could also become pure vowels when the following ablaut vowel reduced to zero.

When u and i came in postvocalic positions, the result was a diphthong. Ablaut is nevertheless regular and looks like this:
Thus, any of these could replace the ablaut vowel when it was reduced to the zero-grade: the pattern CVrC could become CrC.
However, not every PIE syllable was capable of forming a zero grade; some consonant structures inhibited it in particular cases, or completely. Thus, for example, although the preterite plural of a Germanic strong verb is derived from the zero grade, classes 4 and 5 have instead vowels representing the lengthened e-grade, as the stems of these verbs could not have sustained a zero grade in this position.
Zero grade is said to be from pre-Proto-Indo-European syncope in unaccented syllables, but in some cases the lack of accent does not cause zero grade: deywó-, nominative plural -es "god". There does not seem to be a rule governing the unaccented syllables that take zero grade and the ones that take stronger grades.