Rating American English Accents
« previous post | next post »
If you're a native speaker of American English, a Dutch linguist needs your responses to an accent questionnaire:
In this questionnaire we will ask you as a native U.S. English speaker to rate the pronunciation of different speakers, some of whom were born outside the U.S. We ask you to rate how native-like the pronunciations are. While we offer a set of 50 speech fragments, you are free to rate as few or as many as you'd like (of course we'd prefer more, but there is no required minimum).
Coby Lubliner said,
May 19, 2012 @ 10:44 am
The survey does not seem to be limited to native speakers (which I am not); in fact, the questionnaire asks about the respondent's language background. I tried the first ten samples, and not one of them sounded even remotely like a native speaker of American English (though one sounded like an Englishman). I didn't finish, because I really don't understand the 1-7 scale: to me, either the speaker sounds native or they don't.
Alacritas said,
May 19, 2012 @ 10:50 am
Just completed this. When I first filled it out, I put the non-American native-speakers (i.e. Brits and Aussies, maybe there was a New Zealander in there?) in the middle (i.e. on a scale from 1 to 6 I put them at 4). But then I realized that that made no sense, cause they actually *did* sound less American than some of the other people who weren't native English speakers, but who had learned an American accent to some level of competence.
It just felt strange rating them lower than the ones who didn't speak good English at all; but anyway I ended up switching them all from 4 to 1.
It's a rather confusing question, if they sound more American than some of the people who speak bad English but learned something that resembles and American accent.
Michael Briggs said,
May 19, 2012 @ 10:54 am
I did the whole questionnaire. If I remember, I gave all the speakers a 1, except for two who sounded almost US-raised (I gave them a 5 and a 6), and two who I rated as a 7 (US-raised).
#Coby, I have to disagree with you: I believe there are degrees of foreignness in accents. However, that said, for the most part I was able to enter a 1 on the rating scale after hearing only the first few words of the sample. The vowels and consonants in the first three words – Please call Stella – are dead giveaways.
Michael Briggs said,
May 19, 2012 @ 11:02 am
@Alacritas: None of the speakers uses "bad English." The vocabulary and syntax in each sample are identical. It probably does make a difference, though, if you rate the speakers not as to how "American" they sound but as to how "foreign." The Brit-speakers are unmistakably "foreign," to my ear.
Heather said,
May 19, 2012 @ 11:06 am
I have to admit, I found myself giving a higher score to native English speakers, even if they sounded English or Australian. Those I gave a 4, moderate accents I gave a 2 or 3, and those who could barely speak English got a 1. I gave out one 6 to a woman who had a very light lilt to her accent (possibly a woman who moved here from the Caribbean as a child?). I only gave two 7s and those were obviously native born Americans.
Kyle said,
May 19, 2012 @ 11:22 am
@Michael I don't know whether it was impossible for them to use "bad English" – some of the speakers I listened to, although they were reading the same text, dropped the "s" at the end of plural words and even seemed to change "the" to "a".
I found this a weird challenge – I mean, either people sound American or they don't. I did end up giving points for more natural-sounding English, even when it was clearly non-American English. But I wasn't sure if that made sense.
And I spent an inordinate amount of time trying to guess where the speaker was originally from and who they had learned their English from!
rukymoss said,
May 19, 2012 @ 11:30 am
I wish some audio examples of grading had been given. I ended up giving 2 sevens, 2 or 3 sixes, but if I could understand what was being said, I gave at least a 2. I found myself rating on speed, fluency, whether or not articles and plurals were clearly pronounced, so although most were clearly not native speakers of American English, there were still degrees of "foreignness".
Alacritas said,
May 19, 2012 @ 11:45 am
@ Michael: I have to disagree; considering some of their reading skills (as Kyle pointed out, some people dropped some sounds or misread them), it's clear that for some of the readers, their level of English is not good. If they read like that, I doubt that they'll speak with native-like grammar or that they'll have an incredible vocabulary.
As for your second comment, yes, that's the position I took in the end; the Brits do sound foreign to an American, so I had to give them 1s. It was hard for me to do so, though, as I've lived some years outside the US and am quite used to some British accents — and so they didn't really sound "foreign" at all to me. But of course the questionnaire was about American English, so…
Michael Johnson said,
May 19, 2012 @ 11:45 am
I finished the survey, but I was way too biased by the end. It was too obvious that most of these speakers were not native at all. If there were some curveballs– natives with non-standard accents (but not foreign ones), natives who had lived abroad for long periods of time– something to make me second-guess or scrutinize. But I knew that every odd vowel had to be Dutch (or thereabouts), and all it took was one odd vowel to instantly know who was native and who wasn't.
I guess what I mean is that it's hard not to interpret the question as "on a scale of 1 to 7, 1 being most obvious, how obvious is it that these speakers are non-native?" And such obviousness obviously varies with the context– I can more easily tell you're Dutch if you're hanging out with a crowd of 40 other Dutch people.
Sarah Taub said,
May 19, 2012 @ 11:48 am
I feel I did an OK job judging the "foreignness" of most of the speakers (roughly, how many different sounds and combinations did they struggle with), as it seemed clear that they were non-native English speakers. But I had no idea what to do with the 4 or 5 people who sounded like native speakers of non-US English. If the task had been to judge how much the person sounded like a native English speaker of any variety, they would have gotten 7's. I gave them all 5's, but I could easily have decided to give them 1's, as they clearly sounded foreign!
I think my problem is that I wanted to interpret the study as asking how much the speakers sounded like native English speakers. I wish I had had more guidance from the experimenter as to how important the "native vs. non-native" and "US English vs. foreign" dimensions were. Perhaps those two dimensions should not have been combined in one experimental set.
naath said,
May 19, 2012 @ 11:53 am
"sound American or you don't" > that's not quite true; I think there are roughly 6 possibilities:
*Native US accent
*US immigrant accent
*Native non-US English accent (eg, British or Australian)
*Immigrant non-US English accent
*taught English through US-accented material, but not actually lived in the US
*taught English through non-US-accented material, not lived in the US
Only one of these things is "native US", but I think people who have lived in the US a long time develop American accents over time – and the intermediate accents are clearly *more like American accents* than accents of people who have never lived in the US whilst not being 100% "American" yet.
grackle said,
May 19, 2012 @ 11:53 am
I too couldn't comprehend the scale; I started out with 2's and went back and changed them all to 1's except three that sounded to my ear as if they could each be native Americans, perhaps from different parts of the country. Almost all of them took only a few seconds to adjudge.
Michael Johnson said,
May 19, 2012 @ 11:58 am
Sorry for commenting twice in a row, but just one more thought:
I teach in Hong Kong, and while many of my students couldn't well be said to have an accent from any anglophone country, there are students who, while still having Chinese accents, also have British accents, or American ones (yet to meet an "aussie" Chinese student).
So I had difficulty interpreting 'foreign' here– a mainlander with an "American" accent sounds just as 'foreign' as one with a "British" accent, but the former sounds (obviously) more "American" than the latter. Would I rate them all '1' for foreign or would I rate the ones whose teachers had been American more highly?
D Sky Onosson said,
May 19, 2012 @ 12:33 pm
Hmmm, I didn't find the rating system all that hard – I used every number in the scale. I just tried to limit myself to answering the question "how difficult is it for me to understand what words this person is using?"
The non-American native English speakers were generally a 4 or 5 based on how different (subjectively of course) they sounded to my ears. Some of the northern European (?) and South Asian (?) speakers got a 5 or 6, but most speakers were in the 2-4 range for me, with a couple of 1s, and a couple of 7s.
mollymooly said,
May 19, 2012 @ 1:42 pm
I think the post heading "Rating American English Accents" is misleading; I expected it to involve different regional/ethic variants of American English, rather than varying EFL attempts to reproduce a GenAm target accent.
Anton Sherwood said,
May 19, 2012 @ 1:58 pm
I lived in a polyglot city (San Francisco) long enough to say that native and standard are not the same thing!
Rebecca said,
May 19, 2012 @ 2:11 pm
I didn't find it hard to rate, once I committed to focussing on whether the speaker sounded American or not (i.e. – reinterpreting "foreign" to include non-American native varieties of English).
But I only gave one 7 and one 6, and the rest were 4 or lower. I guess, I mostly heard the speakers as American or not, but within the "not", it was relatively easy for me to hear degrees of mastery of the accent. I was hoping there would have been samples aiming at more regional American accents, though. It would have made the task more fun.
I did find myself wondering what store poor Stella needed to find, though, to complete that shopping assignment.
Gene Callahan said,
May 19, 2012 @ 2:12 pm
"I mean, either people sound American or they don't."
I am really surprised that a couple of people think this. If you know people who immigrated here early in life, you know that this is just not true. I know a Jamaican family where you could run through the entire 1-7 scale on the children, guessing with complete accuracy their birth order by the "americanness" ranking (the ones oldest when they arrived sounding the least American, etc.).
Even I, having lived in the US and the UK, sound distinctly American to Brits, but Americans often guess I was born in the British Isles. So clearly, I sound American, but not level 7 American!
Faldone said,
May 19, 2012 @ 2:54 pm
I haven't started the test yet, but I certainly can see the value of rating the accent of a non-American native speaker of English. I know couple of British women who retain much of their British accents but have different degrees of American accent overlaid on that British accent. I also know a couple of Australian women who have the same. I have also heard Jean Redpath say that she used to go back to Scotland regularly to keep her Scottish accent up. When you hear someone who has an identifiable accent of a non-American variety of English it can be hard to filter that out and hear the American accent.
Anton Sherwood said,
May 19, 2012 @ 3:13 pm
Sometimes I give a point for good rhythm and pitch contours even when the phonemes are not quite right.
A few of the samples sound synthesized at first!
I hope the sequence is randomized.
mgh said,
May 19, 2012 @ 3:37 pm
frankly anyone who refers to getting a spoon of snow peas at the store is by definition foreign-sounding!
Chris said,
May 19, 2012 @ 3:43 pm
I'm a native AmE speaker and completed about a dozen of the speakers. None were native speakers. I have serious issues with the methodology. My primary complaint is that there were no fillers and no actual American English speakers (so far). I'm suffering attenuation. Yes, I agree there are degrees of foreignness for accents, but my judgments of severity are being influenced by the each successive trial. Normally this kind of problem is mitigated by creating lists which include both trial stimuli (foreign), some kind of baseline stimuli (actual American English accents), and unrelated fillers (mix of speakers saying sentences NOT related to Stella and her snow peas, these sentences would not be analyzed; they're just there to throw off the test subjects). However, these measures add considerably to the duration, and as such make a simple online format less feasible (typically you have to bribe students with extra credit or a few dollars to coax them into participating).
Jens Fiederer said,
May 19, 2012 @ 3:47 pm
They explicitly ask for native speakers, but ask questions within the survey to confirm – this might be like the Red Cross questionaire where answering certain questions in certain ways will get your unwanted contribution discarded?
They do prefix the instructions with "Please read the following carefully!" so if you waste your time submitting your evaluations when you are not a native English speaker, you have nobody to blame but yourself.
(I'm always conflicted on these, I'm technically not a NATIVE speaker even though English is my best language and most people don't detect any accent when I speak)
Jared W said,
May 19, 2012 @ 3:53 pm
I went through the survey fairly confidently, even naively, before reading these comments. And then I read Sarah Taub's: "I think my problem is that I wanted to interpret the study as asking how much the speakers sounded like native English speakers. I wish I had had more guidance from the experimenter as to how important the "native vs. non-native" and "US English vs. foreign" dimensions were. Perhaps those two dimensions should not have been combined in one experimental set."
Now I'm second-guessing myself. I really had no idea how to balance these two aspects, and I agree with her comment that they should have provided more guidance on separating the two dimensions: native vs non-native and American vs non-American. What I ended up doing was essentially rating them on strictly the native vs non-native dimension (because I've spent too much of my life outside the US, perhaps, to actually CARE about the US vs non-US dimension?). So I gave a lot of scores in the 4 to 6 range (as an ESL teacher I am perhaps too generous – if I can understand them then they don't sound THAT foreign – "foreign" is when you really actually can't even make out what they are talking about!). And then, if someone was a 6, I awarded a 7 to those who were unambiguously American-sounding to me – which was exactly two.
Ethan said,
May 19, 2012 @ 4:09 pm
The intended scoring system is not well explained. I ended up treating it as "assign a percent chance that this person learned to speak English while growing up in the US". So I marked the ones that sounded like UK or Indian subcontinent speakers at the low end of the scale, but ranked two hispanics as 6-7. That is, they had a strong accent in a sense, but to me they sounded like many people I know who grew up in the US Southwest. Most of the samples sounded to me like Europeans with varying degrees of fluency in English. A few sounded like possibly they grew up in the US with non-English speaking parents.
max said,
May 19, 2012 @ 4:30 pm
Perhaps the researchers are not actually concerned with measuring the nativeness of the speakers' English at all, but rather investigating whether Americans recognize degrees of nativeness or have a less nuanced concept?
Michael W said,
May 19, 2012 @ 5:17 pm
I tried to add some meaning to the numeric scale, and judge more on accent than proficiency. Though I think it influenced me as well. If there was someone that sounded to me like an immigrant (they had an obviously foreign accent but seemed more familiar with American speech patterns than, say, a native Irish speaker) I rated them higher. A binary choice would have been easy for almost all of them. There was only one that I thought was close – it could conceivably be American but judged a 6 since it sounded more like one of our northern neighbours. And I assumed USA, not N.A. since they ask for state.
It would have been nice to know for sure if they were all given the same sentence. I wasn't sure since one person seemed to react after possibly saying 'snake' instead of 'snack'; I couldn't tell if she'd just misread the nearby word or thought it odd.
John said,
May 19, 2012 @ 5:18 pm
I grew up in parts of the US that were strongly influenced by immigrant populations who came into the US two or three generations earlier–Irish and Scots in particular, but also Italian and French Canadian. I also lived in areas with high numbers of more recent (i.e., first generation) immigrants from Eastern Europe. I now live in an area where there are many 1st-to-3rd generation immigrants from Latin America.
I can accept a certain amount of accenting as 'Native', realizing that it is heavily colored through living in ethnic neighborhoods. Contours played at last as heavy role in my grading as did pronunciation.
Chris said,
May 19, 2012 @ 5:39 pm
@max, I too wondered about what the goal was (another good reason to use lots of fillers, to dissuade subjects from gaming the system by trying to out-guess the test). I think there is great potential is correlating foreign-soundingness with multiple phonetic features and discovering significant patterns. As has been hinted at above, prosody features seems to be highly correlated with understandability (more so than individual phonetic segments, I think).
But this raises the age-old stink of validity: does this kind of simple experiment actually measure foreign-soundingness? What is the distinction between foreign-soundingness and understandability? Adding samples of American English speakers mumbling these sentences might help tease apart this issue.
Aaron Toivo said,
May 19, 2012 @ 6:20 pm
Don't know if I got the exact same set that everyone else did, but there were three unmistakably native US English speakers in mine. I could identify which state, for two of them.
Michael Watts said,
May 19, 2012 @ 7:54 pm
I too don't understand the rating scale at all. I could interpret it as "how likely do you think it is that [the speaker] is a native speaker of US english, in which case 7 points is far, far more resolution than the scale should have, or "how much difficulty do you think a native speaker would have in communicating with [the speaker]", which makes more sense for a 7-point scale, and is an extremely important question, but isn't really indicated by the instructions.
Someone with an obvious accent and fairly natural-sounding delivery can convince me that they know what they're saying and that I'm unlikely to suffer any major communication mishaps in talking to them, but they're not going to convince me that they're a native speaker. Should I rate them 1/7? 3/7?
JR said,
May 19, 2012 @ 8:18 pm
Wow, hard at first! Then I recalled that at a German university I was at, one could study either American or British English. So then I imagined that I had been asked by a teacher teaching American English to rate her students based on how well they were doing with the American accent. This meant that I gave native British speakers a 1. I gave foreign beginner-level students a 1 basically only if I thought their accent greatly impeded intelligebility. If it didn't, they at least got a 2.
John Bauman said,
May 19, 2012 @ 8:21 pm
I wonder if doing a series of (randomized) pairwise comparisons would be better for getting rankings than trying to sort each individual accent from 1-7. It was pretty difficult trying to stay consistent in my scores.
Teresa Elms said,
May 19, 2012 @ 8:22 pm
Like so many others, I found the 7-point Likert scale ambiguous. Is the researcher asking whether the speaker IS foreign or merely SOUNDS foreign? I opted for the latter. This created two interesting dilemmas for me that I believe the researcher was attempting to probe with this survey.
First, as a resident of southern California, I hear a great deal of non-native English at all levels of proficiency and non-proficiency every day. The sound samples in the survey were, by comparison, all pretty good English. Partly that's because all the samples came from reading a cheat sheet of text rather than generating novel speech from scratch. But these samples were also scrubbed to make them comprehensible — a condition that does not hold in the real world. To me, a score of 1=so badly formed that I can't parse the words out of the noise. None of these samples approached that. So I mentally rescaled the scores so that 7=native speaker, per instructions, but all the others were degrees of foreignness randomly distributed around a mean of 4, where 4=the average level of foreignness I hear in southern California. 1=a non-native math teacher who understands English well but whom I can't understand if I don't already know what he's saying.
The non-native English speaker with the near-perfect American accent therefore scored a 6, while the Brit and the Aussie each got a 5 — despite the fact that the latter were both native English speakers.
I think that felt right to me because foreignness is a rather American thing in southern California. An American accent signals not only migration to this place but also a high degree of assimilation: a "here-person." A British accent, on the other hand, is likely to belong to someone who has either never lived in the States or who lives here but retains the British accent as a differentiating marker of high-class status.
Then I hit the Texan. To a Californian, the normative Texan is socially and politically incomprehensible. A foreigner. But, dang it, an AMERICAN foreigner. So I gave him a 7, even though emotionally he would rate a 5 with the Brits and the Aussies.
It will be interesting to learn with our researcher makes of this.
Liz said,
May 19, 2012 @ 8:29 pm
I completed the entire survey and I think I found only 2 native speaker, one of whom was from the South. "American accents" are quite particular, if you have one of those, then yes, you speak like a "native". If you don't, your English is marked as non-native. Which is not to say that that English is "bad", or "ugly" or "incomprehensible". It could be quite good – clear, easy to understand, it might also mean that you come from an English-speaking country. But nobody is going to mistake a thick Aussie accent for an "American" one.
Andy Averill said,
May 19, 2012 @ 8:33 pm
I agree, too many levels, and too many samples. I got as far as I could stand. But then I discovered you could play all the samples at the same time — that was so fun!
Chris Waters said,
May 19, 2012 @ 9:19 pm
Yes, this was hard. I think the hardest part was distinguishing fluency from accent. I started out downgrading for poor fluency, but then realized that one non-fluent speaker was hitting American vowel sounds much better than the previous, very fluent, speaker. So I had to go back and review all my previous ratings. Another thing that made it difficult was the wide variety of accents. Relative ranking between people with similar accents was easy, but when they had quite different accents, it was tricky to decide which was more, or less, American-like.
I couldn't rate any of the Aussies/Brits lower than about three, because, no matter how far off they were on the vowels, they still did better with the consonants than many of the other samples. (Except 'R', of course.) At the same time, I couldn't rate them much higher.
I'm pretty sure the samples were randomized, since some people have mentioned not hearing any native accents at all, while I got a distinctive NYC accent as my second sample. I also heard one that had me fooled until the third or fourth sentence, when he finally revealed some distinctly non-native markers.
I thought it was interesting that some of them seemed to be hyper-exaggerating American quirks, like the one who put a very hard "ch" at the beginning of "train", I didn't give her any bonus points for accuracy, but I did mentally give her an 'A' for effort. :)
@Michael Watts: I didn't interpret the question as either "how likely do you think it is that this is an American accent?", or "how much difficulty do you have understanding this?", but rather, exactly what was asked: how much is this like an American accent?
JR said,
May 19, 2012 @ 9:58 pm
Doing this brought up a lot of interesting topics in my mind, some of which have been touched on in other comments:
1. Can an adult learn a foreign language and acquire a native accent that even fools native speakers of the country in question? I haven’t met a single one. But, note, this involves making sure this individual didn’t spend a few summers in the US as a kid or didn’t have American playmates as a kid, etc.
2. When learning a foreign language, usually it is very, very hard to distinguish between the various native accents at the beginning. It gets easier with time, but can a fluent non-native speaker correctly guess who is a 6 and who is a 7? For instance, I’ve known people who grew up in the UK, but this only shows up in their accent in just a few words. How well can a fluent non-native notice those few words?
3. Similarly, how well can I fool natives into thinking I’m a native speaker when speaking a foreign language? I’ve done this a few times, but it is probably because they were only short conversations. I learned Mexican Spanish, for instance, and a cab driver in Spain asked me if I was Mexican.
In short, I agree with the commenter who said “Either you sound American or you don’t.” This has nothing to do with where you were born or whether or not you are a citizen. A native speaker of US English should have no problem identifying a native US accent, but can others?
Steve Kass said,
May 19, 2012 @ 10:01 pm
The survey and this comment thread would be a great example to use in a course on survey design. Beyond that, I'm not so sure.
It's clear from the comments that the instructions aren't clear to participants, who chose their answers based on a variety of measures.
The written instructions were at the least inconsistent and confusing. The preamble says
Above the rating form, something else is written:
These are very different requests, in my opinion.
If pressed to guess what Martijn wants rated (before participating), I'd guess it's the nativeness of the speaker, not the nativeness of the pronunciation, because he also writes "Also note you don't need to listen to the complete audio samples if you feel you are able to rate the nativeness of the speaker earlier." He also notes that some of the speakers "were born outside the U.S." (thereby suggesting the some were not).
On the other hand, "nativeness" makes less sense to me as a scale variable with 7 distinct values when applied to a speaker than when applied to pronunciation. There's little middle ground, I would say, regarding the nativeness of a person, but quite a bit of middle ground regarding pronunciation.
Beginning the survey, I'm confused even more. The descriptions of the values 1 and 7 are not opposites or extremes of any clear scale
After listening to the samples, I'm yet more confused. Almost all the speakers were clearly not native speakers. For all but a very few, I'd wager 100-1 or higher odds that (if these were their natural reading voices) they were not native speakers of American English, and "very foreign sounding" (score of 1) characterizes them well to me. How can someone be anything less than "very foreign sounding" if they sound unquestionably foreign?
When I see this sort of confusion in a survey instrument, I wonder if even the study author has a clear understanding of what he or she wants to measure. (I also realize that few participants will puzzle over the task at hand like I do. In some cases, most participants will behave similarly regardless of the instructions, because there is some familiar measure to be assessed, but that's not so much the case here, especially given the clips.)
I'm not sure Martijn knows what he wants to measure, given the audio clips, which exhibit almost no degree of American (as opposed to non-American) English. Maybe he'll pipe up here and clarify his goals and then his survey.
In any case, it's a shame, unless, of course, Martijn's goal is to gain some understanding of how native American English speakers might interpret and implement unclear specifications about accents like "foreign sounding" and "nativeness," or to look for previously unidentified aspects of accents that set them apart with respect to undetermined measures. All of which might be very interesting (but possibly not interesting enough to justify unnecessarily frustrating participants in the research).
A fair bit of research suffers from this sort of problem (assuming no hidden deep goal of this research), though the problem is obscured when published results don't include complete survey instruments. (Pollsters have the edge over authors of peer-reviewed journal articles on this point. Pollsters almost always publish their survey instruments.)
JR said,
May 19, 2012 @ 10:11 pm
@Steve: It seems like we didn't all get the same recordings. For instance, of the English learners, I thought ALL of them, except for two, exhibit American English pronunciaiton, whereas you said "almost no degree of American."
Steve Kass said,
May 19, 2012 @ 10:13 pm
@Chris Waters,
Where do you see that this is "exactly what was asked"? In the survey I got, there are no phrases resembling "like an American accent." As much as I suspect this could have been the goal, it was not exactly what was asked.
Garrett Wollman said,
May 19, 2012 @ 10:20 pm
I went to the page and noted that all of the audio clips were hidden behind "unknown plugin" blocks (which for some reason Opera renders as an iframe with the text of the SVG for the "unknown plugin" icon, rather than rendering the icon itself). Having no interest in digging through the form's HTML source to figure out the actual URLs of the clips, I closed the tab without bothering to read the instructions.
JR said,
May 19, 2012 @ 10:20 pm
@Steve: It says right there at the top: "American English accents." That's what the survey is about. Then it says: "rate how native-like the pronunciations are," i.e. rate how native-American-English sounding are they.
Steve Kass said,
May 19, 2012 @ 10:54 pm
@JR: I listened again, and I think you're right. Many of the speakers do seem to be learning American English as opposed to British English. In almost every case, though, there is another more pronounced "accent," which ranges from Indian to Latin American to European to African to Carribean to French Canadian. (I realize only some of these are accents that exist among native English speakers.) Revisiting these after your question makes me realize how hard it is (for me) to ignore the pronounced "foreign" accent and focus on the question of what kind of English the speaker is likely learning.
Steve Kass said,
May 19, 2012 @ 11:06 pm
@Garrett: The URLs mostly range from http://www.let.rug.nl/accents/soundtracks/L01.mp3 to http://www.let.rug.nl/accents/soundtracks/L50.mp3, but a few files have letters appended (L03B.mp3, L25I.mp3), and in any case, they appear to be randomized for each visitor. Both Firefox and Chrome rendered players on the survey, but neither browser was particularly happy to find dozens of Flash objects on the same page.
Nathan Sanders said,
May 19, 2012 @ 11:30 pm
I did the whole survey and found four that sounded like native American English.
As for the rest, I rated them based on how long it took me to determine that they were foreign. Nearly all were 1s (which translates to being able to determine it from nore more than the first four words. There were eight 2s, most of which took me until I heard a velar fricative in "her" to be able to tell they were foreign. But "her" was only the fifth word, so those 2s could arguably be 1s.
LS said,
May 19, 2012 @ 11:52 pm
As an English teacher, it seemed like a reading test to me. Can these people read this sentence well (i.e. with a suitable American accent) or not? The other problem I had – like others here – was determining just how foreign a native speaker of English not from the US sounds. Do Americans really consider a British accent 'foreign'? Or is it simply a different, yet acceptable accent?
Peter Taylor said,
May 20, 2012 @ 2:46 am
Teresa Elms wrote:
There may be a certain extent to which Brits consider all "foreigners" to be of an inferior social class, but I don't think that's really what's going on. To me it seems more likely that native English speakers from other countries feel less need to adapt their accent because as far as they're concerned they don't sound foreign. So whereas other immigrants may make a conscious effort to sound more American, Brits alter their accent unconsciously and more slowly.
Perhaps this can be tested by comparing the rate at which their accents change to the rate at which their use of distinctively American vocabulary changes? Anecdotally, I've observed that Latin Americans in Spain make some vocabulary changes within a few months.
Alan Curry said,
May 20, 2012 @ 3:35 am
Yes, Britain is foreign, i.e. not the 51st state yet.
I'm another non-survey-taker who just didn't want to jump through the hoops. No, I won't run your Javascript or your SWF or whatever other dumb thing would have come next if I hadn't been stopped by those. Repeat after me: A HREF=FOO.MP3 dammit. I'm willing to listen to your sound files, but not willing to execute your code.
The mp3 URLs weren't actually hard to dig out, but the need to do so is a deterrent. Once they've decided that Javascript is mandatory they're gonna use it for EVERYTHING, so even if I can answer all the questions my submission will probably fail and it will have been a giant waste of time.
Teresa Elms said,
May 20, 2012 @ 4:09 am
I'm nearly certain that one and probably two of the sound samples were computer generated. The one I'm nearly certain about was one of the most native sounding, without obvious tempo problems, but there was still something subtly strange about it. The one I'm less sure of was more "foreign" sounding, but not by no means extremely so.
So it may be that this project is not the simple perceptual assessment it appears to be. If you embed computer-generated samples that are DESIGNED to be "more native" or "less native" within a larger collection of human samples, any ratings of foreignness for computer versus human can be analyzed within rater. That would ameliorate the problems of interpretation presented by that blasted Likert scale. With some clever regression analysis on purely physical acoustic dimensions rather than psychophysical auditory dimensions, the researcher may well tease out new criteria for what makes a MACHINE voice sound more or less realistic.
Teresa Elms said,
May 20, 2012 @ 4:31 am
BTW, my comment about the status value of a British accent in the United States refers to a cultural feature of the United States, rather than any particular attachment to home or attitudes concerning civilized versus barbarian behavior retained by individual Brits. Americans adore any flavor of Britishness in speech, and we provide a wide range of social incentives that encourage Brits to retain that speech pattern.
BenHemmens said,
May 20, 2012 @ 4:55 am
Can an adult learn a foreign language and acquire a native accent that even fools native speakers of the country in question?
Oh yes. I don't think anyone manages it (as an adult) without a special effort, and possibly special training. I've known one English native speaker who speaks German with a flawless accent; she's a classical singer, and they work hard on these things. And I'm married to an Austrian whose accent when speaking English comes across to most people as very neutral; but she's a lecturer in Anglistik and they also have systematic pronunciation training as part of their degree.
(I for my part speak German very fluently with a very noticeable foreign accent)
David Eddyshaw said,
May 20, 2012 @ 7:11 am
@BenHemmens:
I was on a bus in Anatolia once and was greeted from the seat behind me in what I took to be a completely authentic Yorkshire English accent, which turned out to belong to a Turkish soldier who'd done a science degree in Leeds. I also once had a Jordanian colleague who had an accent which seemed like some subtly unfamiliar regional English dialect one couldn't quite place, but certainly not foreign; he'd come to the UK at 18. In both cases there had been no special training as such, just lots and lots of exposure; both had essentially been interacting almost exclusively in English for several years.
Mind you, the Jordanian chap possibly was aided by a noticeably polyglot family background; he was Circassian, and spoke native-level Turkish too, among other accomplishments.
(His grandmother was the last person left in his family who could actually speak Circassian; according to him, it sounded like someone with a bad cough …)
John said,
May 20, 2012 @ 7:46 am
1) I also use Opera, but I checked "enable plugins on demand", so I only load one flash object at a time. It was extremely fast at the start, but by the time I got to the second half, the delay became noticeable. This may just be because I only have 1GB RAM (remember the days when 128MB was loads?)
2) I am not American but grew up with US TV shows. I didn't listen very carefully or for more than 5 seconds for most of them, but there were definitely 2 Brits – one RP and one Scottish (or maybe Irish). I also identified one southern and two general Americans (one slightly foreign who I would have given a 6). I couldn't identify an Aussie for some reason.
However I thought I could hear a Shanghainese, Beijinger and northeastern Chinese. There was also possibly a Vietnamese. (I'm ethnic Chinese British-Australian) I suppose I could make fairly good guesses at who was a Germanic or Slavic speaker etc. It would be interesting to see how accurate Chinese are at identifying where in China someone comes from based on their English accent. Would Indians be better at this since English is spoken more natively in India?
Eneri Rose said,
May 20, 2012 @ 9:46 am
I have lived my whole life (58 yrs) in the USA. My parents were native AE speakers. I read all of the above comments before I took the survey. I approached the survey as though I were a teacher of AE who was now grading her students on how well they were speaking AE. I found this to be pretty straight forward. I gave 3 of my "students" a 1 (a woman and 2 men) for speech that was indistinguishable from a native AE speaker. I reserved a grade of 6 for students whose speach was so close to AE as to make me wonder whether they were native. I gave no grades of 6, but I used all other grades (1-5).
I gave 5s to speakers I would assume were native speakers of other English dialects simply because I found their speach to be closer to AE than the other speakers. I gave 1s to speakers that I actually had some difficulty understanding.
I had fun doing this survey. I was paying attention to the characteristics that indicated to me the native language family of the speaker. Obviously, these are the things the speakers need to work on the most.
Once, a coworker of mine asked me to handle a phone call because she could not understand the accent of the caller. As I was talking on the phone, my coworker started laughing. After I hung up, my coworker said that the longer I spoke with the caller, the more I started talking like him.
All this reminds me of The Big Bang Theory episode in which Rajesh, who is a native speaker of Indian English, complains to Sheldon that he would be winning their argument if they were arguing "in my native language." To which Sheldon retorts that English is Raj's native language.
Leo said,
May 20, 2012 @ 11:02 am
I'm from England, and none of the voices sound to me like native British speakers of any variety. Which clips sound "British" to the native AmE speakers?
Morten Jonsson said,
May 20, 2012 @ 11:47 am
It isn't that some of the voices sound native British. It's that the English the speakers learned was evidently British English.
Chance said,
May 20, 2012 @ 1:12 pm
I'm with some others; either someone sounds like a native or they don't. I gave some a one, some a two and two or three of them got a three, but that's it. Except for two near the bottom who I suspect of being Americans and got 7s. So the graded system seemed not helpful to me.
I wonder how many people rated the clearly British or Australian voices as less foreign, even though they are just as foreign as a Chinese or African accent?
Mark F. said,
May 20, 2012 @ 2:26 pm
I am one of those who decided that, as far as I was concerned, a native-speaking British accent was more like a native-speaking American accent than was a pronounced non-native American accent. They were really asking about two different variables, native-sounding-ness and American-sounding-ness, and it was easier to prioritize one over the other than to try to estimate the vector norm.
If you are in the "either they sound like a native or they don't" camp, do expressions like "thick foreign accent" or "almost no accent" have no meaning to you? To be sure, my ratings were almost all 3 or lower, but for me 1's were rare because I have heard people with worse accents than any of those in the study.
Chris Sundita said,
May 20, 2012 @ 3:42 pm
The vast majority of the recordings were of non-native speakers of American English, so I gave them 1s. Even the people who were British & Australian. I gave the Texan/Southerner a 7 and to those who spoke other regional US accents. There was one instance where the person sounded like she moved to the US in her teens but still had a noticeable accent – I rated her a 4 or 5.
Chance said,
May 20, 2012 @ 6:30 pm
Mark F. said, "If you are in the "either they sound like a native or they don't" camp, do expressions like "thick foreign accent" or "almost no accent" have no meaning to you?"
Of course they do. But to me, the question, "How native are these speakers?" is a different question than "How thick is their accent?". I think the former was asked. So to me, they were all in the non-native camp.
blahedo said,
May 20, 2012 @ 7:59 pm
I started out giving 1s, treating the scale as something like "certainty that this person is native/non-native AmE speaker", but quickly saw that it would be nothing but 1s and 7s if I did that. (Looking back, there probably would have been one or two in the middle, but those came later.) So I went back and revised it to something like 7=certain they're native AmE, 6=really close (either a teenage immigrant or someone who's worked really hard at it), on down the line. I thought that for a majority of them the variety of English they'd been taught was not American, which counted as a "dead giveaway" and thus made them max out around 4 or 5 even if otherwise nearly perfect.
The only ones I heard with anything like a regional American accent sounded quite authentic (I had two, a Texan and one that was I think either a Tennesseean or a southern Virginian), and got 7s; I wonder if the best strategy for "sounding native" is to aim not for the high-standard dialect but for a regional dialect, which most people don't have as good an ear for and thus might be more easily fooled.
lucia said,
May 20, 2012 @ 10:06 pm
I interpreted this as follows:
If these people were cast as American's in a movie, would I believe they were an American? If I would absolutely believe it, I gave them a 7. If I absolutely would not believe it, I gave them a 1. Had anyone sounded like they were from the UK, I would have given them a 1. This is not a value judgement about good or bad nore even facility. It was just a question of "Would I believe them cast as an American in a movie?"
British accent are not American.
Nearly all recordings earned 1s and I checked 1 within the first 4 words. It wasn't even close. I may have given a few 2's to people who managed 10 words before you could tell. But this was rare.
I think I gave two or three 6's to a few people who sounded like BBC actors trying to sound American or possibly foreigners who got pretty close. (Or maybe they are American's from some part of the country whose accent I haven't heard. Maybe Canadian. It's just a few sounded like they might be American, but I just couldn't be sure.)
I gave a few 7s. (I think one of the American's may have been from Arkansas. ) I gave no scores between 3 & 5.
If I'd been asked to rate for sounding British, none would have gotten a 7 from me. That said, I'm not familiar with all possible British accents. Some sounded Indian. Some sounded 1/2 way between Indian and British, which I thought would be someone from India who moved to the UK and adapted. They didn't sound American; I gave them 1's within the first 4 words.
Chris Waters said,
May 20, 2012 @ 10:19 pm
@Chance: again, it did not say, "how native are these speakers". It asked how native-like they were. If you can admit the existence of thick vs. mild accents, you can grade degrees of native-likeness. Which is what was asked for.
BenHemmens said,
May 21, 2012 @ 4:28 am
@David Eddyshaw:
Yes, I'm sure there are cases where people pick up an accent accurately by themselves, but I believe it's pretty rare for someone to achieve a native-like accent and pronunciation without any noticeably odd features at all.
It may have to do with the way the person hears their own voice. When speaking German, I hardly notice my foreign accent but when I hear a recording of my voice (or speak through a pa system), I can hear very clearly what's not right. Maybe some people hear themselves more objectively.
Ken Brown said,
May 21, 2012 @ 8:49 am
How can "…either someone sounds like a native or they don't." possibly be true?
What about a native speaker (of any language) who moves to another country and whose accent changes? They won't sound like a native speaker any more, but it will be gradual process, not a sudden flip. Americans living in England often don't quite sound like Americans any more (not do they sound like English people living in America – I'm not sure what the difference is, but I suspect that people going one way gain and lose features in a different order from those going in the other)
What about people living near a political boundary that is not a language boundary. On each side "native" to their own part, but foreign to the other, yet with similar accents. Again a cline, a gradual change. This surely exists in some parts of the US/Canadian border. (and of course most non-North-American English speakers can't tell US from Canadian accents anyway)
Others have mentioned recent immigrants. But what about some non-immigrant native speakers of US English who mostly use Spanish at home and have an accent, when speaking English, that many Anglo-Americans might not think native? There are, I think, Spanish-speaking communities in Texas and New Mexico and probably other places that have existed for centuries.
Jonathan Mayhew said,
May 21, 2012 @ 9:50 am
I am puzzled by the dichotomous thinking here. I know people whose English has a faint patina of foreignness, so by the logic of some commenters here I would have to put those people as either totally foreign (because they are) or totally native Americans (which they aren't). In any case, a badly designed survey because its instructions are so ambiguous.
JR said,
May 21, 2012 @ 11:44 am
@Ken: A native speaker is a native speaker. If an American moves to England and takes on aspects of a British accent, it is no longer a native American accent. But he is still a native speaker.
As for the border thing, you didn't give a specific example, so I'm not sure I understand, but I did question if one is able to pick out a native accent in another language or, in this case, another dialect. I'm not a Southerner and have never lived in the South, so, maybe I couldn't tell if someone was faking a Southern accent (though I could probably tell if a foreigner was faking it.) In the same way, a German in the north with a standard German accent may not be able to tell if someone way in the south is from the German side or the Austrian side.
As for immigrant children or ethnic minority groups: I'm not saying that there is only ONE native, US accent, obviously. Some have said that certain Asian Americans have a certain accent and having lived in California, I think I know what they are talking about. But it was still American-sounding. And I know some families who lived in the southwest when it was still Mexico, and, well, I haven't noticed a non-US accent. In any case: Why are you bringing up things you have no personal knowledge about? Simple question: Have you ever met anyone that you thought was from a different country but turned out to be a native US speaker? (And again, I'm not talking about whether that person is a US citizen or not, who may have spent time in another country.)
@Jonathan: Only native US accents get a 7. Foreigners with almost perfect accents get a 6. Can a foreigner who learned English as an adult get a 7? Well, as someone mentioned above, maybe actors or people trained in singing can, but I've personally never seen it.
Morten Jonsson said,
May 21, 2012 @ 1:41 pm
My German former girlfriend is extremely–miraculously–fluent in English. She would be miffed when I told her that some other Germans, though not nearly as fluent, actually sounded more American. But they did. I thought of that in answering this questionnaire, because some of the speakers, though clearly not comfortable in English, still made recognizably American sounds, while others, with much better command of the language, didn't sound at all like Americans.
ohwilleke said,
May 21, 2012 @ 6:40 pm
"Can an adult learn a foreign language and acquire a native accent that even fools native speakers of the country in question? I haven’t met a single one."
I do think that the threshold is fairly fluid. My inlaws both came to the U.S. as medical residents at ca. 24-25 years of age. One is distinguishably not a native speaker in a lengthy conservation, but could pass as a native for a few sentences and is clearly totally fluent. The other has much less fluidity, but enough to conduct a professional career and reasonably ordinary private law mostly around people who spoke exclusively in English. I know of a man who moved to Peru at around the same age who went onto marry a local woman who is reportedly close to a native speaker accent in Peruvian Spanish.
Two important factors for when the cutoff age hits are the prior history of language learning (someone already bilingual can reach native speaker accent a bit later), and the intensity of the immersion of the learner (the more complete the immersion the better and an intense romatic relationship with a native speaker will often make the difference at the margins). There are also just differences in personal abilities to learn languages to some extent – the door closes for some at 15 and for others at 25, even if all other things are equal. There is also a legitimate argument, although I haven't seen any studies to that effect, that classroom instruction from a non-native speaker prior to immersion can actually be counterproductive because it ingrains bad habits before good ones take hold.
ohwilleke said,
May 21, 2012 @ 6:47 pm
"British accent are not American."
Following up on the score on the can adults sound like natives, I know multiple native speakers of the English language who have completely banished the accent of their childhood in favor of another accent of the English language (or can switch it on and off). For example, I've known Aussies and South Africans who moved to the U.S. as adults who can speak in a Native speaker American accent, and I've known Americans who have acquired pretty much native British speaker accents as adults. I also know a fair number of people who have managed to wipe a South Asian, American South, or New York accent in favor of a different accent. Quite a few actors learn to do this as well.
lucia said,
May 21, 2012 @ 9:31 pm
Never. Ever.
I know people who did not grow up in the US who mastered American accents sufficiently to pass for American's most of the time but I have never identified an accent as foreign only to discover that person grew up in the US. Some vowels, consonants and cadences do not match any American accent.
Aaron Toivo said,
May 22, 2012 @ 4:05 am
For a moment of comic relief after all this discussion, there exists a recording of the Stella text being read with a voice appropriate for… well, you'll see.
http://www.kneequickie.com/w/kq/images/c/c7/Stellajlynesexy.mp3
Bobbie said,
May 22, 2012 @ 8:25 am
I heard several native Americans, some who had definite regional accents. Since Brits or Aussies or Canadians are not "native" Americans by my definition, I generally gave them a 5 or a 6, but not a 7.
BenHemmens said,
May 22, 2012 @ 10:43 am
native speakers of the English language who have completely banished the accent of their childhood in favor of another accent of the English language
Yes, that's possible. But it's a complex process. What non-experts perceive as "accent" is probably a composite of several different phenomena. I lived on the east coast of Scotland for some years. A while after that, I once took a phone call from a Scottish friend while my brother was in the room. He commented on how I had apparently switched completely to Scottish. But whenever I raised the issue of whether I had picked up a Scottish way of speaking with my Scottish friends, they had fits of laughter before they could even attempt a reply. For them my way of speaking is more or less the epitome of Irishness.
Probably what I picked up best was the sentence intonation and rhythm, and the different vocabulary. This sounds like a very significant shift to a non-native of the target region but not at all convincing to a native. The full deal would involve not only these aspects but a set of specific pronunciations, including the effects of neighbouring sounds on them; exact collocational and syntactical properties of dialectal expressions, and overall voice timbre. It's certainly not impossible to pick up all the dimensions accurately, but most people are likely to have weaknesses in one area or the other – I'd guess mine are in some specific areas of pronunciation and the overall sound of my voice.
(For a foreign language, add complete mastery of the standard and relevant dialectal lexicon and grammar, and awareness of different registers – slips in any of these areas are liable to make native speakers listen for what is odd about the way you speak).
Victoria Simmons said,
May 22, 2012 @ 1:40 pm
I was looking for an American accent. If they were almost unintelligible they got a 1, but most were 2s or 3s. There were a few who had a natural American-sounding rhythm and tone to whom I gave 4s. I gave no 5s or 6s, although I would have if I had heard any samples that were American except for a trilled 'r' or an awkward 'th'. I too had trouble trying to figure out what to do with non-American native speakers of English, and I probably rated them higher than I should have.
My selection had, I think, 5 native American-English-speakers, 3 of whom were Southern.
That made me wonder if the purpose was to see how "American" regional American English sounds to native speakers, depending on where the respondents were raised. But then I would have expected to hear some Baltimore or New Orleans accents, both of which can be hard to place. Or the even-harder-to-place traditional Outer Banks accents of North Carolina or Gullah-influenced English. They are all American. There are also native speakers of American English whose accents are definitely Spanish- or French-tinged–sometimes more a matter of rhythms than pronunciation.
On the other hand, I teach a writing enrichment course to the children of Chinese immigrants. Most of the kids were born in the US, but some immigrated, some as late as the age of 6. Most speak Chinese at home. All, without exception, have perfect American-English accents.
Victoria Simmons said,
May 22, 2012 @ 1:42 pm
I have an old operating system on my computer, and as I proceeded things slowed down more and more. The submission took so long that it probably failed. Frustrating.
Steven said,
May 22, 2012 @ 10:57 pm
Just to confuse some of you:
http://newstalgia.crooksandliars.com/gordonskene/newstalgia-reference-room-eleanor-roos
blahedo said,
May 23, 2012 @ 11:23 am
@BenHemmens: I had a friend in grad school (in the US) who was from England, and while he was (of course) perfectly intelligible, nobody around us would have said he was anything close to a native American accent… but when he called home once to a relative still in England, his relative's flatmate called out, "hey Xxxx, some American wants to talk to you." It's the curse of being in the middle; we're better at distinguishing "us" vs "not us" so people on both sides will categorically perceive the differences from "us".
lucia said,
May 23, 2012 @ 5:32 pm
Was the that the actress who played Aunt Bea in Mayberry RFD cast as Eleanor Roosevelt?
Michael Briggs said,
May 23, 2012 @ 8:03 pm
@Lucia: Not confusing at all. Mrs Roosevelt spoke with a perfectly authentic US accent — the accent of New York high society. I don't see how she could be mistaken for anything but a native US English speaker.
Bloix said,
May 23, 2012 @ 9:40 pm
Mrs.Roosevelt's accent is extinct. I doubt there's a single living American who has that accent.
lucia said,
May 23, 2012 @ 9:47 pm
Michael–
The confusing claim was me quoting someone else. I disagree with it– I don't think anyone would think Eleanor Roosevelt sounded like anything other than American.
I think Eleanor Roosevelt reminds me of lots of actresses I heard watching black and white movies. I think she sounds like Aunt Bee. The actress is listed as being born in NY, NY in 1902.
SSH said,
May 24, 2012 @ 3:28 am
I read all the comments (except for a couple that didn't seem to be going anywhere) and then listened to the survey. I'm a native U.S. English speaker. I counted exactly six perfectly native accents. I listened very closely to make sure, because everyone else has said they heard three at the most.
I don't understand the confusion. The survey could be improved, but it was very straightforward. The scale should clear up any potential misunderstanding. If he wanted to know whether you thought the speaker was native or not, there would only be two options, not seven. The objective was not "guess which are foreign and which are native." There was an enormous difference between the nativeness of the accents. Some were very native-like, but not perfectly so, and others were very foreign. Most were somewhere in between. Only a horse is a horse. Every other species is a nonhorse. But certainly a camel is more like a horse than a jellyfish. We can compare which is more horse-like. How is this concept really that difficult to grasp?
Walt said,
May 24, 2012 @ 6:55 am
I think seven levels may be too many from a statistical modelling standpoint, but FWIW I tried to apply the following system to the full set:
(1) barely recognizable as English of any native-speaking region;
(2) many words difficult to understand;
(3) some words difficult to understand;
(4) a few words difficult to understand;
(5) fully understandable but inconsistent, e.g. peas & cheese not both /z/;
(6) consistent and correct but not quite American to my ear, e.g. vowel length;
(7) apparently American.
I didn't give any 1s though several 2s. I think I marked four 7s.
EorrFU said,
May 24, 2012 @ 9:23 am
The Southern accents caught me out the worst. I felt like I had been primed to down-rate those examples because they sounded strange at first but then I realized that there is no other place but the US those accents could be from.
nyb said,
May 27, 2012 @ 5:56 am
JR says:
Perhaps you've met many, but were fooled.