Gāndhārī is the modern name, coined by scholar Harold Walter Bailey (in 1946), for a Prakrit language found mainly in texts dated between the 3rd century BCE and 4th century CE in the northwestern region of Gandhāra. The language was heavily used by the former Buddhist cultures of Central Asia and has been found as far away as eastern China, in inscriptions at Luoyang and Anyang. Gāndhārī Prakrit appears to be descended from, or heavily influenced by, Vedic Sanskrit or a closely related language, although there is an ongoing debate about the question of whether some Prakrits were originally non-prestige contemporaries and/or antecedents of Sanskrit.
It appears on coins, inscriptions and texts, notably the Gandhāran Buddhist texts. It is notable among the Prakrits for having some archaic phonology (some being characteristic of the Dardic languages of the modern region), for its relative isolation and independence, for being partially within the influence of the ancient Near East and Mediterranean and for its use of the Kharoṣṭhī script, a unique sister to the Brahmic scripts used by other Prakrits.
Gāndhārī is a late Middle Indo-Aryan language - a Prakrit - with unique features that distinguish it from all other known Prakrits. Phonetically, it maintained all three Old Indo-Aryan sibilants - s, ś and ṣ - as distinct sounds where they fell together as [s] in other Prakrits, a change that is considered one of the earliest Middle Indo-Aryan shifts. Gāndhārī also preserves certain Old Indo-Aryan consonant clusters, mostly those involving v and r. In addition, intervocalic Old Indo-Aryan th and dh are written early on with a special letter (noted by scholars as an underlined s, [s]), which later is used interchangeably with s, suggesting an early change to a sound, likely the voiced dental fricative ð, and a later shift to z and then a plain s.
In general terms, Gāndhārī is a Middle Prakrit, a term for middle-stage Middle Indo-Aryan languages. It only begins to show the characteristics of the Late Prakrits in the 1st century of the Common Era. The Middle Prakrit phonetic features are the weakening of intervocalic consonants: degemination and voicing, such as the shift of OIA *k to g. The most rapid loss was the dentals, which started to disappear completely even before the late period as with *t > ∅ as in *pitar > piu; in contrast, retroflex consonants were never lost. There is also evidence of the loss of a distinction between aspirates and plain stops as well, which is unusual in the Indo-Aryan languages.
In Central Asian Gāndhārī, there is often confusion in writing nasals with homorganic stops; it is unclear if this might represent assimilation of the stop or the appearance of prenasalized consonants to the phonetic inventory.
Gāndhārī grammar is difficult to analyse; endings were eroded not only by the loss of final consonants and cluster simplification of all Prakrits but also by the apparent weakening of final vowels "'to the point that they were no longer differentiated'". Nonetheless, there was still at least a rudimentary system of grammatical case. Verbal forms are highly restricted in usage due to the primary usage of longer texts to translations of religious documents and the narrative nature of the sutras but seem to parallel changes in other Prakrits.
The lexicon of Gāndhārī is also limited by its textual usage; it is still possible to determine unusual forms, such as Gāndhārī forms that show commonalities with forms in modern Indo-Iranian languages of the area, notably the Dardic languages. An example is the word for sister, which is a descendant of Old Indo-Aryan svasṛ- as in the Dardic languages, whereas all the Indo-Aryan languages have replaced that term with reflexes of bhaginī.
Initial identification of a distinct language occurred through study of one of the Buddhist āgamas, the Dīrghāgama, which had been translated into Chinese by Buddhayaśas (Chinese: 佛陀耶舍) and Zhu Fonian (Chinese: 竺佛念).
The now dominant hypothesis on the propagation of Buddhism in Central Asia goes back to 1932 when E. Waldschmidt remarked that the names quoted in the Chinese Dīrghāgama (T. 1), which had been translated by the avowedly Dharmaguptaka monk Buddhayaśas (who also translated the Dharmaguptakavinaya), were not rendered from Sanskrit, but from a then undetermined Prākrit also found in the Khotan Dharmapada. In 1946, Bailey identified this Prākrit, which he named Gāndhārī, as corresponding to the language of most Kharoṣṭhī inscriptions from Northwestern India.
Since this time, a consensus has grown in scholarship which sees the first wave of Buddhist missionary work as associated with Gāndhārī and the Kharoṣṭhī script, and tentatively with the Dharmaguptaka sect. Available evidence also indicates that the first Buddhist missions to Khotan were carried out by the Dharmaguptaka sect, and used a Kharoṣṭhī-written Gāndhārī. However, there is evidence that other sects and traditions of Buddhism also used Gāndhārī, and evidence that the Dharmaguptaka sect also used Sanskrit at times.
It is true that most manuscripts in Gāndhārī belong to the Dharmaguptakas, but virtually all schools — inclusive Mahāyāna — used some Gāndhārī. Von Hinüber (1982b and 1983) has pointed out incompletely Sanskritised Gāndhārī words in works heretofore ascribed to the Sarvāstivādins and drew the conclusion that either the sectarian attribution had to be revised, or the tacit dogma "Gāndhārī equals Dharmaguptaka" is wrong. Conversely, Dharmaguptakas also resorted to Sanskrit.
Starting in the first century of the common era, there was a large trend toward a type of Gāndhārī which was heavily Sanskritized.
Until 1994, the only Gāndhāri manuscript available to the scholars was a birch bark manuscript of a Buddhist text, the Dharmapāda, discovered at Kohmāri Mazār near Hotan in Xinjiang in 1893 CE. From 1994 on, a large number of fragmentary manuscripts of Buddhist texts, seventy-seven altogether, were discovered in eastern Afghanistan and Western Pakistan. These include:
Mahayana Buddhist Pure Land sūtras were brought from Gandhāra to China as early as 147 CE, when the Kushan monk Lokakṣema began translating the first Buddhist sutras into Chinese. The earliest of these translations show evidence of having been translated from Gāndhārī. It is also known that manuscripts in the Kharoṣṭhī script existed in China during this period.