How to use the SynAlm database

The results of our investigations are represented in so-called datasheets. These give a compact overview about:

  • the tested examples (with a literal English translation)
  • the number of informants for this specific question
  • the number of acceptance/rejection, divided according to the regions investigated. These numerical results are represented as actual numbers, percentage and a diagram, visualizing the numbers
  • a link to an interactive map where the areal distribution can be analyzed
  • the tagged and properly glossed example (in progress)

A datasheet contains usually several examples (a block) since in most cases, SynAlm asked for contrastive judgments. This means the sentences were systematically varied suitable for the concrete phenomenon. E.g., a relative clause was presented with only a particle, then with a relative pronoun, then with a relative pronoun plus the particle, the particle with/without resumptive pronoun and so on. The rest of the sentence was kept constant. All these sentences build one block and appear together on the datasheet - but for each single sentence, the above mentioned attributes are present.

The datasheets are accessible in two ways.

  1. There is a regular listing which presents the results ordered according to the single questionnaires and the original ordering in them. The questionnaires can be accessed here. (only in German, resp. the various Alemannic variants)
    In addition to the original questionnaire numbers, there are B-numbers. This is an additional numbering, singling out phenomena that were categorized in translation tasks and which deal with a different type of phenomenon than the translation task originally aimed at. (E.g. a task aimed at the left periphery of a relative clause – but in addition, interesting results showed up for the different forms and position of the participle used in this clause – so these are then categorized under a different number and grouped together under a different construction type, see below.)
  2. via the interactive Search. Here, a selection filters out all examples and results that were grouped according to the levels of description explained below.

There are five aspects with which one can narrow down the search:

Construction type: This is the label for the linguistic issue, e.g. Relative clauses, Possessives, W-Extraction, Pronouns, etc. Listing all construction types gives an overview about all the issues that are dealt with in SynAlm. A mouse-over gives more information why and/or under which aspect this construction type was chosen for investigation.

Phenomenon: This gives a finer grained and more specific characterization of the various aspects of the construction types. E.g., there is a construction type W-Extraction. SynAlm has results for the extraction of the subject, the direct object, various types of adjuncts etc. These are specified on the phenomena level: for all the examples that test for subject extraction, the ‘phenomenon’ will be: w-extraction(subject).
The labels for the phenomena were often built in an ad-hoc way – we tried to be as theory neutral and transparent as possible. Again, a mouse over gives you more information (still under construction!).

Description level: the linguistic description level of the example in question: syntax, semantics, morphology, phonology, and lexicon

Task type (judgment task, translation etc.): the task type used for the example. For more information about the types and how to interpret the scales read here.

Tag: Examples can also be searched by using a selection of tags. These can be used either in addition to a selection on the above mentioned levels or as the sole way to search the database. (list of tags used in SynAlm)

The selections made in the search are, by default, combined with an 'AND' operation and will therefore narrow down the remaining options and results.
The switch, however, allows to combine the selections of the same aspect (e.g. construction type) with an 'OR' operation which will actually broaden the results. Choosing 'INFINITIVES' and 'WORD-ORDER' for example with 'AND' results in all entries containing BOTH types while 'OR' delivers the entries where ANY of them is present. The type of operation among the aspects on the other hand is always 'AND'

Every datasheet is marked-up with at least one value for each of these labels. Some examples have more than one value (e.g. prepositional dative is grouped under CASE but since the example occurs in a dative-possessor construction, it is additionally grouped under POSSESSIVES). The same holds for phenomena and sometimes even for description level (e.g. morph and phon). So one single datasheet can be accessed by various paths.
By choosing ‘regular’, the datasheets with the compact information about the results will appear. ‘Data’ lists the original examples as they were given by our informants. (note: these may be very long lists of single sentences – up to 900!)
For information about the representation of the results, read here.

How to read the results in SynAlm

SynAlm uses several task types; the following list gives an overview about the standardized answers given for the various task types in the result tables and also the maps:

numbers (1-6): the numbers are listed according to the different judgment tasks, whereby 1 is always the highest (best) rating. The 'x' in these tables means that the informant did not evaluate this sentence.

no data: occurs in translation and own versions (ov) and means that the informant did not translate the sentence or did not give an ov.

not collected: some questionnaires were split up for different countries (e.g. Switzerland vs. Germany) and thus there is no data available. These are sentences/constructions that were collected already in other dialect syntax projects. Eventually, the SynAlm data will be merged with these results.

categorization values: SynAlm does not give the whole list of original sentences (the way they were written down by the informants) as the result of a query. Instead, the given sentences are already grouped, abstracting away first from the (often rather unconventional) orthography and lexical variation - if it is not relevant for the syntactic construction under examination. In a second step, the different syntactic constructions are given a categorization value and it is these values that occur in the result table – together with the numerical and areal distribution of the single values.
Example: We wanted to find out about the usage of articles with mass nouns and collected translations for a clause like 'this looks like gold'. In the result table, the results are represented like the following:

indef indefinite determiner used (= a gold)
def definite determiner used (= the gold)
zero no determiner (= gold)
If necessary, the specific meaning of the categorization values is described (and illustrated with an example) in the 'specification slot', directly above the result table.

For all categorization values, one tagged and properly glossed example – illustrating the construction - will be made available.

n.a.: occurs in categorization columns and indicates that the sentence given by the informant does not entail the necessary input for categorization according to the specific research question, due to e.g. different lexical items or a different syntactic construction chosen by the informant.

Task types used in SynAlm

Alemannic is a dialect of which certain properties are already discussed on a broad basis in the literature. Most notably infinitival syntax (verb raising, different types of infinitival introducers, etc. ), relative clauses, and possessive constructions. Therefore, it was possible for us to start already with (nearly) exhaustive alternative sets of test sentences since the base-line – so to speak - is already established.
Because of this situation, SynAlm explicitly heads for fine grained judgments (scale 1-5) of clauses that stem from the above mentioned areas. Nevertheless, translation tasks and fill-in tasks are also used. Especially the translation tasks often yield unexpected results which open up new areas of investigation. Below, there is a list of the tasks used in SynAlm with a few comments.

[task-type] := 'judgbin'|'judg3|'judg5'|judg6'|'trans'|'ov'|'pref'|'choice|'insert'|'comment'

Judgment tasks

judgbin: the informant is asked whether a given construction is possible in his/her dialect according to his personal judgment. The answer is simply Yes/No. In this type of question, the informant has the possibility to give his own version (ov). If appropriate, these own versions are categorized according to the relevant construction types, respectively new construction types are defined.

judg3: this task heads directly for the competence/knowledge of the informant about his/her dialect, whereby the following is given to the informant as a guideline:

1= I know this construction and use it on my own on a regular basis
2= I know this construction but I do not use it (sounds e.g. old-fashioned)
3= I have never heard this construction/this is completely unknown to me
The idea behind it is that the informant abstracts away from his every-day usage but rather reports on the dialect and his knowledge about it.

judg5: This is the usual 5-point scale used in linguistic judgments which can easily be transformed into the three judgments ok - ? - and *.

judg6: This is the school rating (1= best – 6 = worst) used in Germany. In former studies this was used very successfully since the informants have a good intuition about this scale and the rating is basically conventionalized. Unfortunately, this can't be used in the comparative approach in SynAlm since Switzerland has a different scale and thus, the results are not comparable.

Tasks providing full sentences

trans: These are translation tasks of sentences we constructed aiming at certain constructions. In FB1, 3 sentences of Wenker were to be translated. The goal here is to compare them eventually with the original Wenker translations in order to see to which extent the dialects are stable. The Wenker-translations stem from about 1870 and at least for certain morphological properties, we can already prove that the dialect is very stable.
Translation tasks are categorized according to the construction types and if appropriate, new construction types are added.

ov (own version): Under nearly all judgment tasks, the informant has the possibility to write down his/her own version of the construction asked for in the judg-tasks. This possibility is used if the informant feels that none of the offered versions is correct in his dialect. Quite often, these 'own versions' offer different lexical items; however sometimes they reveal constructions that have been overlooked until now.

pref: The informant has to choose from the offered versions the one which is the most natural for him/her.

choice: Several lexical items (mostly function words) are offered and the informant chooses which one is for him/her the most appropriate in the given context/sentence.

insert: This is a combination of choice and translation as the informant has to translate a given function word from Standard German into his dialect and then insert it into a clause/construction.

comment: this is not a real task – under some (larger) tasks we gave space to write down all types of comments/other versions/general remarks etc.