The prosody of station and train announcements in German – sound examples

This post contains the sound examples to my paper on the prosody of station and train announcements in German .

Gilles, P. (2014). Zur Prosodie von Bahnhofsansagen. In K. Birkner, P. Bergmann, P. Gilles, H. Spiekermann, & T. Streck (Eds.), Sprache im Gebrauch: räumlich, zeitlich, interaktional: Festschrift für Peter Auer (pp. 95–108). Heidelberg: Universitätsverlag Winter.

(1) Weibliche Ansagerin, Hbf Ulm (ulm, sec. 11)

1 ↑↑<<f, all>mEine damen und> hErren ʔam:: gleis ʔEins hat <ʔEINfah:rt>–
2 ↑↑ʔI: ze<:> e: sechs ʔEinunneunzi[k] <nach berlin><OSTbahnho::f>–
3 über stUttgart MANNheim;
4 frAnkfurt-FLUGhafen–
5 ↑↑<Abfahrt> nEun uhr einundFÜNFzi[k];
6 ↑↑<bitte> vOrsicht bei <der ʔʔEINfahrt>;

(2) Männlicher Ansager, Hbf Freiburg (bahnhof 1, sec. 9)

1 <<f, all>mEine damen und herren> am gleis ZWO:?
2 ʔes fährt ʔʔEIN, (-)
3 <regio>nAlbahn von: (-) STAUfen,=
4 =ʔAnkunft neun uhr SECHS, (--)
5 <in dIesen> zug bitte nicht ʔEINsteigen-
6 dIeser zug ʔEndet hier ʔam gleis ZWO-=
7 =<bitte> vOrsicht bei der ʔEINfahrt;

(3) Weibliche Ansagerin, Hbf Freiburg (bahnhof 1, sec. 125)

1 meine damen und herren an gleis fÜnf fährt EIN?=
2 =intercItyexpress einhundertEINundachzi[ç]?=
3 =von FRANKfurt?=
4 =zur wEi[th]erfahrt nach ZÜ:rich? (-)
5 über sIngen schaffhAusen WIN[th]erthur;
6 Ankunft um neun uhr vierundZWANzi[k]?
7 wEiterfahrt um neun uhr sechsundZWANzi[k]- (-)
8 die wAgen der ers[tɛn] klasse halten im abschnitt BE:, (-)
9 °h die wA[gəәn] der zwO[thəә]n klasse im abschnitt bE: bis
DE:; (-)
10 nÄchster halt SINgen, (--)
11 am gleis fÜnf bitte vOrsicht bei der EINfahrt;


↑↑<<f, all>meine damen und> hErren ʔam:: gleis ʔEins hat <ʔEINfah:rt>–


über WÖRlen-
ALLmendingen; (ulm, sec. 5)


a. über sIngen schaffhAusen WIN[th]erthur; (fr01, sec. 133)
b. über bad krOzingen STAUfen; (fr01, sec. 60)
c. über dEnzlingen Emmendinge LAHR, (fr02, sec. 35)
d. über dEnzlingen WALDkirch, (fr02, sec. 284; fr02, sec. 882)


a. über kArlsruhe MANNheim-
frAnkfurt hanNOver, (fr01, sec. 101, fr02, sec. 57)
b. über stUttgart MANNheim;
frAnkfurt-FLUGhafen– (fr02, sec. 68)
c. über kArlsruhe mAnnheim FRANKfurt- (.)
gÖttingen hanNOver, (fr02, sec. 161)
d. über DENZlingen-
ʔEmmendinge LAHR, (fr02, sec. 278)
e. über bad krOzingen MÜLLheim-
bAsel BAdischer bahnhof, (fr02, sec. 709; fr02, sec. 892)


über kArlsruhe MANNheimfrAnkfurt-
köln WUPpertal- (fr02, sec. 315; fr02, sec. 353; fr02, sec. 411)

(9) Weibliche Ansagerin, Hbf Freiburg (fr02, sec. 519-528)

1 meine dAmen und herren am gleis vIer fährt EIN,
2 rIngzug von Immendingen nach ROTTweil-=
3 =Ankunft nEun uhr siebensiebenundVIERzig?=
4 =wEiterfahrt zehn Uhr bitte vOrsicht bei der bei der Einfahrt am glEis VIER.


a. ... regionAlbahn von: (-) STAUfen (fr1, sec. 11)
b. ... wird heute voraussichtlich (--) ZEHN minuten später eintreffen (fr01, sec. 253)
c. ... (1.4) wird heute voraussichtlich (--) fÜnf (.) bis ZEHN minuten später eintreffen (fr2, sec 582)

Mapping name frequencies

I always wanted to make this map.

Surname frequencies in Germany

While it is evident from an onomastic point of view that surnames offer various patterns of regional distribution, e.g., due to migration, linguistic structure, genealogical reasons, it seems less clear – at least to me – whether there exist also regional patters for the amount of different surnames. Are there regions with a larger or smaller inventory of surnames compared to others? It seems evident of course that the melting pots of big cities would show larger inventories due to their history of migration. But is it always true that the more people live at a specific location, the more different surnames we get?

In order to get some grip on this question for the German surname landscape, I calculated first the absolute number of different surnames for all three-digit postal code areas in Germany. The data basis consists of the 23.526.460 registered people in the German telephone book of 2009. The count for different surnames in a certain postal code area then is correlated with the total number of persons registered for this area. In order to account for the varying people count of these regions – a postal code area of Berlin certainly has higher counts than a more rural one – it was decided to calculate the number of different surnames per 1000 persons.

The resulting map (made with QGIS) firstly shows that German regions indeed differ with respect to the number of different surnames. The mean is approximately 270 different names per 1000 persons and we can clearly identify regions containing more respectively less names. Unsurprisingly, it is confirmed that big cities and urban agglomerates are characterised by having more different surnames. Note the red areas on the map, indicating heavily urbanised regions like, e.g. Munich, Stuttgart, Frankfurt, the Ruhrgebiet, Hamburg, Berlin etc. A high degree of urbanity also correlates with more diverse surname inventories. Nevertheless, we also find more rural areas, e.g. in the northeast, also exhibiting a large inventory of surnames and the reasons for this remain to be investigated. Light blue areas, on the other hand, indicate areas with a very low surname diversity and again it remains to be investigated why certain areas in Bavaria, Brandenburg or in the Eifel seem to have developed less surnames than other regions. It would be interesting correlating these figures with some kind of ‘rurality index’. 


Using CartoDb’s torque option to display lexical diffusion on Twitter

Twitter offers wonderful data for linguists. As a tweet contains information about time and space, the diffusion of a certain word can be traced in the mediascape. The following example shows how a newly created word spreads out in time and space on Twitter within minutes.

In a TV show on the 1st of September 2013 featuring the last federal elections in Germany, the necklace of Kanzerlin Merkel was heavily commented on Twitter as it seemed to contain the colors of the German flag but assembled in an unusual order allowing allusions to the Belgian flag. The necklace quickly was termed ‘Schlandkette’ (‘Germany necklace’), containing a neologism coined in the realm of the last football World Cup when ‘Deutschland’ was shortened to ‘Schland‘. This humorous idea was taken up immediately in the Twitter community through commenting, modifying and retweeting.

As a neologism ‘Schlandkette’ rapidly made its way into 4000 tweets within two hours. Displaying them in the following chart shows how the word came up around 8:47 p.m. and reached a frequency peak within the next ten minutes.


Why not combining the temporal distribution of the spread for this word with the spatial dimension? Unfortunately, only a very small portion of tweets are geocoded (this is especially true for Twitter users in Germany), the attribution of a given tweet, i.e. Twitter user, to a geographic location is more complicated. As a workaround I extracted the geographical location users entered in their Twitter profile and assigned the values for latitude and longitude. Of course, this couldn’t be done when a user entered fictitious locations like ‘Middle Earth’, ‘On the Internet’, ‘daheim’ and the like. Nevertheless, a more or less reliable geocoding could be achieved for most of the data.

Having now the necessary information on time and space I used CartoDB‘s amazing Torque rendering technique, which creates animated maps displaying time-dependent data. The following map on CartoDB nicely shows how ‘Schlandkette’ originates in some locations in the west of Germany and then very very rapidly spreads all over Germany before then also being taken up in neighboring countries.

Bildschirmfoto 2014-01-18 um 10.52.27Click to enlarge and watch the animation!