Why do I get "ValueError: Inconsistent number of columns" when reading sentences from .ConLL file?

Question

from nltk.corpus.reader.conll import ConllCorpusReader

READER = ConllCorpusReader(root="./", fileids=".conll", 
                           columntypes=('words','pos','tree','chunk','ne','srl','ignore')
                          )

READER_sents(myConLLfile)

I'm extracting sentence as a list of strings from a .conll file. The code above doesn't report any errors, so I think something is extracted for each sentence. Yet when I try to print out or add POS-tags to every sentence, the Value Error below occurs for every sentence after the 1007th one.

1. What's happening? Is there a way to see those extracted but ill-structured sentences?
2. How can I extract the sentences properly? I guess some tokens are represented as a tuple of string and OBI instead of string. But then it's weird to have the same error report for many sentences.
3. Worse case, can I only extract those sentences in good structure?

i = 0
for sentence in READER_sents(myConLLfile):
    print(i)
    print(sentence)
    i += 1

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-125-9c03d8d69ec0> in <module>()
      1 i = 0
----> 2 for sentence in READER.sents(myConLLfile):
      3     print(i)
      4     print(sentence)
      5     i += 1

2 frames
/usr/local/lib/python3.6/dist-packages/nltk/corpus/reader/conll.py in _read_grid_block(self, stream)
    206                 if len(row) != len(grid[0]):
    207                     raise ValueError('Inconsistent number of columns:\n%s'
--> 208                                      % block)
    209             grids.append(grid)
    210         return grids

ValueError: Inconsistent number of columns:
This    O
guy O
needs   O
his O
own O
show    O
on  O
Discivery   B-corporation
Channel I-corporation
!   O