World Speech Accents Online

In a previous post I discussed Forvo, an online database for teaching and learning pronunciation.  In a similar collaborative resource approach, a group at George Mason University maintains an archive of speech accents for English, spoken by both native and non-native speakers of English from around the world.

Each speaker records the same paragraph of English text, allowing comparative studies such as contrastive analysis of different language groups.  This “elicitation paragraph” has been constructed to contain most of the sounds used in common English, together with some words that use “difficult sounds and sound sequences”.

The site is collaborative in that it solicits recordings from its visitors (described as “remote researchers”) and provides a detailed set of instructions for making high quality recordings for inclusion in the archive. Samples must be 44.1 KHz., 16-bit mono recordings and each sample is vetted by the GMU archive team, cosisting of qualified linguists.  Samples also need a signoff by subjects, ensuring that all required ethical conditions have been met.  These include a minimal age of 18 and an understanding of how the recording will be used, etc.

One nice touch about this site is the Atlas page, allowing visitors to browse accents by geographic region. I couldn’t resist checking out a regional accent for Sydney, Australia.  To my ear, the provided sample sounds like a caricature of Australian English, more like a exagerrated version of Paul Hogan’s speech than a typical Sydneysider.

Forvo: learning and teaching pronunciation – online

Here’s a great example of a collaborative online resource that relies on its users for most of its content – not unlike the model used by Wikipedia:

Forvo is an online pronunciation resource that is built by its many users. It currently has a database of 210 languages including all major western, african and asian languages and some more obscure ones such as; Occitan (Southern France), Hawaiian, Sudanese (East Africa) and Quechua (Peru). The total number of words currently held is 203,463 but only around 66% of these (133,030) have recorded pronunciations so far. The site lists words awaiting pronunciation recordings as “pending”, and encourages users to add their own recorded pronunciation of the word.  The recorded file is tagged with the user’s location so that anyone using the guide has an idea of how likely the recording is to be influenced by regional accents or dialects.  Someone fromn North Carolina, for example, would probably pronounce “house” rather differently from a resident of Manchester in the UK.  For this reason, Forvo encourages its users to add their own pronunciation of a word, even if a recording already exists.