According to the Sindh Information Department, 78 newspapers of Sindhi language are regularly published and numbers don’t include monthly magazines but very few of these newspapers are available online.

Around 50 private radio stations and eight private TV channels, including KTN, Kashish, Sindh TV, Awaz TV, Mehran TV and Dharti TV, are also broadcasting in Sindhi but the impact of the language on multimedia is quite limited.

Sindhi is the only regional language in Pakistan that is the medium of instruction in education in a province (Sindh). It has its own robust media while Sindhi literature is also being written regularly.

Mohan Madhosh has been publishing Sindhi books in Hyderabad since 1996. According to him, about 35 organisations are in publishing business, printing about 250 books in Sindhi language annually but there are almost no websites for these publishers.

It seems that despite its strength and maturity, the Sindhi language is reluctant to embrace new technology or is unable to keep up with the rapid development in this new field.

Amar Fayyaz Buriro, the director of the Abdul Majid Bhurgri Institute of Language Engineering (AMBILE), believes that languages that cannot adopt innovation and accept change will be extinct and only those languages will survive that would become a part of software, mobile applications, and online learning systems.

In Pakistan, there are only two institutions focused on language engineering, which aim to bring regional languages into the computing world. One of them, AMBILE, was launched by the Sindh government in 2017 and it was named after Abdul Majid Bhurgri whose efforts led to Sindhi's entry into the computer world in 1987-88.

More language users, more online presence

Dr Wazir Ali Jamro, head of the Data Science Department at the Institute of Business Management (IBM), Karachi, has been working on computer technology for the Sindhi language for the past 10 years. He explains that Sindhi is available on platforms like ChatGPT, Google, DeepSeek and Meta, and translation from Sindhi into English and vice versa is very much possible. A response in Sindhi can also be obtained but its quality is not on a par with English.

“There is not much material in Sindhi on the internet whereas machines require huge data. The more a language is used on the internet, the more accurately the machine will learn to use it.”

Dr Wazir Ali believes that language is interconnected with the market and business. The world of algorithms and browsers depends on ‘usage;’ hence, the larger the portion of the population that uses Sindhi, the more companies will design their products considering the Sindhi users. However, he adds, most educated users in Sindh use English on the internet instead of Sindhi, that’s why Sindhi has not yet left its print on the internet or flourished there.

Prof Muteeur Rahman, who holds a PhD in Computational Linguistics and heads the Computer Science Department at Isra University, Hyderabad, also believes that a language will survive only with its presence in IT applications, software and devices. 

“Now it depends on Sindhi speakers to what extent they use their mother tongue on the internet. The companies need users, and whichever language has more users will naturally become a part of the internet.”

He said the learning algorithms of Meta, ChatGPT, DeepSeek, etc., have learned from the data available in Sindhi on the internet. When we ask them something in Sindhi, they take relevant information from that data or they take it from English and other languages’ data and translate it into Sindhi.

“All the work done for the Sindhi language has been at an individual level, with no visible role of government institutions,” he adds.

Regional languages need machine learning models 

Professor Mutiah Rahman believes that regional and local languages need to develop machine learning models, considering their cultures and values, which can accommodate regional accents, words and meanings.

He stressed that for the development of machine learning models, experts in Sindhi linguistics and digital libraries are required, which can assist in processing Sindhi material. However, no significant work has been done in this regard so far.

“We need to create models and software where Sindhi is not only the language for machines interaction but also the language for storing, processing and presenting information.”

Dr Wazir Ali from the IBM also points out that no work is being done on conversion of Sindhi sounds into script or on speech-to-text and text-to-speech. The limited capability available on YouTube can only recognise Sindhi written in Roman script.

He also complains that all the work related to the Sindhi language and technology has been done at the individual level and no efforts have been made by the government institutions.

However, Sindhi Language Authority Chairperson Dr Ishaq Samejo disagrees with this objection. He believes that work at the institutional level is also being done. 

“The establishment of the Institute of Language Engineering in Hyderabad in 2017 is a clear example of this and the institute is playing an important role in this regard.”

The Sindh government established the Sindhi Language Authority in 1990 under the Teaching, Promotion and Use of Sindhi Language (Amendment) Act, 1990. Its chairperson says that the institution is dealing with issues related to linguistics, grammar, dictionaries, the Indus script, as well as the development of the language. It is also encouraging the ‘computationalisation’ of Sindhi through awards and scholarships

Amar Fayyaz from the Institute of Language Engineering claims, “Thanks to our efforts, last year Sindhi Optical Character Recognition (OCR) was developed, and now even scanned documents in Sindhi can be read when posted online in the image format”.
(OCR is the technology that converts text into images or scanned documents into machine-readable text. It can also be used to recognize text in subtitles.)

Regarding the Sindhi OCR claim, Dr Wazir Ali states that the OCR being hailed as a major success is not very efficient.

Professor Muteeur Rahman says that this OCR has neither been widely used nor has any innovation been introduced in it. The internet companies have developed their own OCRs, which are being used.