Feature 6. Mirror-image effect

found in question(s): 16a, 16b, 16c, 17a, 17b, 17c

Definition and illustration

Mirror-image effects refer to the observation that if two categories are in a fixed order preverbally, we find the reverse order postverbally (part of Greenberg's U20-effects, see Cinque 2009, Abels & Neeleman 2012, overview in Pregla 2024).

Assuming that the scope of modifiers matches the order in which they combine with the head, e.g., temporal adverbials being combined later with the verb than manner adverbials, we can observe the following mirror-image effect in English: we obtain the neutral order Temporal-Manner preverbally, but Manner-Temporal postverbally.

(1)

In the project, Mirror-image effects were mainly used as a diagnostic for head displacement (also see questions 2, 3, 4, 11): The presence of a mirror image effect indicates that English does not feature movement of the lexical verb. Therefore any findings on the adjacency of V and O (features 2,3,4) can be taken at face value.

If, on the other hand, V-Temporal-Manner is a neutral order, verb movement has been diagnosed. As a result, surface VAdvO/VXO orders (features 2,3,4) can only be interpreted if verb movement is controlled for, as illustrated for German (Germanic, Europe) below:

(2)

The b-example shows the verb in its base-position (while the Aux undergoes V2-movement), and hence we obtain the order Temporal>Manner>V. In the a-example, the verb precedes both adverbials, V>Temporal>Manner, but this is due to the fact that the verb moves tot C. Such examples are thus not informative regarding adverbial order relative to the verb.

Assuming cross-categorial harmony we expect the direction in which dependents combine with the head to be consistent resulting in statistical implicational universals. Hence, in OV-languages, these adverbials are more likely to occur preverbally, while in VO-languages, they are more likely to be placed postverbally (English being an atypical VO-language in this respect).

Apart from serving as a diagnostic, mirror-image effects also directly test for the universality of U20-effects in the adverbial domain: the order Manner-Temporal-V is supposed to be universally ruled out as a neutral order (Cinque 1997):

(3)

Different focus contexts have been tested to differentiate clearly between neutral and marked orders in languages with more than one word order, and to determine the function of marked orders.

Correlations

Question 16

• Statistical Prediction: O-V ↝ *V-Loc-Temp
• Statistical Prediction: V-O ↝ *Loc-Temp-V
• Prediction: V-Temp-Loc ↔ V-movement
• Prediction: V-Temp-Loc ↔ neutral VAdvO/VXO

Question 17

• Absolute Universal: *Manner-Temp-V
• Statistical Prediction: O-V ↝ *V-Manner-Temp
• Statistical Prediction: V-O ↝ *Manner-Temp-V
• Prediction: V-Temp-Manner ↔ V-movement
• Prediction: V-Temp-Manner ↔ neutral VAdvO/VXO
• Prediction: O-V → *Manner-Temp (underlying structure)

Results

The data collection showed that the relative order of temporals and locatives is not as fixed preverbally as the relative order of temporals and manners. Therefore question 17 is more informative than question 16.

References