 The methods involved in dialect geography are the same as in many other branches of linguistics. Basically, we are concerned with methods of data recording and data analysis. However, the way the data is collected differs from other branches of linguistics, especially due to the amount of information required. The following variants of investigating dialects will be discussed in this e-lecture. First, we will look at questionnaires and then we will briefly deal with the technique of using the telephone in dialect surveys. Let's start with questionnaires. From the very beginning, interviews of informants by field workers have been conducted via questionnaires, allowing the interviews to be carried out by different field workers under varying circumstances. The questionnaires contain questions, lists of words and sentences, and various instructions to the field workers as to how to elicitate the subject's responses. In the German Sprachatlas, for example, or German Language Atlas, questionnaires constituted one method to elicitate particular keywords in order to find out regional variation. Let's look at telephone surveys next. Telephone surveys have been used with success at several points in the history of dialectology. They are especially useful if large geographical areas are to be explored. Today, we have modern versions of telephone surveys that are used on the web. For example, web conferences or peer-to-peer interview technologies such as voice over IP software, for example, most of you will know Skype. In both cases, that is questionnaires and telephone surveys, the method of data elicitation can be distinguished along various lines. Let's look at them in detail. Now, the main goal of dialectal fieldwork is to trigger a set of phonologically relevant keywords in the most natural way. This can be achieved using the following methods. First of all, there is the indirect method. Now, indirect questions are often felt to encourage informants to give more natural responses. Using this method, field workers would ask, for example, what is this? And now I would certainly expect the answer. This is a cup or mug trying to trigger the low central vowel a. Or I could ask the subjects, please count from 1 to 10 to trigger the pronunciation of the numbers in a most natural environment. What is the opposite of good expecting bad to be the answer trying to find out variation of this vowel that is involved in bad? And so on. Most surveys have used such an indirect technique. A modern example is the Atlas of North American English where data elicitation was carried out with telephone interviews where keywords were triggered indirectly via questions or via talking about particular topics such as housing. The next method is the direct method. The direct use of questions in questionnaires involves lists about particular items resulting in questions such as how would you pronounce cup or how do you say to? And so on. The German flag that you can see here indicates that this method was relatively popular in the original versions of the Sprach Atlas where the direct method was used to ask for regional variants of particular words. The disadvantage of this method is that the interviewer's pronunciation might have already biased the subject. So when I pronounce the word to first, then it influences the person who has to repeat it. The formal method is the next one. Using the formal method, all questions have to be presented in advance. Thus the field workers are bound to these questions and have little or no degree of freedom to alter the questions. This approach dominated the British dialect surveys indicated by the Union Jack here. And finally we have the informal approach. Now this method allows the field workers to freely frame their questions as long as they elicit the desired response. In American dialectology for example, the informal approach has dominated work. In practice however, most field workers discovered that a combination of these methods is best. Let us now illustrate the combination of these methods. In an interview with David Hornsby from the University of Kent, we will try to trigger as many representative keywords of his variety of English as possible. My name is Ritfa and I'm a student in the VLC seminar Varieties of English and I will now illustrate the indirect method using David Hornsby and some selected questions. Hello, I'm David Hornsby. I work at the University of Kent in the southeast of England but I was brought up in Essex also in the southeast of England. So I've spent most of my life in what's known as the Estuary English area of the southeast near to London. Okay, so David, here are my questions. While I'm asking these questions, I will record you using Audacity, the sound editor that is freely available on the web. First question. Eyes, nose and mouth are located in your face. What do you call the cost of something in pounds? Price. Where is your tongue located at? In your mouth. And all soccer players want to? Score. Score. Goal. A goal. Okay, that's it. Now, Jürgen Hunker will explain how this small data set and further data will be processed in order to create an entry in the VLC language index. Having collected the data, numerous additional actions have to be carried out. They depend on how the data is going to be presented. On the Virtual Linguistics Campus, the data must be compatible with our multimedia guidelines. First, we need some general information about the speaker. Now, here's the picture that our graphics team designed with our speaker, David, in the center. In addition to the picture information, we need information about David's place of residence, which will eventually be represented as a pin on a zoomable map. We need to inform the users of this entry about his ethnic background to be able to judge external influences on his personal variety. And we need personal data such as affiliation and age for further search options. And, of course, we have to specify the recording date. Last but not least, we need copyright information and, if available, literature references plus the permission of the speaker to use the data for scientific purposes. Now, the audio data has to be refined in several ways before it can become part of our language entry. First of all, we have to apply principles of general editing. For example, we have to set the sample rate, the bit depth and the number of channels. And we have to isolate our keywords first and the key passages. And we have to cut out unwanted stuff, unwanted material, as you can see over here. So materials such as coughs or repetitions, etc. And then we have to normalize the data. That is, we have to generate a standard volume level. And, very importantly, we have to find the starting or cue points for each keyword indicated by these red lines here. Now, these cue points are necessary to trigger particular portions of the overall sound file during the representation of sounds. With this brushed up information and these now well formed sound files, we can now perform various analyses. For example, we can submit our files to further acoustic analysis such as the generation of spectrograms. Now, here you see the spectrograms of the items that Ritva and David recorded earlier on. And, quite clearly, you can see that the fricative portion in the word mouth is different from that in the words price and face. In fact, David produced a dental fricative here. And I will say something about this later on. So with this spectrographic information, we can look at the frequency patterns involved in the vowels or we can identify selected consonants. The most common approach towards the analysis of sound data, however, is an impressionistic one. Many professional linguists have undergone intensive ear training enabling them to exactly identify the quality of particular speech sounds and convert their findings into a detailed phonetic code referred to as narrow phonetic transcription. As far as vowels are concerned, they can be placed on the cardinal vowel chart that you can see over here. Now, here is David's monothongal system as an example. Now, this is already part of the final VLC language index entry. Let's concentrate on two vowels. Now, David's data contains two keywords which were triggered using the indirect method. The first one is this C, C. And as you can hear, this monothong, this vowel sound which is normally classified as a high front monothong, is really some sort of diphthong in character, C. And we represent it down here by means of this sort of narrow transcription. Similarly, his O is not a clear cut high back vowel. Two. Rather, it involves some sort of off glide from a relatively central position of the O. Two. Two. To allow us to identify this speech sound also as some sort of diphthong. Now, in looking at these vowel sounds, so this is what we had. We have an O and we have an E. Here represented phonemically on the chart, but really, as we heard, they should be represented like this here. And the position is the sort of position phoneticians would identify in terms of the placement of vowels on the cardinal vowel chart. Well, and the keyword that Ritva recorded earlier on, well, here they are in a clean sound file. Let's listen. Face. Price. Mouth. Again, face. Price. Mouth. So, well, what we heard is something like this. The onset of face is relatively low in David's speech and narrowly we can transcribe it like this. The onset of price is, well, here I should use narrow transcription. The onset of eye is clearly back. Price. And the diphthong in mouth is well relatively front. And the glide, in fact, is not a glide that leads us towards a high back vowel, but somewhere in the mid-high region. So let's transcribe it with an O symbol, mouth. Well, and the final consonant in mouth is really a labiodental fricative in David's speech. Now this reminds us of his former Cockney roots. David lived in London for some time where Cockney speakers would regularly replace final dental fricatives with labiodental fricatives, so mouth instead of mouth. So we triggered the relevant data and could show that David's diphthongal onsets have moved. A typical feature of a speaker of estuary English. That is, the English that can, among others, be found in Kent. This can now be represented in the entry in the virtual linguistics campus language index. And if we use multimedia techniques, we can even make the system of David's diphthong interactive and show the glides as animated transitions. So let's take a look. Here are David's diphthongs. And now you see, here are the three keywords. Now we have different keywords here. Now, bye. Where's the last one? Say. Okay, so these are the three diphthongs. There are further diphthongs. There. Near. And so on. Now the combination of the indirect and the direct method, which we also used, now constitutes the backbone of the content of each entry in the VLC language index. Let us illustrate this with two examples from David's language index entry, the monophongs we've already seen. Now here is his monophongo system. Put. Co. Be. Be. And here is the morphology and syntax information, the very big book. The man sees the woman. And so on and so forth. Last but not least, the sound files are now freely available on the web. So in the language index, we have the download option where we can now find David's entry. So here we are. And here you find all the sound files ready to be downloaded and freely available for further linguistic research. However, you have to be a member of the VLC community. So join us on the virtual linguistics campus, create your free account and make use of the VLC language index with currently more than 1000 speakers.