Question About the Speech Accent Archive

A recent piece at WAMU profiles the Speech Accent Archive, an online database of accent samples compiled by the George Washington University‘s Linguistics Department.  A vast collection of recordings of English speakers around the globe, this eleven-year-old resource is one of the first accent-centric websites I ever encountered.

A good example of the SAA’s unique format is this accent sample, from Kilkenny, Ireland. As you can see, the site provides a useful International Phonetic Alphabet transcription to go along with each subject.

I have tremendous respect for the people who’ve created this database. And I appreciate that the study focuses an uncommon amount of attention on non-native English speakers. As such, I don’t intend the following to be taken as outright criticism.  But I’ve always been a little confused by the passage that the SAA’s subjects read:

Please call Stella. Ask her to bring these things with her from the store: six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob. We also need a small plastic snake and a big toy frog for the kids. She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.

In the above-linked article, Steven Weinberger, director of George Mason’s linguistics program, states the following about the sample text:

“[The paragraph has] difficult clusters of consonants like P-Ls and S-Ts, just about every vowel in English, and just about every consonant …”

While it’s technically true that this passage features “just about every vowel in English,” the few omissions don’t strike me as exactly trivial. Notably, the passage lacks:

1.) “Mouth” words: house, loud, about, etc.
2.) “Thought” words: fought, caught, flawed, etc.*
3.) Words with stressed syllables that end in “r:” start, near, fair, etc.

I admit this sounds like nitpicking (or worse, backseat academia). But some of these phonemes are important, in my opinion, for getting the full picture of what an accent sounds like. For example, without “thought” words, it’s difficult to deduce if an American has the Caught-Cot merger.  The passage doesn’t leave out too many of English’s vowel sounds, but those it does are pretty noticeable.

Conversely, what I do like about the passage is that the researchers seem to have geared it more toward consonants (particularly consonant clusters), which are often neglected in the face of English’s diverse vowel pronunciations. At the same time, the lack of some of English’s most important vowels has always puzzled me.

Again, I don’t mean this as any kind of take-down. I’m just a little shaky on the researcher’s methodology, and am perhaps unclear on the aim of this text in particular. Anyone care to enlighten me?

*The passage DOES have words like “small” or “call,” however.  But due to “l-coloring,” someone with the Cot-Caught merger can pronounce these words in the same way as someone who doesn’t have the merger.


About Ben

Ben T. Smith launched his dialect fascination while working in theatre. He has worked as an actor, playwright, director, critic and dialect coach. Other passions include linguistics, urban development, philosophy and film.
This entry was posted in English Phonetics and tagged , . Bookmark the permalink.

13 Responses to Question About the Speech Accent Archive

  1. lynneguist says:

    Another thing that’s interesting here is that there’s at least one downright Americanism (‘snow pea’) and a kind-of Americanism (‘train station’–though it’s widely said in Britain now, it’s perceived by many as an Americanism). If you don’t know what a snow pea is, will it affect the delivery of the compound?

    • trawicks says:

      That’s a good question. I’ve never considered whether the use of regionalisms might affect elicitation texts. Compounds seem especially tricky if you’re contrasting AmE vs. BrE because Americans use adjectival stress more than Brits, and the placement of stress in compounds would be expected to impact pronunciation. Although I’m not sure if a Brit would say “snow PEA,” necessarily!

  2. Dw says:

    “store”, “go meet” and (maybe not so much nowadays) “kids” are also Americanisms. “She can scoop” for “please tell her to scoop” sounds a little odd to me, but I don’t think it’s necessarily an Americanism.

  3. dw says:

    As you say, there’s an absence of any stressed words from MOUTH and also all the r-colored lexical sets (with the exception of FORCE)

    There are also no stressed words from the FOOT or PALM sets. And the words that do appear seem very biased towards the KIT and FLEECE lexical sets for some reason:

    Perhaps someone could come up with an extra sentence that fixes all these shortcomings!

    parentheses = unstressed

    KIT bring things (with) six thick big kids (into) (will)
    DRESS Stella fresh red Wednesday
    TRAP slabs (and) snack plastic (can) bags (at)
    BATH ask
    LOT (from) (of?) Bob
    CLOTH frog?
    STRUT (of?) brother
    FOOT (to?)

    FLEECE please these peas cheese (we) (need) (she) (these) three meet
    FACE maybe snake train station
    THOUGHT call also small
    GOAT snow go
    GOOSE (to?) spoons blue scoop

    PRICE five
    CHOICE toy

    NURSE (her)

    NORTH (for)
    FORCE store

    • Austin says:

      Of usually has STRUT in the U.S., with the exception of the South. Maybe that’s why you put a question mark after it.

    • trawicks says:

      I can’t tell if the inclusion of “frog” is supposed to be a nod to the CLOTH set or not.
      If so, I’d probably suggest they go with a voiceless fricative. Even in American accents with the CLOTH phoneme, “-og” words seem inconsistent.

  4. Austin says:

    Yes, I’ve never thought that was a good passage since the first time I saw it. Comma Gets a Cure is much better.

  5. m.m. says:

    I have to admit to falling to ‘backseat academia’ [god that sounds terrible in my ears] when I came upon the “worldwide accent project” on youtube Their passage is a bit better than the “speech accent archive” one; though I find the “International Dialects of English Archive”s use of the ‘comma gets a cure’ and ‘rainbow passage’ to be the better elicitation readings so far. A lot of the elicitation writings are ok for general vowels and consonants, but miss many of the pre-L mergers/changes or other conditional mergers that are particular of many accents.

  6. Julie says:

    It’s also a bit of a tongue-twister. I can’t read it without stumbling over the “thick slabs.” I wonder if that means people rehearse it a few times before recording it. At any rate, it’s a long way from natural speech. It seems like a longer passage with shorter, more natural-sounding sentences would be more useful?

  7. Austin says:

    This one, maybe?

  8. Amy Stoller says:

    I am an associate editor at IDEA, and proud to consider its creator, Paul Meier, and one of the writers of “Comma Gets a Cure,” Doug Honorof, among my friends. “Comma” went through an extensive period of drafting and redrafting before reaching its present form. It was specifically designed to include all of the words in J.C. Wells’s standard lexical sets. I, too, agree that it is superior to “Please Call Stella” in many regards, though it does contain a few words and phrases that even most native speakers of English stumble over. (I keep meaning to write to Doug about them.)

    “Comma” is written in American English, and I use it for gathering accent samples. I used to supplement it with passages from Real Accents, which are written in British English, but nearly all of my clients stumbled so badly over them that I now reserve those passages for my clients who are native speakers of BrE, or who learned BrE as their first English. (I live and teach in the US, so most of my ESOL clients want to learn a non-regional accent of AmE, but a couple prefer working on contemporary RP.)

    Hickey’s expanded lexical sets for Irish English are fabulous. When I teach Scottish accents and some regional accents of BrE and AmE, I expand the the sets myself as needed.

  9. trawicks says:


    “Comma gets a cure” is a lot more effective in my ears. I also like that it’s long. It gives you a much better idea (no pun intended) about various allophones and points of overlap.


    I’m deducing from the quote from the project’s director that it’s partially designed to be difficult to pronounce. It seems especially focused on consonant clusters and tricky word combinations. Which is useful if you’re researching L2 speakers, but arguably less so for native English speakers.


    I’m glad you mentioned Hickey! Some of his additions are useful for North American English as well–for example he makes a distinction between “price” and “pride,” which is becoming increasingly important in the accents that use a raised pronunciation for the former.