Publishing Partner: Cambridge University Press CUP Extra Publisher Login

New from Cambridge University Press!


Revitalizing Endangered Languages

Edited by Justyna Olko & Julia Sallabank

Revitalizing Endangered Languages "This guidebook provides ideas and strategies, as well as some background, to help with the effective revitalization of endangered languages. It covers a broad scope of themes including effective planning, benefits, wellbeing, economic aspects, attitudes and ideologies."

New from Wiley!


We Have a New Site!

With the help of your donations we have been making good progress on designing and launching our new website! Check it out at!
***We are still in our beta stages for the new site--if you have any feedback, be sure to let us know at***

Review of  German Pronunciation & Accents

Reviewer: Franz Dotter
Book Title: German Pronunciation & Accents
Book Author: Luciano Canepari
Publisher: Lincom GmbH
Linguistic Field(s): Phonetics
Subject Language(s): German
Issue Number: 28.5087

Discuss this Review
Help on Posting
REVIEWS EDITOR: Helen Aristar-Dry


“German Pronunciation and Accents,” by Luciano Canepari follows the outlines which the author has formulated in his ''Natural Phonetics & Tonetics Method'' (for an overview and several special files cf., applied to German. The first six chapters contain an introduction to the method and its application to ''international German'': the ''Foreword'' introduces Canepari's terminology of ''international'', ''neutral'', ''traditional'' and ''mediatic pronunciation'', related to German:

''However, the often capricious behavior of purely native accents has led us to the 'wise' decision to primarily present an 'international' German pronunciation …, in a 'native-like form'. This has the practical advantage of providing an easier access to the 'spirit' governing German phonemics, precisely because the international phonetic realizations are more straightforwardly - yet realistically - connected to the phonemic level than most national or regional accents are. … this 'international pronunciation' has proven to be a good teaching and learning device. In fact, it is still 'German German', but with some simplifications.'' (p. 10)

''Neutral'' is used instead of ''standard'', ''an impeccably articulated pronunciation which gives no clue about the speaker's geographical and/or social origin'' (p. 9). Canepari gives the following reason: ''because we are convinced that the term 'standard' stands more for something which is common and … 'good enough', rather than excellent and really recommendable.'' (p. 9). Following the author, ''traditional pronunciation … is less recommendable, by now, unlike 50 or 30 years ago'' (p. 9).

''Mediatic German pronunciation, in itself, is not a regional kind, rather, something which is not completely neutral, although some peculiarities can coincide with actual regional accents. This pronunciation can be heard on radio and TV, by people who could not achieve neutral pronunciation, starting from localizable personal situations, which have been attenuated (trying to avoid revealing their origin) but with no tangible result. It is something acquired more by practical assimilation than through any learning method.'' (p.135)

Having started in Ch. 1, Canepari goes on to discuss his view of phonetics in Ch. 2 (''Pronunciation & phonetics''). ''The phono-articulatory apparatus'' (Ch. 3) presents some articulatory descriptions. It is followed by ''The classification of sounds'' (Ch. 4), specifying into ''Vowels and vocoids'' (Ch. 5) and ''Consonants and contoids'' (Ch. 6).

The presentation of German turns in Ch. 7 towards ''The consonants of international German'', followed by ''Structures'' (Ch. 8: ''Taxophonics'' and ''Stress''), ''Reduced Forms'' (Ch. 9) and ''Intonation'' (Ch. 10).

Chapter 11 offers ''Some texts in phonotonetic transcription'' (where - under others - the famous ''Nordwind und Sonne''/''North wind and sun'' appears in 8 variants).

Ch. 12 describes ''Neutral German pronunciation'', Ch. 13 ''Traditional German pronunciation''. Concerning the vowels of these two postulated variants, Canepari assumes only a difference in [ɛ:] (maintained in the ''traditional'' variant) and in the colouring of diphthongs; for consonants there is a strong ''traditional'' resistance against assimilation (p. 132). Ch. 14 presents ''Mediatic German pronunciation'' with the following restriction: ''Obviously, here we present a slightly idealized normalization. This is necessary, because (as it happens with all mediatic pronunciation, for any language) oscillation between mediatic and neutral characteristics is quite normal'' (p. 135).

Chapters 15 to 18 then deal with the variants in ''North-east Germany'' (the area of the former German Democratic Republic), Austria, Switzerland and South Tyrol/Alto Adige. Different from the variants quoted so far - which are subsumed under ''Native accents'', Chapters 19-21 give ''Regional accents'' for Germany (Ch. 19; offering 19 variants), Austria (Ch. 20; offering 9 variants) and Switzerland (Ch. 21; 7 variants, including Italian, French and Romansch areas).

In the ''Appendices'' section, we find Liechtenstein, Alsace, Lorraine, Luxemburg and East Wallonia as ''German speaking areas in other countries'' (Ch. 22), ''Some dialects'' (Ch. 23, 15 variants) and ''Some foreign accents'' [of German] (Ch. 24, 17 variants), ''Some diachronic stages'' [of German] (Ch. 25, containing Proto-Germanic up to Middle High German).

Chapter 26 provides a ''Mini-phono-dictionary with about 2300 words in Canepari's International German version; Ch. 27 provides ''Phonopses of 26 languages'' and ''principal consonant orograms''. Ch. 28 deals with ''English pronunciation by German speakers''.
Ch. 29 offers an ''Annotated bibliography''. The last page of the book shows the IPA chart.


Canepari is the author of 7 books of the same type as the one discussed here. For a first 2008 review on Canepari's method cf. ; cf. also . I regret, but I have to sharpen the critical points found in this review.

1. Canepari's model and claims

Let us first look at Canepari's model: he criticizes the IPA as being more phonemic than phonetic. A real phonetic alphabet should follow the ''ideal of «one sound, one symbol»'' (; FAQ 10) and puts his ''can-IPA'' instead:

''Natural Phonetics aims at achieving the essence of linguistic sounds:
- by determining their exact articulations through the awareness of what our mouths do to produce them;
- by showing them with accurate articulatory diagrams (i.e. orograms, vocograms, labiograms, palatograms, dorsograms), and auditory diagrams (mainly tonograms);
- by assigning them appropriate phonetic symbols (which must not be too general with the risk of being useless).
What I mean is that even today there is a terrible need for good phonotonetic descriptions, based on a rich set of symbols, as is my own canIPA inventory, which comprises 52 basic vocoid symbols and several hundreds of contoids. If we then consider that there are 8 further potential vocoids, and that taking into account intermediate lip-positions we could get at least as many as 26 more, and as many nasalized vocoids as there are oral, we could reckon with about 1000 linguistic sounds altogether, which are all to be found in my Handbook of Phonetics (HPh), each with its own symbol (generally with no «second-class» diacritics): 500 basic, 300 complementary, and 200 supplementary ones.''

Nobody would contradict Canepari when he states that we should hear to the languages and not do phonetics via written language. But his essay titled ''Writing systems: the utmost monstrosity of alphabets and 'orthographies' ( shows his illusions concerning his model.

The quotation also uncovers the author's illusion of the benefit of a very narrow phonetic transcription. This comprises not only the illusion that he can discriminate the 1000 sounds he defines articulatorily, but also that teachers and learners can profit from these transcriptions. If we took Canepari's ideal, namely ''one sound, one symbol'' literally, we would need to create a symbol of its own for every realization of any sound because no sound equals any other completely. I suppose that not even Canepari would like to be understood in this sense. Therefore he would have to concede his need to form classes. He does this by taking each of his assumed articulatory values implicitly as a class and by presenting the classes by his idealized ''vocogram'', ''orogram'', ''linguogram'', ''palatogram'' and ''labiogram'' drawings. Let's compare that to his postulated method:

''To be of any practical use in the learning/teaching of pronunciation, but also as a plausible basis for any successive abstract speculation/theorizing, phonetics should be ‹natural›, in the sense that it should be possible to do it without the help of any instruments other than one’s own articulatory apparatus and ears.'' (

By this practical contradiction of working ''naturally'' and only implicitly assuming classes, Canepari ignores the fact that the variation of vowels is not restricted in a way that would keep the realizations of different phonemes away from each other. He also ignores the fact that the realization of sounds is influenced by coarticulation: in a very narrow transcription he would have to differentiate between vowels framed by different classes of consonants concerning their articulation place, etc. Canepari wants to be seen as working purely phonetically but often states a fixed relation between one of the many sounds he transcribes and the respective phoneme (without describing the phonemic system of the respective variant).

There is a basic contradiction between the announced exact transcription of the sounds detected and the apparent normative construction of ''international German'' (cf. above) which is shown by Canepari's FAQ 21:

''Do you approve of normative phonetics?
I do, however old-fashioned this might make me appear! To me, it’s a way of showing respect to an aspect of a language which is secondary to none. Exactly as people strive hard to use correct grammar and proper vocabulary, why shouldn’t they try to pronounce their language correctly?''

Therefore, as we do not have any audio data, we cannot be sure to which extent Canepari applies a normative or idealizing (normalizing) perspective to all other variants of German, including the dialects, when he transcribes them.

A chart of the ''vocoids'' can be found at p. 48; for the ''contoids'' the reader has to consult files NPT 09 and NPT 10 (cf. table at pp. 166-168) at Canepari's homepage.

Canepari's FAQ 14 reads as follows:

''14. But can you really distinguish between all those sounds?
Yes! Many people are able to distinguish a great deal of different car models, including the year they have been made. Of course, they must be particularly interested in cars. When you’re very much interested in sounds, you know that you’re able to feel and appreciate their differences. Try it and see… Obviously, some particular or newer sounds can be more difficult to grasp, at first: a good tape recorder is very important.''

2. Missing data and information

However, Canepari does not inform the reader about acoustic text sources where the reader could hear examples of the sound transcribed or check the correctness of a transcription of a certain text. He also does not tell us how he collected the data. This is a crucial violation of scientific rules, as without examples no transcription can be verified.

From a functional point of view, Canepari does not offer any systematic distributional analysis of the inventories of the variants of German; there are no contrastive diagrams which would allow the reader to compare diverse variants on the spot. E.g. open [ɔ] appears in ''international German'' for short /o/ while it represents parts of /a/ in Southern dialects. This remains uncommented.

3. Narrow transcription

One of the central claims of Canepari's approach is: he wants to be ''as precise as possible'' in transcription. This reminds me of a phase in Austrian German dialectology, linked to the name of Eberhard Kranzmayer: there the linguists thought they could systematically differentiate three or four degrees of opening for German dialect vowels. But soon it turned out that variation over generations, regions, persons and contexts (e.g. over different stress of words/syllables in the sentence or different sound context) was bigger than what they believed to be a stable articulation for a certain functional sound, to be transcribed by a certain symbol. What we can learn in general from that example: the narrower the transcription, the more idiosyncratic its outcome will be.

Moreover, experience in phonetic fieldwork shows that the single transcriber's practice sometimes causes shifts in the evaluation of certain sounds within the course of a long transcription. Without intense checking of results with native speakers, counter-checking with previous own transcriptions and transcriptions of colleagues it is even hard to use a rather limited phonetic inventory consistently. Here I do not even speak of perception tests or technical aids.

Additionally, we may ask what the function of a very narrow transcription may be. Naturally, done previously to the assumption of a phonemic model, it makes us aware of variation related to linguistic or other criteria and by that allows us to create an adequate phonological (= functional) model of the respective language. Therefore narrow transcription, accompanied by technical and perceptual information, delivers important facts about intricate phenomena of spoken language.

But we must be aware not to be trapped in the illusion of transcription as a complete sound mapping of any language: transcription aims predominantly at reflecting the quality of sounds, related to their main features of articulatory production (e.g. articulation manner and place for consonants, tongue height, position, and rounding for vowels) or acoustic description (main characteristics; e.g. type of signal, formants for vowels).

4. Narrow transcription as a partially normative or idealizing means for the construction of language variants

But the situation with Canepari's work is even worse than to be trapped in this illusion: the author uses narrow transcription in order to construct systematic differences between the variants postulated: he wants to prove fine-grained differences between variants of languages - in this book: German - by way of systematically differentiating transcriptions without documenting the respective data. Naturally, there is something which has been described as ''Basis of Articulation'' and which allows us to identify the regional or national linguistic background of a speaker. But this metaphor summarizes a lot of factors which contribute to such an identification: besides the specific articulation of single sounds we find suprasegmental factors like forming resonance space throughout articulation, timing, stress, etc. It cannot be shown by transcription as this is mainly oriented to the quality of vowels and the articulation manner and place of consonants and only a little bit to duration and other factors. To give an example: the duration ratio between the accented vowel and the following consonant is very different over the German area though it preserves the main functional difference between the ''long'' and ''short'' vowels.

Canepari uses apparently idiosyncratic transcription strategies in order to construct differences between the 8 variants of ''Nordwind und Sonne'' (''North wind and the sun'') p. 104-106). I can only give some examples here; the numbers in the columns denote the following variants: 1=German word, 2=international pronunciation, 3=neutral pronunc., 4=traditional pronunc., 5=mediatic pronunc., 6=ex-DDR pronunc., 7=mediatic Austrian

1 2 3 4 5 6 7
der; dʌ; d[ʁ+]; dəʁ; dɛɜ; d[ʁ+]; dɐ
des; dəs; [d+]s; [d+]s; [flap]ɘs; [flap]ɜs; dəs
den; dn; dn; [d+]n; [flap]ɘ[n+]; dn; dn
ihnen; ʔinm̩; ʔi[m̩+]; ʔin[ɨ+]n; ʔinɘm; ʔi[m̩+]; i[m̩+]
von; fən; f[ɨ+]n; f[ɨ+]n; fɔn; fɜn; fən
sollte; zɔltə; z[ɔ+]l[t+][ɨ+]; zɔlth[ɨ+]; zɔltɘ; z[ɔ+]ltɜ; s[ɔ+]ltɘ

Note: forms like [d+] represent characters used by Canepari which cannot be given in IPA.

An inspection of the transcriptions of ''North wind and the sun'' shows that Canepari uses a more or less rule-driven transcription for every variant of German he defined: e.g. the realization of unaccented /e/ is systematically transcribed by [ɨ+] for ''traditional'' and ''neutral'' or with [ə] for ''international German'', etc.; as well, unaccented /-er/ is transcribed by [ʌ] for ''international'' or [ɐ] for Austrian; all cases irrespective to the real pronunciation. Respecting the latter, other factors - e.g. formality or position in a sentence - decide about the vocalized or consonantal realization of /r/ in final position; they are valid throughout many German variants. For this reason, foreigners are easily detectable when they try to systematically realize a consonantal version of this /r/. An omission of the vowel is clearly interpretable as regional (e.g. Alemannic). It is absolutely unusual in the articles [dɛs] or [den]. Also the different diphthong transcriptions for the regional variants of German (p. 53) seem to be more or less invented in order to construct systematic differences.

There is an overwhelming use of assimilation rules, contradictory to explicit reading or speaking, as we can see from the examples for ''ihnen'' in the table above (the word is followed by a bilabial consonant). Conversely, the glottal stop is inserted almost everywhere in all variants except the Austrian one.

5. Phonetically incorrect transcriptions

Compared to Canepari's pretensions, there are many systematic transcriptions which are not phonetically adequate, even if we take the perspective of a broader transcription: the realizations of unaccented /e/ in German supraregional variants, often represented by the neutral Schwa [ə], have to be characterized with an open quality, namely [ɜ]. Higher variants as e.g. in some Bavarian dialects show the quality of a reduced closed [e] while an i-colouring (which Canepari postulates for standard variants, cf. p. 136) cannot be found. An a-coloured Schwa [ɐ] represents the unaccented syllable /-er/ with vocalized /r/.

Standard German diphthongs are another case where Canepari follows some stereotyped transcriptions without respecting the quality: backward perception experiments show that the diphthongs {eu} and {au} are best transcribed by [ɔɛ] instead of [ɔy] - for Austrian German even [ɔø] - (the rounding disappears in the second part of the diphthong) and [ɑɔ] (both parts have open quality) instead of [ao].

The realization of unaccented /-en/ after /n/ (examples: ''seinen'', ''spannen'') by [nn̩] which is - normatively - set for ''international'' and ''neutral'' as well as even for ''mediatic Austrian'', is a clear regional variant of Northern and middle Germany.

6. Phonological and other points

Concerning phonology, Canepari claims that the vowel combinations in German words like ''nahe'' or ''Ruhe'' are ''real diphthongs'' (p. 53) which is clearly not the case, comparing them with the true diphthongs.

In his tour de force through German dialects, the author does not give a correct picture, but often uses single features for a specific dialect which are also characteristic of others, e.g. he localizes the vocalization of /l/ in Salzburg and Vienna only (pp. 230, 275), the unrounding of vowels only for South Tyrol. Concerning Bavarian (p. 270), he mentions ''vowel doubling before lenis'' and speaks of ''nasalized vowels and (peculiar) diphthongs''.

A historical faux-pas is Canepari's use of ''Danubia'' instead of ''Upper'' and ''Lower Austria'' including Vienna. He may not know that the terms ''Upper'' and ''Lower Danubia'' were invented by the Nazis in order to extinguish the name of Austria.

I do not comment on the assumed ''phonetics'' of Gothic or Old Saxon, as well as on the foreign accents of German and the ''Phonopsis'' the other languages (altogether 100 pages) as there is no information at all how the author got the respective audios.

7. Summative evaluation

Summarizing, Caneparis' book shows so many phonetic inadequacies, methodologically questionable shortcomings, basic contradictions in the model itself, or crooked statements (cf. the definition of variants and their evaluation) that it cannot be positively evaluated. Conversely, the readers have to be warned not to take the author's transcriptions as an adequate representation of today's German variants. Adding the sociolinguistic comments on diverse variants and the unclear relation of normative and descriptive factors, the book is a danger for linguists and students to be misled completely when trying to get information about German, phonetically and sociolinguistically.
Franz Dotter, born 1948 in Salzburg, Austria; Associate Professor for General Linguistics, retired; worked from 1973 to 2013 at the Institute for Linguistics of Klagenfurt University, then at its Centre for Sign Language and Deaf Communication. Main interests: Typology and cognitive linguistics, sociolinguistics of politics and minorities, sign languages, applied toponomastics.