One chapter of my Master’s thesis focused on patterns of wh-extraction from the than-complement in English. The basic idea was that different patterns of wh-extraction permit an indirect analysis of the syntax of the than-complement in different comparative constructions. Thanks to all those who helped with grammaticality and acceptability judgements!

English is typologically unusual in permitting wh-extraction from the than-complement in comparative constructions, as in (1).

(1)        Who is John taller than?

(1) is a phrasal comparative, i.e. ‘than’ is underlyingly followed by a phrase (as opposed to a clause). Wh-extraction is only permitted from phrasal comparatives, not clausal comparatives, as in (2), the clausal counterpart of (1).

(2)        *Who is John taller than is?

Wh-extraction from phrasal comparatives exhibits at least two patterns.

The first is seen in nominal comparatives, as in (3).

(3)        a.         Who did John give more presents to Mary than?

            b.         *What did John give more presents to Mary than?

            c.         *To whom did John give more presents to Mary than?

Wh-extraction of phrasal than-complements which are associated with non-subjects is ungrammatical (or at least degraded), as in (3b) and (3c), whereas wh-extraction of phrasal than-complements associated with subjects is acceptable, as in (3a).

A sub-pattern within nominal comparatives is that, when the nominal modified by more is the (external argument) subject, no wh-extraction (even wh-extraction of subject-associated phrasal than-complements) is possible, as in (4).

(4)        *Who did more children give presents to Mary than?

I argued that the patterns in (3) and (4) follow if phrasal than-complements in nominal comparatives are typically small clausal than-complements.

The second pattern is seen in adverbial comparatives, as in (5).

(5)        a.         Who did John give presents to Mary more than?

            b.         What did John give presents to Mary more than?

            c.         To whom did John give presents to Mary more than?

Although the sentences in (5) sound a bit clunky, (5b) and (5c) are significantly better than (3b) and (3c). It seems that wh-extraction from the phrasal than-complement of adverbial comparatives is generally unrestricted.

An important sub-pattern concerns adverbial comparatives with gradable verbs such as like. Compare (6) and (7).

(6)        a.         John likes chocolate more than Bill.

            b.         Who does John like chocolate more than?

(7)        a.         John likes Mary more than Bill.

            b.         Who does John like Mary more than?

(6a) and (7a) are ambiguous, for example, in (6) Bill could be associated with either the subject John (i.e. John likes chocolate more than Bill does) or the object chocolate (i.e. John likes chocolate more than he likes Bill). (6b) is also ambiguous in the same way, i.e. who can be associated with the subject John or the object chocolate. However, (7b) is unambiguous – it can only have the interpretation where who is associated with the object Mary, i.e. John likes Mary more than he likes who? Descriptively, when the object is [+human], only the object-association interpretation is available for the wh-phrase.

This pattern is not found with non-gradable verbs such as visit, as in (8) and (9).

(8)        a.         John visits museums more than Bill.

            b.         Who does John visit museums more than?

(9)        a.         John visits Mary more than Bill.

            b.         Who does John visit Mary more than?

(8b) and (9b) are both ambiguous in the same way as (8a) and (9a).

So, the sub-pattern is that wh-extraction of phrasal than-complements is prohibited when the verb is gradable, the direct object is [+human] and the extracted phrase is associated with the subject. I argued that these patterns follow if phrasal than-complements in adverbial comparatives are genuine phrasal than-complements.

In summary, wh-extraction is prohibited from clausal than-complements, but is allowed from small clausal and genuine phrasal than-complements (though it exhibits different patterns concerning the available interpretations of the extracted element). Nominal comparatives may use clausal or small clausal than-complements, whilst adverbial comparatives may use clausal or genuine phrasal than-complements.

19,995 words, 116 examples (mainly English examples but also including French, German, Dutch, Swedish, Bulgarian, Polish, Serbo-Croatian, Hungarian, Malagasy, Turkish, Hindi-Urdu, Japanese, Mandarin, Romanian, Menomini and Latin), and hopefully some good ideas and arguments about the structure of the than-complement in comparative constructions and its consequences for syntactic and formal typological theory! :)

No matter if I’m frustrated with linguistics or enthused by it, I’m always asking myself this same question. What counts is how I look at it.

Would anyone mind helping with some more grammaticality judgement tests?

Basically, in the following are the question-answer pairs acceptable given the context?

(1)

Context: John visits Alice more than Sue does.

Question: Who does John visit Alice more than?

Answer: Sue.

(2)

Context: John visits Alice more than he visits Mary.

Question: Who does John visit Alice more than?

Answer: Mary.

(3)

Context: John visits museums more than he visits Mary.

Question: Who does John visit museums more than?

Answer: Mary.

(4)

Context: John visits Alice more than he visits museums.

Question: What does John visit Alice more than?

Answer: Museums.

Thank you! :)

Could you all help me with a grammaticality judgement test?

Context: John likes Mary more than Bill does.

Question: Who does John like Mary more than?

Answer: Bill.

Is the question and answer pair a legitimate one given the context? I just can’t tell anymore!

Thank you! :)

Definition:

(1) The sound your grammaticality judgements make as they fly out the window.

(2) The herald of doubts that you no longer know your own native language.

Diagnosis:

Judgement fatigue!

Something that soaks up water could be described as:

a) a soaker up

b) a soak upper

c) a soaker upper

d) a soaker upperer

e) other variant

:)

Last week I was at a talk on sound change in Southern British English given by Christian Uffmann, specifically related to a vowel shift which seems to be ongoing particularly among younger speakers.

In brief (and much simplified), mid-front vowels (e.g. in ‘dress’) are lowering, low-back and mid-back vowels (e.g. in ‘part, thought’) are raising, and high-back vowels (e.g. in ‘rule’) are fronting.

Some consequences of this are that the words ‘true, looked, could’ now sound very similar to ‘tree, licked, kid’ respectively amongst the speakers with the most advanced shift, and the word ‘mess’ is becoming more like ‘mass’.

Before a syllable-coda /l/, the mid-back and high-back vowels are continuing to merge, e.g. ‘Paul, pool, pull’ sound virtually homophonous (also ‘call, cool’, although ‘calling, cooling’ are still distinct perhaps because these speakers have not lexicalised the merged vowel so that, when /l/ is the onset of the next syllable, the vowels don’t merge).

There also seems to be a split going on which, interestingly, is re-introducing the vowel /y/ (a high-front-rounded vowel) making English more like its Germanic cousins again, e.g. the word ‘ruler’ is pronounced with a high-back-rounded /u/ vowel in ‘ruler = monarch’ but with a high-front rounded /y/ in ‘ruler = item of stationery’. In general, it seems to be that the more morphologically transparent a word is, the more likely it is to keep the /u/, i.e. ‘ruler = monarch’ is still transparently ‘rule + er’ whereas ‘ruler = item of stationery’ is perceived by most to be monomorphemic.

As was implied in the speaker’s summary: Southern Britain is seeing a lowering of ‘dresses’ and a rise in ‘thought’. It seems the best advice then to any who would seek to condemn and counteract such sound change would be to lower their thoughts and raise their dresses…

Stassen (1985) (from whom the examples and typology are taken) identifies a number of different types of comparative construction found in the world’s languages.

First, a language’s primary comparative construction may be a ‘derived case comparative’ or a ‘fixed case comparative.’ Starting with the latter, a fixed case comparative is one where the than-phrase (called the standard phrase) always has a fixed morphological case. There are two sub-types: exceed-comparatives and adverbial comparatives. Of the latter there are three sub-types depending on whether the adverbial is based around the locational concepts of goal, source, or the here-and-now:

1.      Allative (used for motion towards a goal; ‘for, to’):

         Maasai (Nilo-Saharan language from Kenya/Tanzania)

         Sapuk ol-kondi to l-kibulekeny

         Is.big the-deer to the-waterbuck

         ‘The deer is bigger than the waterbuck’

NB ‘than the waterbuck’ is translated as ‘to the-waterbuck.’

2.      Separative (used for motion away from a source; ‘from’):

         Japanese

         Nihon-go wa doits-go yori muzukashi

         Japanese TOP German from difficult

         ‘Japanese is more difficult than German’

NB ‘than German’ is translated as ‘German from.’ This is the most common type of comparative construction.

3.      Locative (not involving motion; ‘on, beside’):

         Chuckchee (Palaeosiberian language from Siberia)

         Gamga-qla’ul-ik qetvu-ci-um

         All-men-on strong-more-1sg

         ‘I am stronger than all men’

NB ‘than all men’ is translated as ‘all-men-on.’

The final type of fixed case comparative is the exceed-comparative where the case of the standard phrase is determined by a verb meaning something like ‘exceed, surpass.’

4.      Exceed comparatives

         Vietnamese

         Vang qui hon bac

         Gold valuable exceed silver

         ‘Gold is worth more than silver’

NB ‘than silver’ is translated as ‘exceed silver.’

Returning to derived case comparatives, the case of the standard phrase in these comparative constructions is not fixed by some element in the sentence, but by the role it plays in the clause. There are two types: conjoined comparatives and particle comparatives.

1.      Conjoined comparatives (two adversative clauses are co-ordinated or juxtaposed):

         Hixkaryana (Carib language from Brazil)

         Kaw-ohra naha Waraka, kaw naha Kaywerye

         Tall-not he-is Waraka tall he-is Kaywerye

         ‘Kaywerye is taller than Waraka’

NB there are two clauses here and no direct translation of ‘than Waraka.’

2.      Particle comparatives (a diverse grouping, the particle does not fix the case of the standard phrase):

         Latin

         Brutum ego non minus amo quam tu

         Brutus-ACC I-NOM not less love-1sg than you-NOM

         ‘I love Brutus no less than you (love Brutus)’

Compare:

         Brutum ego non minus amo quam te

         Brutus-ACC I-NOM not less love-1sg than you-ACC

         ‘I love Brutus no less than (I love) you’

NB the case of the standard noun-phrase ‘you’ depends on its grammatical role (subject or object) in the clause, and not on the particle ‘quam.’

English is considered to have a particle comparative construction (although ‘than’ shows many characteristics of being a preposition, its status is quite a controversial topic). Despite the particle comparative being widespread in the languages of Europe, cross-linguistically it is rare – some might even say it’s exotic!

Reference

Stassen, L. (1985). Comparison and Universal Grammar. Oxford: Blackwell.

You know how articles reporting on psycholinguistic experiments often say something like ‘X number of people took part but Y number of people’s results were discounted for various reasons…didn’t understand the instructions, wasn’t paying attention, was generally incompetent…’, well now I can say I have probably just been admitted to that inevitable and illustrious group of rejected data-providers.

First, I did some example tests to get used to the computer and the instructions for the task which involved learning a made-up language. That was all well and good. Then I started the experiment proper. I was plugging away at the exercises, tapping here, tapping there as required until the researcher came in mid-way and told me, in a kindly, roundabout sort of way, that I was being too slow (it was meant to be a short-term memory test after all)!

After that, I sped up as best I could. After the first short-term memory part, I moved on to the second long-term memory one. Essentially they were testing to see what kinds of rules I had learned from part 1. After the experiment there was a quick interview-like section where the experimenter asked me to describe the rules I had learned from the exercises and what I thought this made-up language was. Now here’s the bizarre bit…I correctly spotted that the made-up language was essentially an ergative language - hooray! However, virtually all the rules that I had been using during the experiment (and consequently my answers) were completely wrong!

It turns out they are testing whether certain alignments (e.g. ergative alignment or accusative alignment or some unattested one) are equally learnable or not. I reckon their null hypothesis will be that unattested and attested systems are equally learnable with the aim of demonstrating that typologically-unattested systems are harder to learn. Evidently I find even attested systems hard to learn! I suppose (and hope, for dignity’s sake!) that in every experiment there’s always one such person!